2025-08-27 10:44:02 +00:00
69 changed files with 3827 additions and 189 deletions
--- a/.gitignore
+++ b/.gitignore
@@ -38,3 +38,7 @@ __pycache__/
 */.env

 wg_config/wg_confs/
+records/
+src/auracast/server/stream_settings.json
+src/auracast/server/certs/per_device/
+src/auracast/.env
--- a/4
+++ b/4
@@ -1,9 +1,9 @@
 # TODO: investigate using -alpine in the future
-FROM python:3.11
+FROM python:3.11-bookworm

 # Install system dependencies and poetry
 RUN apt-get update && apt-get install -y \
-    iputils-ping \
+    iputils-ping portaudio19-dev\
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*

--- a/README.md
+++ b/README.md
@@ -0,0 +1,194 @@
+## Local HTTP/HTTPS Setup with Custom CA
+
+This project provides a dual-port Streamlit server setup for local networks:
+
+- **HTTP** available on port **8502**
+- **HTTPS** (trusted with custom CA) available on port **8503**
+
+### How it works
+- A custom Certificate Authority (CA) is generated for your organization.
+- Each device/server is issued a certificate signed by this CA.
+- Customers can import the CA certificate into their OS/browser trust store, so the device's HTTPS connection is fully trusted (no browser warnings).
+
+### Usage
+
+1. **Generate Certificates**
+    - Run `generate_ca_cert.sh` in `src/auracast/server/`.
+    - This creates:
+        - `certs/ca/ca_cert.pem` / `ca_key.pem` (CA cert/key)
+    - **Distribute `ca_cert.pem` or `ca_cert.crt` to customers** for installation in their trust store.
+    - This is a one-time operation for your organization.
+
+2. **Start the Server**
+    - Run `run_http_and_https.sh` in `src/auracast/server/`.
+    - This starts:
+        - HTTP Streamlit on port 8500
+        - HTTPS Streamlit on port 8501 (using the signed device cert)
+
+3. **Client Trust Setup**
+    - Customers should install `ca_cert.pem` in their operating system or browser trust store to trust the HTTPS connection.
+    - After this, browsers will show a secure HTTPS connection to the device (no warnings).
+
+### Why this setup?
+- **WebRTC and other browser features require HTTPS for local devices.**
+- Using a local CA allows trusted HTTPS without needing a public certificate or exposing devices to the internet.
+- HTTP is also available for compatibility/testing.
+
+### Advertise Hostname with mDNS
+```bash
+cd src/auracast/server
+sudo ./provision_domain_hostname.sh <new_hostname> <new_domain>
+```
+- Example:
+  ```bash
+  sudo ./provision_domain_hostname.sh box1 auracast.local
+  ```
+- The script will:
+  - Validate your input (no dots in hostname)
+  - Set the system hostname
+  - Update `/etc/hosts`
+  - Set the Avahi domain in `/etc/avahi/avahi-daemon.conf`
+  - Restart Avahi
+  - Generate a unique per-device certificate and key signed by your CA, stored in `certs/per_device/<hostname>.<domain>/`.
+  - The certificate will have a SAN matching the device's mDNS name (e.g., `box1-summitwave.local`).
+---
+
+### Troubleshooting & Tips
+- **Use .local domain** (e.g., `box1-summitwave.local`) - most clients will not resolve multi-label domains.
+- **Hostnames must not contain dots** (`.`). Only use single-label names for the system hostname.
+- **Avahi domain** can be multi-label (e.g., `auracast.local`).
+- **Clients may need** `libnss-mdns` installed and `/etc/nsswitch.conf` configured with `mdns4_minimal` and `mdns4` for multi-label mDNS names.
+- If you have issues with mDNS name resolution, check for conflicting mDNS stacks (e.g., systemd-resolved, Bonjour, or other daemons).
+- Some Linux clients may not resolve multi-label mDNS names via NSS—test with `avahi-resolve-host-name` and try from another device if needed.
+
+---
+
+After completing these steps, your device will be discoverable as `<hostname>.<domain>` (e.g., `box1.auracast.local`) on the local network via mDNS.
+
+---
+
+## Checking Advertised mDNS Services
+Once your device is configured, you can verify that its mDNS advertisement is visible on the network:
+
+- **List all mDNS services:**
+  ```bash
+  avahi-browse -a
+  ```
+  Look for your hostname and service (e.g., `box1.auracast.local`).
+- **Check specific hostname resolution:**
+  ```bash
+  avahi-resolve-host-name box1.auracast.local
+  avahi-resolve-host-name -4 box1.auracast.local  # IPv4 only
+  avahi-resolve-host-name -6 box1.auracast.local  # IPv6 only
+  ```
+
+## Run the application with local webui
+- for microphone streaming via the browser, https is required
+- poetry run multicast_server.py
+- sudo -E PATH="$PATH" bash ./start_frontend_https.sh
+- bash start_mdns.sh
+
+
+## Managing Auracast systemd Services
+
+You can run the backend and frontend as systemd services for easier management and automatic startup on boot.
+
+### 1. Install the service files
+Copy the provided service files to your systemd directory (requires sudo):
+
+```bash
+sudo cp auracast-server.service /etc/systemd/system/
+sudo cp auracast-frontend.service /etc/systemd/system/
+```
+
+### 2. Reload systemd
+```bash
+sudo systemctl daemon-reload
+```
+
+### 3. Enable services to start at boot
+```bash
+sudo systemctl enable auracast-server
+sudo systemctl enable auracast-frontend
+```
+
+### 4. Start the services
+```bash
+sudo systemctl start auracast-server
+sudo systemctl start auracast-frontend
+```
+
+### 5. Stop the services
+```bash
+sudo systemctl stop auracast-server
+sudo systemctl stop auracast-frontend
+```
+
+### 6. Disable services to start at boot
+```bash
+sudo systemctl disable auracast-server
+sudo systemctl disable auracast-frontend
+```
+
+### 7. Check service status
+```bash
+sudo systemctl status auracast-server
+sudo systemctl status auracast-frontend
+```
+
+If you want to run the services as a specific user, edit the `User=` line in the service files accordingly.
+
+# Setup the audio system
+sudo apt update
+
+sudo apt remove -y libportaudio2 portaudio19-dev libportaudiocpp0
+echo "y" | rpi-update stable
+
+sudo apt install -y pipewire wireplumber pipewire-audio-client-libraries rtkit cpufrequtils
+cp src/service/pipewire/99-lowlatency.conf ~/.config/pipewire/pipewire.conf.d/
+sudo cpufreq-set -g performance
+
+/etc/modprobe.d/usb-audio-lowlatency.conf
+option snd_usb_audio nrpacks=1
+
+sudo apt install -y --no-install-recommends \
+        git build-essential cmake pkg-config \
+        libasound2-dev libpulse-devpipewire ethtool linuxptp
+
+git clone https://github.com/PortAudio/portaudio.git
+cd portaudio 
+git checkout 9abe5fe7db729280080a0bbc1397a528cd3ce658     
+rm -rf build
+cmake -S . -B build -G"Unix Makefiles" \
+      -DBUILD_SHARED_LIBS=ON \
+      -DPA_USE_ALSA=OFF \
+      -DPA_USE_PULSEAUDIO=ON \
+      -DPA_USE_JACK=OFF
+cmake --build build -j$(nproc)
+sudo cmake --install build          # installs to /usr/local/lib
+sudo ldconfig                        # refresh linker cache
+
+
+# Device commisioning
+- generate id_ed25519 keypair
+- setup hostname
+  - sudo bash src/auracast/server/provision_domain_hostname.sh box7-summitwave local
+- activate aes67 service
+  - install udev rule for ptp4l
+    - sudo cp src/service/aes67/90-pipewire-aes67-ptp.rules /etc/udev/rules.d/
+    - sudo udevadm control --log-priority=debug --reload-rules
+    - sudo udevadm trigger
+  - bash src/service/update_and_run_aes67.sh
+  - poetry config virtualenvs.in-project true
+  - poetry install
+- activate server and frontend
+  - bash srv/service/update_and_run_server_and_frontend.sh
+- update to latest stable kernel
+    - echo "y" | rpi-update stable
+- place cert
+- disable pw login
+- reboot
+
+# Known issues:
+- When running on a laptop there might be issues switching between usb and browser audio input since they use the same audio device
+
--- a/docker-compose-cloud.yaml
+++ b/docker-compose-cloud.yaml
--- a/docker-compose-test.yaml
+++ b/docker-compose-test.yaml
@@ -1,6 +1,6 @@
 services:
  multicaster:
-    container_name: multicaster-test
+    container_name: multicaster
    privileged: true        # Grants full access to all devices (for serial access)
    restart: unless-stopped
    ports:
--- a/docker-compose-webui.yaml
+++ b/docker-compose-webui.yaml
@@ -0,0 +1,24 @@
+services:
+  multicaster:
+    container_name: multicast-webapp
+    privileged: true        # Grants full access to all devices (for serial access)
+    restart: unless-stopped
+    ports:
+      - "8501:8501" 
+    build:
+      dockerfile: Dockerfile
+      ssh: 
+        - default=~/.ssh/id_ed25519 #lappi
+        #- default=~/.ssh/id_rsa #raspi
+    volumes:
+      - "/dev/serial:/dev/serial"
+      - "/dev/snd:/dev/snd" 
+    #devices:
+      # - /dev/serial/by-id/usb-ZEPHYR_Zephyr_HCI_UART_sample_81BD14B8D71B5662-if00
+    environment:
+      LOG_LEVEL: INFO
+    
+    # start the server and the frontend 
+    command: >
+      bash -c "python ./auracast/server/multicast_server.py & streamlit run ./auracast/server/multicast_frontend.py"
+
--- a/poetry.lock
+++ b/poetry.lock
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -6,13 +6,16 @@ requires-python = ">=3.11"
 dependencies = [
    "bumble @ git+ssh://git@ssh.pstruebi.xyz:222/auracaster/bumble_mirror.git@12bcdb7770c0d57a094bc0a96cd52e701f97fece",
    "lc3 @ git+ssh://git@ssh.pstruebi.xyz:222/auracaster/liblc3.git@7558637303106c7ea971e7bb8cedf379d3e08bcc",
-    "sounddevice",
    "aioconsole",
    "fastapi==0.115.11",
    "uvicorn==0.34.0",
    "aiohttp==3.9.3",
-    "sounddevice (>=0.5.1,<0.6.0)",
-    "aioconsole (>=0.8.1,<0.9.0)"
+    "aioconsole (>=0.8.1,<0.9.0)",
+    "numpy (>=2.2.6,<3.0.0)",
+    "streamlit (>=1.45.1,<2.0.0)",
+    "aiortc (>=1.13.0,<2.0.0)",
+    "sounddevice (>=0.5.2,<0.6.0)",
+    "python-dotenv (>=1.1.1,<2.0.0)"
 ]

 [project.optional-dependencies]
--- a/src/auracast/auracast_config.py
+++ b/src/auracast/auracast_config.py
@@ -35,6 +35,10 @@ class AuracastGlobalConfig(BaseModel):
    presentation_delay_us: int = 40000
    # TODO:pydantic does not support bytes serialization - use .hex and np.fromhex()
    manufacturer_data: tuple[int, bytes] | tuple[None, None] = (None, None)
+    # LE Audio: Broadcast Audio Immediate Rendering (metadata type 0x09)
+    # When true, include a zero-length LTV with type 0x09 in the subgroup metadata
+    # so receivers may render earlier than the presentation delay for lower latency.
+    immediate_rendering: bool = False

 # "Audio input. "
 # "'device' -> use the host's default sound input device, "
@@ -58,42 +62,52 @@ class AuracastBigConfig(BaseModel):
 class AuracastBigConfigDeu(AuracastBigConfig):
    id: int = 12
    random_address: str = 'F1:F1:F2:F3:F4:F5'
-    name: str = 'Broadcast0'
+    name: str = 'Hörsaal A'
    language: str ='deu'
-    program_info: str = 'Announcements German'
-    audio_source: str = 'file:./testdata/announcement_de.wav'
+    program_info: str = 'Vorlesung DE'
+    audio_source: str = 'file:./testdata/wave_particle_5min_de.wav'

 class AuracastBigConfigEng(AuracastBigConfig):
    id: int = 123
    random_address: str = 'F2:F1:F2:F3:F4:F5'
-    name: str  = 'Broadcast1'
+    name: str  = 'Lecture Hall A'
    language: str ='eng'
-    program_info: str  = 'Announcements English'
-    audio_source: str  = 'file:./testdata/announcement_en.wav'
+    program_info: str  = 'Lecture EN'
+    audio_source: str  = 'file:./testdata/wave_particle_5min_en.wav'

 class AuracastBigConfigFra(AuracastBigConfig):
    id: int = 1234
    random_address: str = 'F3:F1:F2:F3:F4:F5'
-    name: str  = 'Broadcast2'
+    # French
+    name: str  = 'Auditoire A'
    language: str ='fra'
-    program_info: str  = 'Announcements French'
-    audio_source: str  = 'file:./testdata/announcement_fr.wav'
+    program_info: str  = 'Auditoire FR'
+    audio_source: str  = 'file:./testdata/wave_particle_5min_fr.wav'

 class AuracastBigConfigSpa(AuracastBigConfig):
    id: int =12345
    random_address: str = 'F4:F1:F2:F3:F4:F5'
-    name: str  = 'Broadcast3'
+    name: str  = 'Auditorio A'
    language: str ='spa'
-    program_info: str  = 'Announcements Spanish'
-    audio_source: str  = 'file:./testdata/announcement_es.wav'
+    program_info: str  = 'Auditorio ES'
+    audio_source: str  = 'file:./testdata/wave_particle_5min_es.wav'

 class AuracastBigConfigIta(AuracastBigConfig):
    id: int =1234567
    random_address: str = 'F5:F1:F2:F3:F4:F5'
-    name: str  = 'Broadcast4'
+    name: str  = 'Aula A'
    language: str ='ita'
-    program_info: str  = 'Announcements Italian'
-    audio_source: str  = 'file:./testdata/announcement_it.wav'
+    program_info: str  = 'Aula IT'
+    audio_source: str  = 'file:./testdata/wave_particle_5min_it.wav'
+
+
+class AuracastBigConfigPol(AuracastBigConfig):
+    id: int =12345678
+    random_address: str = 'F6:F1:F2:F3:F4:F5'
+    name: str  = 'Sala Wykładowa'
+    language: str ='pol'
+    program_info: str  = 'Sala Wykładowa PL'
+    audio_source: str  = 'file:./testdata/wave_particle_5min_pl.wav'


 class AuracastConfigGroup(AuracastGlobalConfig):
--- a/src/auracast/multicast.py
+++ b/src/auracast/multicast.py
@@ -44,13 +44,18 @@ from bumble.profiles import bass
 import bumble.device
 import bumble.transport
 import bumble.utils
+import numpy as np  # for audio down-mix
 from bumble.device import Host, BIGInfoAdvertisement, AdvertisingChannelMap
 from bumble.audio import io as audio_io

 from auracast import auracast_config
 from auracast.utils.read_lc3_file import read_lc3_file
+from auracast.utils.network_audio_receiver import NetworkAudioReceiverUncoded
+from auracast.utils.webrtc_audio_input import WebRTCAudioInput


+# Instantiate WebRTC audio input for streaming (can be used per-BIG or globally)
+
 # modified from bumble
 class ModWaveAudioInput(audio_io.ThreadedAudioInput):
    """Audio input that reads PCM samples from a .wav file."""
@@ -148,6 +153,30 @@ async def init_broadcast(
    bap_sampling_freq = getattr(bap.SamplingFrequency, f"FREQ_{global_config.auracast_sampling_rate_hz}")
    bigs = {}
    for i, conf in enumerate(big_config):
+        metadata=le_audio.Metadata(
+            [
+                le_audio.Metadata.Entry(
+                    tag=le_audio.Metadata.Tag.LANGUAGE, data=conf.language.encode()
+                ),
+                le_audio.Metadata.Entry(
+                    tag=le_audio.Metadata.Tag.PROGRAM_INFO, data=conf.program_info.encode()
+                ),
+                le_audio.Metadata.Entry(
+                    tag=le_audio.Metadata.Tag.BROADCAST_NAME, data=conf.name.encode()
+                ),
+            ]
+            + (
+                [
+                    # Broadcast Audio Immediate Rendering flag (type 0x09), zero-length value
+                    le_audio.Metadata.Entry(tag = le_audio.Metadata.Tag.BROADCAST_AUDIO_IMMEDIATE_RENDERING_FLAG, data=b"")
+                ]
+                if global_config.immediate_rendering #TODO: verify this
+                else []
+            )
+        )
+        logging.info(
+            metadata.pretty_print("\n")
+        )
        bigs[f'big{i}'] = {}
        # Config advertising set
        bigs[f'big{i}']['basic_audio_announcement'] = bap.BasicAudioAnnouncement(
@@ -160,19 +189,7 @@ async def init_broadcast(
                        frame_duration=bap.FrameDuration.DURATION_10000_US,
                        octets_per_codec_frame=global_config.octets_per_frame,
                    ),
-                    metadata=le_audio.Metadata(
-                        [
-                            le_audio.Metadata.Entry(
-                                tag=le_audio.Metadata.Tag.LANGUAGE, data=conf.language.encode()
-                            ),
-                            le_audio.Metadata.Entry(
-                                tag=le_audio.Metadata.Tag.PROGRAM_INFO, data=conf.program_info.encode()
-                            ),
-                            le_audio.Metadata.Entry(
-                            tag=le_audio.Metadata.Tag.BROADCAST_NAME, data=conf.name.encode()
-                            ),
-                        ]
-                    ),
+                    metadata=metadata,
                    bis=[
                        bap.BasicAudioAnnouncement.BIS(
                            index=1,
@@ -211,7 +228,7 @@ async def init_broadcast(
                primary_advertising_interval_max=200,
                advertising_sid=i,
                primary_advertising_phy=hci.Phy.LE_1M, # 2m phy config throws error - because for primary advertising channels, 1mbit is only supported
-                secondary_advertising_phy=hci.Phy.LE_2M, # this is the secondary advertising beeing send on non advertising channels (extendend advertising)
+                secondary_advertising_phy=hci.Phy.LE_1M, # this is the secondary advertising beeing send on non advertising channels (extendend advertising)
                #advertising_tx_power= # tx power in dbm (max 20)
                #secondary_advertising_max_skip=10,
            ),
@@ -272,7 +289,7 @@ async def init_broadcast(

        logging.debug(f'big{i} parameters are:')
        logging.debug('%s', pprint.pformat(vars(big)))
-        logging.debug(f'Finished setup of big{i}.')
+        logging.info(f'Finished setup of big{i}.')

    await asyncio.sleep(i+1) # Wait for advertising to set up

@@ -322,26 +339,75 @@ class Streamer():
        else:
            logging.warning('Streamer is already running')

-    def stop_streaming(self):
-        """Stops the background task if running."""
-        if self.is_streaming:
-            self.is_streaming = False
-            if self.task:
-                self.task.cancel()  # Cancel the task safely
-                self.task = None
+    async def stop_streaming(self):
+        """Gracefully stop streaming and release audio devices."""
+        if not self.is_streaming and self.task is None:
+            return
+
+        # Ask the streaming loop to finish
+        self.is_streaming = False
+        if self.task is not None:
+            self.task.cancel()
+
+            self.task = None
+
+        # Close audio inputs (await to ensure ALSA devices are released)
+        close_tasks = []
+        for big in self.bigs.values():
+            ai = big.get("audio_input")
+            if ai and hasattr(ai, "close"):
+                close_tasks.append(ai.close())
+                # Remove reference so a fresh one is created next time
+                big.pop("audio_input", None)
+        if close_tasks:
+            await asyncio.gather(*close_tasks, return_exceptions=True)

    async def stream(self):
+
        bigs = self.bigs
        big_config = self.big_config
        global_config = self.global_config
-        # init
        for i, big in enumerate(bigs.values()):
            audio_source = big_config[i].audio_source
            input_format = big_config[i].input_format

+            # --- New: network_uncoded mode using NetworkAudioReceiver ---
+            if isinstance(audio_source, NetworkAudioReceiverUncoded):
+                # Start the UDP receiver coroutine so packets are actually received
+                asyncio.create_task(audio_source.receive())
+                encoder = lc3.Encoder(
+                    frame_duration_us=global_config.frame_duration_us,
+                    sample_rate_hz=global_config.auracast_sampling_rate_hz,
+                    num_channels=1,
+                    input_sample_rate_hz=audio_source.samplerate,
+                )
+                lc3_frame_samples = encoder.get_frame_samples()
+                big['pcm_bit_depth'] = 16
+                big['lc3_frame_samples'] = lc3_frame_samples
+                big['lc3_bytes_per_frame'] = global_config.octets_per_frame
+                big['audio_input'] = audio_source
+                big['encoder'] = encoder
+                big['precoded'] = False
+
+            elif audio_source == 'webrtc':
+                big['audio_input'] = WebRTCAudioInput()
+                encoder = lc3.Encoder(
+                    frame_duration_us=global_config.frame_duration_us,
+                    sample_rate_hz=global_config.auracast_sampling_rate_hz,
+                    num_channels=1,
+                    input_sample_rate_hz=48000, # TODO: get samplerate from webrtc
+                )
+                lc3_frame_samples = encoder.get_frame_samples()
+                big['pcm_bit_depth'] = 16
+                big['lc3_frame_samples'] = lc3_frame_samples
+                big['lc3_bytes_per_frame'] = global_config.octets_per_frame
+                big['encoder'] = encoder
+                big['precoded'] = False
+
            # precoded lc3 from ram
-            if isinstance(big_config[i].audio_source, bytes):
+            elif isinstance(big_config[i].audio_source, bytes):
                big['precoded'] = True
+                big['lc3_bytes_per_frame'] = global_config.octets_per_frame

                lc3_frames = iter(big_config[i].audio_source)

@@ -352,6 +418,7 @@ class Streamer():
            # precoded lc3 file
            elif big_config[i].audio_source.endswith('.lc3'):
                big['precoded'] = True
+                big['lc3_bytes_per_frame'] = global_config.octets_per_frame
                filename = big_config[i].audio_source.replace('file:', '')

                lc3_bytes = read_lc3_file(filename)
@@ -363,21 +430,23 @@ class Streamer():

            # use wav files and code them entirely before streaming
            elif big_config[i].precode_wav and big_config[i].audio_source.endswith('.wav'):
+                logging.info('Precoding wav file: %s, this may take a while', big_config[i].audio_source)
                big['precoded'] = True
+                big['lc3_bytes_per_frame'] = global_config.octets_per_frame

                audio_input = await audio_io.create_audio_input(audio_source, input_format)
                audio_input.rewind = False
                pcm_format = await audio_input.open()

                if pcm_format.channels != 1:
-                    print("Only 1 channels PCM configurations are supported")
+                    logging.error("Only 1 channels PCM configurations are supported")
                    return
                if pcm_format.sample_type == audio_io.PcmFormat.SampleType.INT16:
                    pcm_bit_depth = 16
                elif pcm_format.sample_type == audio_io.PcmFormat.SampleType.FLOAT32:
                    pcm_bit_depth = None
                else:
-                    print("Only INT16 and FLOAT32 sample types are supported")
+                    logging.error("Only INT16 and FLOAT32 sample types are supported")
                    return
                encoder = lc3.Encoder(
                    frame_duration_us=global_config.frame_duration_us,
@@ -402,40 +471,92 @@ class Streamer():
            # anything else, e.g. realtime stream from device (bumble)
            else:
                audio_input = await audio_io.create_audio_input(audio_source, input_format)
-                audio_input.rewind = big_config[i].loop
-                pcm_format = await audio_input.open()
+                # Store early so stop_streaming can close even if open() fails
+                big['audio_input'] = audio_input
+                # SoundDeviceAudioInput (used for `mic:<device>` captures) has no `.rewind`.
+                if hasattr(audio_input, "rewind"):
+                    audio_input.rewind = big_config[i].loop

-                #try:
-                if pcm_format.channels != 1:
-                    print("Only 1 channels PCM configurations are supported")
+                # Retry logic – ALSA sometimes keeps the device busy for a short time after the
+                # previous stream has closed. Handle PortAudioError -9985 with back-off retries.
+                import sounddevice as _sd
+                max_attempts = 3
+                for attempt in range(1, max_attempts + 1):
+                    try:
+                        pcm_format = await audio_input.open()
+                        break  # success
+                    except _sd.PortAudioError as err:
+                        # -9985 == paDeviceUnavailable
+                        logging.error('Could not open audio device %s with error %s', audio_source, err)
+                        code = None
+                        if hasattr(err, 'errno'):
+                            code = err.errno
+                        elif len(err.args) > 1 and isinstance(err.args[1], int):
+                            code = err.args[1]
+                        if code == -9985 and attempt < max_attempts:
+                            backoff_ms = 200 * attempt
+                            logging.warning("PortAudio device busy (attempt %d/%d). Retrying in %.1f ms…", attempt, max_attempts, backoff_ms)
+                            # ensure device handle and PortAudio context are closed before retrying
+                            try:
+                                if hasattr(audio_input, "aclose"):
+                                    await audio_input.aclose()
+                                elif hasattr(audio_input, "close"):
+                                    audio_input.close()
+                            except Exception:
+                                pass
+                            # Fully terminate PortAudio to drop lingering handles (sounddevice quirk)
+                            if hasattr(_sd, "_terminate"):
+                                try:
+                                    _sd._terminate()
+                                except Exception:
+                                    pass
+                            # Small pause then re-initialize PortAudio
+                            await asyncio.sleep(0.1)
+                            if hasattr(_sd, "_initialize"):
+                                try:
+                                    _sd._initialize()
+                                except Exception:
+                                    pass
+
+                            # Back-off before next attempt
+                            await asyncio.sleep(backoff_ms / 1000)
+                            # Recreate audio_input fresh for next attempt
+                            audio_input = await audio_io.create_audio_input(audio_source, input_format)
+                            continue
+                        # Other errors or final attempt – re-raise so caller can abort gracefully
+                        raise
+                else:
+                    # Loop exhausted without break
+                    logging.error("Unable to open audio device after %d attempts – giving up", max_attempts)
                    return
+
+                if pcm_format.channels != 1:
+                    logging.info("Input device provides %d channels – will down-mix to mono for LC3", pcm_format.channels)
                if pcm_format.sample_type == audio_io.PcmFormat.SampleType.INT16:
                    pcm_bit_depth = 16
                elif pcm_format.sample_type == audio_io.PcmFormat.SampleType.FLOAT32:
                    pcm_bit_depth = None
                else:
-                    print("Only INT16 and FLOAT32 sample types are supported")
+                    logging.error("Only INT16 and FLOAT32 sample types are supported")
                    return
+                
                encoder = lc3.Encoder(
                    frame_duration_us=global_config.frame_duration_us,
                    sample_rate_hz=global_config.auracast_sampling_rate_hz,
                    num_channels=1,
                    input_sample_rate_hz=pcm_format.sample_rate,
                )
-                lc3_frame_samples = encoder.get_frame_samples() # number of the pcm samples per lc3 frame
                
+                lc3_frame_samples = encoder.get_frame_samples() # number of the pcm samples per lc3 frame
                big['pcm_bit_depth'] = pcm_bit_depth
+                big['channels'] = pcm_format.channels
                big['lc3_frame_samples'] = lc3_frame_samples
+                big['lc3_bytes_per_frame'] = global_config.octets_per_frame
                big['audio_input'] = audio_input
                big['encoder'] = encoder
                big['precoded'] = False

-            # Need for coded an uncoded audio
-            lc3_frame_size = global_config.octets_per_frame #encoder.get_frame_bytes(bitrate)
-            lc3_bytes_per_frame = lc3_frame_size #* 2 #multiplied by number of channels
-            big['lc3_bytes_per_frame'] = lc3_bytes_per_frame

-        # TODO: Maybe do some pre buffering so the stream is stable from the beginning. One half iso queue would be appropriate
        logging.info("Streaming audio...")
        bigs = self.bigs
        self.is_streaming = True
@@ -443,7 +564,6 @@ class Streamer():
        while self.is_streaming:
            stream_finished = [False for _ in range(len(bigs))]
            for i, big in enumerate(bigs.values()):
-
                if big['precoded']:# everything was already lc3 coded beforehand
                    lc3_frame = bytes(
                        itertools.islice(big['lc3_frames'], big['lc3_bytes_per_frame'])
@@ -452,13 +572,26 @@ class Streamer():
                    if lc3_frame == b'': # Not all streams may stop at the same time
                        stream_finished[i] = True
                        continue
-                else:
+                else: # code lc3 on the fly
                    pcm_frame = await anext(big['audio_input'].frames(big['lc3_frame_samples']), None)

                    if pcm_frame is None: # Not all streams may stop at the same time
                        stream_finished[i] = True
                        continue

+                    # Down-mix multi-channel PCM to mono for LC3 encoder if needed
+                    if big.get('channels', 1) > 1:
+                        if isinstance(pcm_frame, np.ndarray):
+                            if pcm_frame.ndim > 1:
+                                mono = pcm_frame.mean(axis=1).astype(pcm_frame.dtype)
+                                pcm_frame = mono
+                        else:
+                            # Convert raw bytes to numpy, average channels, convert back
+                            dtype = np.int16 if big['pcm_bit_depth'] == 16 else np.float32
+                            samples = np.frombuffer(pcm_frame, dtype=dtype)
+                            samples = samples.reshape(-1, big['channels']).mean(axis=1)
+                            pcm_frame = samples.astype(dtype).tobytes()
+
                    lc3_frame = big['encoder'].encode(
                        pcm_frame, num_bytes=big['lc3_bytes_per_frame'], bit_depth=big['pcm_bit_depth']
                    )
@@ -511,13 +644,12 @@ async def broadcast(global_conf: auracast_config.AuracastGlobalConfig, big_conf:
 if __name__ == "__main__":
    import os
    
-    logging.basicConfig( #export LOG_LEVEL=INFO
-        level=os.environ.get('LOG_LEVEL', logging.DEBUG),
+    logging.basicConfig( #export LOG_LEVEL=DEBUG
+        level=os.environ.get('LOG_LEVEL', logging.INFO),
        format='%(module)s.py:%(lineno)d %(levelname)s: %(message)s'
    )
    os.chdir(os.path.dirname(__file__))

-
    config = auracast_config.AuracastConfigGroup(
        bigs = [
            auracast_config.AuracastBigConfigDeu(),
@@ -537,15 +669,19 @@ if __name__ == "__main__":
    #config.transport='serial:/dev/serial/by-id/usb-ZEPHYR_Zephyr_HCI_UART_sample_95A087EADB030B24-if00,115200,rtscts' #nrf52dongle hci_uart usb cdc
    #config.transport='usb:2fe3:000b' #nrf52dongle hci_usb # TODO: iso packet over usb not supported
    #config.transport= 'auto'
-    config.transport='serial:/dev/ttyAMA2,1000000,rtscts' # transport for raspberry pi
+    config.transport='serial:/dev/ttyAMA4,1000000,rtscts' # transport for raspberry pi

+    # TODO: encrypted streams are not working

-    for big in config.bigs: # TODO: encrypted streams are not working
+    for big in config.bigs: 
        #big.code = 'ff'*16 # returns hci/HCI_ENCRYPTION_MODE_NOT_ACCEPTABLE_ERROR
        #big.code = '78 e5 dc f1 34 ab 42 bf c1 92 ef dd 3a fd 67 ae'
-        big.precode_wav = True
-        big.audio_source = big.audio_source.replace('.wav', '_10_16_32.lc3') #lc3 precoded files
-        big.audio_source = read_lc3_file(big.audio_source)  # load files in advance
+        big.precode_wav = False
+        #big.audio_source = big.audio_source.replace('.wav', '_10_16_32.lc3') #lc3 precoded files
+        #big.audio_source = read_lc3_file(big.audio_source)  # load files in advance
+
+        # --- Network_uncoded mode using NetworkAudioReceiver ---
+        #big.audio_source = NetworkAudioReceiverUncoded(port=50007, samplerate=16000, channels=1, chunk_size=1024)
    
    # 16kHz works reliably with 3 streams
    # 24kHz is only working with 2 streams - probably airtime constraint
--- a/src/auracast/multicast_control.py
+++ b/src/auracast/multicast_control.py
@@ -52,13 +52,19 @@ class Multicaster:
        self.device = device
        self.is_auracast_init =  True

-    def start_streaming(self):
+    async def start_streaming(self):
+        """Start streaming; if an old stream is running, stop it first to release audio devices."""
+        if self.streamer is not None:
+            await self.stop_streaming()
+            # Brief pause to ensure ALSA/PortAudio fully releases the input device
+            await asyncio.sleep(0.5)
        self.streamer = multicast.Streamer(self.bigs, self.global_conf, self.big_conf)
        self.streamer.start_streaming()

-    def stop_streaming(self):
+
+    async def stop_streaming(self):
        if self.streamer is not None:
-            self.streamer.stop_streaming()
+            await self.streamer.stop_streaming()
            self.streamer = None

    async def reset(self):
@@ -66,18 +72,28 @@ class Multicaster:
        self.__init__(self.global_conf, self.big_conf)

    async def shutdown(self):
+        # Ensure streaming is fully stopped before tearing down Bluetooth resources
+        if self.streamer is not None:
+            await self.stop_streaming()
+
        self.is_auracast_init = False
-        self. is_audio_init = False
+        self.is_audio_init = False
+        
+        if self.bigs:
+            for big in self.bigs.values():
+                if big.get('audio_input'):
+                    if hasattr(big['audio_input'], 'aclose'):
+                        await big['audio_input'].aclose()
+
        if self.device:
            await self.device.stop_advertising()
            if self.bigs:
                for big in self.bigs.values():
-                    if big['advertising_set']:
+                    if big.get('advertising_set'):
                        await big['advertising_set'].stop()
            await self.device_acm.__aexit__(None, None, None)  # Manually triggering teardown


-
 # example commandline ui
 async def command_line_ui(caster: Multicaster):
    while True:
--- a/src/auracast/multicast_script.py
+++ b/src/auracast/multicast_script.py
@@ -0,0 +1,155 @@
+"""
+multicast_script
+=================
+
+Loads environment variables from a .env file located next to this script
+and configures the multicast broadcast. Only UPPERCASE keys are read.
+
+Environment variables
+---------------------
+- LOG_LEVEL: Logging level for the script.
+  Default: INFO. Examples: DEBUG, INFO, WARNING, ERROR.
+
+- INPUT: Select audio capture source.
+  Values:
+    - "usb" (default): first available USB input device.
+    - "aes67": select AES67 inputs. Two forms:
+        * INPUT=aes67            -> first available AES67 input.
+        * INPUT=aes67,<substr>   -> case-insensitive substring match against
+          the device name, e.g. INPUT=aes67,8f6326.
+
+- BROADCAST_NAME: Name of the broadcast (Auracast BIG name).
+  Default: "Broadcast0".
+
+- PROGRAM_INFO: Free-text program/broadcast info.
+  Default: "Some Announcements".
+
+- LANGUATE: ISO 639-3 language code used by config (intentional key name).
+  Default: "deu".
+
+- PULSE_LATENCY_MSEC: Pulse/PipeWire latency hint in milliseconds.
+  Default: 3.
+
+Examples (.env)
+---------------
+LOG_LEVEL=DEBUG
+INPUT=aes67,8f6326
+BROADCAST_NAME=MyBroadcast
+PROGRAM_INFO="Live announcements"
+LANGUATE=deu
+"""
+import logging
+import os
+import time
+from dotenv import load_dotenv
+from auracast import multicast
+from auracast import auracast_config
+from auracast.utils.sounddevice_utils import list_usb_pw_inputs, list_network_pw_inputs
+
+
+if __name__ == "__main__":
+    
+    logging.basicConfig( #export LOG_LEVEL=DEBUG
+        level=os.environ.get('LOG_LEVEL', logging.INFO),
+        format='%(module)s.py:%(lineno)d %(levelname)s: %(message)s'
+    )
+    os.chdir(os.path.dirname(__file__))
+    # Load .env located next to this script (only uppercase keys will be referenced)
+    load_dotenv(dotenv_path='.env')
+    
+    os.environ.setdefault("PULSE_LATENCY_MSEC", "3")
+  
+    usb_inputs = list_usb_pw_inputs()
+    logging.info("USB pw inputs:")
+    for i, d in usb_inputs:
+        logging.info(f"{i}: {d['name']}  in={d['max_input_channels']}")
+  
+    aes67_inputs = list_network_pw_inputs()
+    logging.info("AES67 pw inputs:")
+    for i, d in aes67_inputs:
+        logging.info(f"{i}: {d['name']}  in={d['max_input_channels']}")
+    
+    # Input selection (usb | aes67). Default to usb.
+    # Allows specifying an AES67 device by substring: INPUT=aes67,<substring>
+    # Example: INPUT=aes67,8f6326 will match a device name containing "8f6326".
+    input_env = os.environ.get('INPUT', 'usb') or 'usb'
+    parts = [p.strip() for p in input_env.split(',', 1)]
+    input_mode = (parts[0] or 'usb').lower()
+    iface_substr = (parts[1].lower() if len(parts) > 1 and parts[1] else None)
+
+    selected_dev = None
+    if input_mode == 'aes67':
+        if not aes67_inputs and not iface_substr:
+            # No AES67 inputs and no specific target -> fail fast
+            raise RuntimeError("No AES67 audio inputs found.")
+        if iface_substr:
+            # Loop until a matching AES67 input becomes available
+            while True:
+                current = list_network_pw_inputs()
+                sel = next(((i, d) for i, d in current if iface_substr in (d.get('name','').lower())), None)
+                if sel:
+                    input_sel = sel[0]
+                    selected_dev = sel[1]
+                    logging.info(f"Selected AES67 input by match '{iface_substr}': index={input_sel}")
+                    break
+                logging.info(f"Waiting for AES67 input matching '{iface_substr}'... retrying in 2s")
+                time.sleep(2)
+        else:
+            input_sel, selected_dev = aes67_inputs[0]
+            logging.info(f"Selected first AES67 input: index={input_sel}, device={selected_dev['name']}")
+    else:
+        if usb_inputs:
+            input_sel, selected_dev = usb_inputs[0]
+            logging.info(f"Selected first USB input: index={input_sel}, device={selected_dev['name']}")
+        else:
+            raise RuntimeError("No USB audio inputs found.")
+
+    TRANSPORT1 = 'serial:/dev/ttyAMA3,1000000,rtscts' # transport for raspberry pi gpio header
+    TRANSPORT2 = 'serial:/dev/ttyAMA4,1000000,rtscts' # transport for raspberry pi gpio header
+    # Capture at 48 kHz to avoid PipeWire resampler latency; encode LC3 at 24 kHz
+    CAPTURE_SRATE = 48000
+    LC3_SRATE = 24000
+    OCTETS_PER_FRAME=60
+
+    # Read uppercase-only settings from environment/.env
+    broadcast_name = os.environ.get('BROADCAST_NAME', 'Broadcast0')
+    program_info = os.environ.get('PROGRAM_INFO', 'Some Announcements')
+    # Note: 'LANGUATE' (typo) is intentionally used as requested, maps to config.language
+    language = os.environ.get('LANGUATE', 'deu')
+
+    # Determine capture channel count based on selected device (prefer up to 2)
+    try:
+        max_in = int((selected_dev or {}).get('max_input_channels', 1))
+    except Exception:
+        max_in = 1
+    channels = max(1, min(2, max_in))
+
+    config = auracast_config.AuracastConfigGroup(
+        bigs = [
+            auracast_config.AuracastBigConfig(
+                name=broadcast_name,
+                program_info=program_info,
+                language=language,
+                iso_que_len=1,
+                audio_source=f'device:{input_sel}',
+                input_format=f"int16le,{CAPTURE_SRATE},{channels}",
+                sampling_frequency=LC3_SRATE,
+                octets_per_frame=OCTETS_PER_FRAME,
+            ),
+            #auracast_config.AuracastBigConfigEng(),
+        ],
+        immediate_rendering=True,
+        presentation_delay_us=40000,
+        qos_config=auracast_config.AuracastQosHigh(),
+        auracast_sampling_rate_hz = LC3_SRATE,
+        octets_per_frame = OCTETS_PER_FRAME, # 32kbps@16kHz
+        transport=TRANSPORT1
+    )
+    #config.debug = True
+
+    multicast.run_async(
+        multicast.broadcast(
+            config,
+            config.bigs
+        )
+    )
--- a/src/auracast/multicast_server.py
+++ b/src/auracast/multicast_server.py
@@ -1,104 +0,0 @@
-import glob
-import logging as log
-from fastapi import FastAPI, HTTPException
-from auracast import multicast_control, auracast_config
-
-app = FastAPI()
-
-# Initialize global configuration
-global_config_group = auracast_config.AuracastConfigGroup()
-
-# Create multicast controller
-multicaster: multicast_control.Multicaster | None = None
-
-
-@app.post("/init")
-async def initialize(conf: auracast_config.AuracastConfigGroup):
-    """Initializes the broadcasters."""
-    global global_config_group
-    global multicaster
-    try:
-        if conf.transport == 'auto':
-            serial_devices = glob.glob('/dev/serial/by-id/*')
-            log.info('Found serial devices: %s', serial_devices)
-            for device in serial_devices:
-                if 'usb-ZEPHYR_Zephyr_HCI_UART_sample' in device:
-                    log.info('Using: %s', device)
-                    conf.transport = f'serial:{device},115200,rtscts'
-                    break
-
-        # check again if transport is still auto
-        if conf.transport == 'auto':
-            HTTPException(status_code=500, detail='No suitable transport found.')
-
-        # initialize the streams dict
-        global_config_group = conf
-        log.info(
-            'Initializing multicaster with config:\n %s', conf.model_dump_json(indent=2)
-        )
-        multicaster = multicast_control.Multicaster(
-            conf,
-            conf.bigs,
-        )
-        await multicaster.init_broadcast()
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-
-@app.post("/stream_lc3")
-async def send_audio(audio_data: dict[str, str]):
-    """Streams pre-coded LC3 audio."""
-    if multicaster is None:
-        raise HTTPException(status_code=500, detail='Auracast endpoint was never intialized')
-    try:
-        for big in global_config_group.bigs:    
-            assert big.language in audio_data, HTTPException(status_code=500, detail='language len missmatch')
-            log.info('Received a send audio request for %s', big.language)
-            big.audio_source = audio_data[big.language].encode('latin-1') # TODO: use base64 encoding
-
-        multicaster.big_conf = global_config_group.bigs
-        multicaster.start_streaming()
-        return {"status": "audio_sent"}
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-
-@app.post("/shutdown")
-async def shutdown():
-    """Stops broadcasting."""
-    try:
-        await multicaster.reset()
-        return {"status": "stopped"}
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-
-@app.post("/stop_audio")
-async def stop_audio():
-    """Stops streaming."""
-    try:
-        multicaster.stop_streaming()
-        return {"status": "stopped"}
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-
-@app.get("/status")
-async def get_status():
-    """Gets the current status of the multicaster."""
-    if multicaster:
-        return multicaster.get_status()
-    else:
-        return {
-            'is_initialized': False,
-            'is_streaming': False,
-        }
-
-
-if __name__ == '__main__':
-    import uvicorn
-    log.basicConfig(
-        level=log.INFO,
-        format='%(module)s.py:%(lineno)d %(levelname)s: %(message)s'
-    )
-    uvicorn.run(app, host="0.0.0.0", port=5000)
--- a/src/auracast/server/certs/ca/ca_cert.crt
+++ b/src/auracast/server/certs/ca/ca_cert.crt
@@ -0,0 +1,30 @@
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----
--- a/src/auracast/server/certs/ca/ca_cert.der
+++ b/src/auracast/server/certs/ca/ca_cert.der
--- a/src/auracast/server/certs/ca/ca_cert.pem
+++ b/src/auracast/server/certs/ca/ca_cert.pem
@@ -0,0 +1,30 @@
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----
--- a/src/auracast/server/certs/ca/ca_cert.srl
+++ b/src/auracast/server/certs/ca/ca_cert.srl
@@ -0,0 +1 @@
+5078804E6FBCF893D5537715FD928E46AD576ECA
--- a/src/auracast/server/certs/ca/ca_key.pem
+++ b/src/auracast/server/certs/ca/ca_key.pem
@@ -0,0 +1,52 @@
+-----BEGIN PRIVATE KEY-----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+-----END PRIVATE KEY-----
--- a/src/auracast/server/generate_ca_cert.sh
+++ b/src/auracast/server/generate_ca_cert.sh
@@ -0,0 +1,30 @@
+#!/bin/bash
+# Script to generate a CA cert/key and a device/server cert signed by this CA
+# Outputs: ca_cert.pem, ca_key.pem, device_cert.pem, device_key.pem
+
+CA_DIR=certs/ca
+mkdir -p "$CA_DIR"
+CA_CERT=$CA_DIR/ca_cert.pem
+CA_KEY=$CA_DIR/ca_key.pem
+
+# Generate CA key and cert (20 year expiry)
+echo "Generating CA key and certificate (20 year expiry)..."
+openssl req -x509 -newkey rsa:4096 -days 7300 -nodes -subj "/CN=SummitWaveCA" -keyout "$CA_KEY" -out "$CA_CERT"
+
+# PEM version (for most browsers)
+cp "$CA_CERT" "$CA_DIR/ca_cert.crt"
+# DER version (for Windows)
+openssl x509 -in "$CA_CERT" -outform der -out "$CA_DIR/ca_cert.der"
+
+# Output summary
+echo "CA cert: $CA_CERT"
+echo "CA cert (CRT for browser import): $CA_DIR/ca_cert.crt"
+echo "CA key: $CA_KEY"
+echo "Distribute $CA_CERT or $CA_DIR/ca_cert.crt to clients to trust this device."
+echo "Keep $CA_KEY secret and offline except when signing device CSRs."
+echo "CA cert: $CA_CERT"
+echo "CA cert (CRT for browser import): $CERT_DIR/ca_cert.crt"
+echo "CA key: $CA_KEY"
+echo "Device cert: $DEVICE_CERT"
+echo "Device key: $DEVICE_KEY"
+echo "Distribute $CA_CERT or $CERT_DIR/ca_cert.crt to clients to trust this device."
--- a/src/auracast/server/multicast_frontend.py
+++ b/src/auracast/server/multicast_frontend.py
@@ -0,0 +1,399 @@
+# frontend/app.py
+import os
+import time
+import streamlit as st
+import requests
+from auracast import auracast_config
+import logging as log
+
+# Track whether WebRTC stream is active across Streamlit reruns
+if 'stream_started' not in st.session_state:
+    st.session_state['stream_started'] = False
+
+# Global: desired packetization time in ms for Opus (should match backend)
+PTIME = 40
+BACKEND_URL = "http://localhost:5000"
+#TRANSPORT1 = "serial:/dev/serial/by-id/usb-ZEPHYR_Zephyr_HCI_UART_sample_B53C372677E14460-if00,115200,rtscts"
+#TRANSPORT2 = "serial:/dev/serial/by-id/usb-ZEPHYR_Zephyr_HCI_UART_sample_CC69A2912F84AE5E-if00,115200,rtscts"
+
+TRANSPORT1 = 'serial:/dev/ttyAMA3,1000000,rtscts' # transport for raspberry pi gpio header
+TRANSPORT2 = 'serial:/dev/ttyAMA4,1000000,rtscts' # transport for raspberry pi gpio header
+QUALITY_MAP = {
+    "High (48kHz)": {"rate": 48000, "octets": 120},
+    "Good (32kHz)": {"rate": 32000, "octets": 80},
+    "Medium (24kHz)": {"rate": 24000, "octets": 60},
+    "Fair (16kHz)": {"rate": 16000, "octets": 40},
+}
+
+# Try loading persisted settings from backend
+saved_settings = {}
+try:
+    resp = requests.get(f"{BACKEND_URL}/status", timeout=1)
+    if resp.status_code == 200:
+        saved_settings = resp.json()
+except Exception:
+    saved_settings = {}
+
+st.title("🎙️ Auracast Audio Mode Control")
+
+# Audio mode selection with persisted default
+options = ["Webapp", "USB/Network", "Demo"]
+saved_audio_mode = saved_settings.get("audio_mode", "Webapp")
+if saved_audio_mode not in options:
+    saved_audio_mode = "Webapp"
+
+audio_mode = st.selectbox(
+    "Audio Mode",
+    options,
+    index=options.index(saved_audio_mode),
+    help="Select the audio input source. Choose 'Webapp' for browser microphone, 'USB/Network' for a connected hardware device, or 'Demo' for a simulated stream."
+)
+
+if audio_mode == "Demo":
+    demo_stream_map = {
+        "1 × 48kHz": {"quality": "High (48kHz)", "streams": 1},
+        "2 × 24kHz": {"quality": "Medium (24kHz)", "streams": 2},
+        "3 × 16kHz": {"quality": "Fair (16kHz)", "streams": 3},
+        "2 × 48kHz": {"quality": "High (48kHz)", "streams": 2},
+        "4 × 24kHz": {"quality": "Medium (24kHz)", "streams": 4},
+        "6 × 16kHz": {"quality": "Fair (16kHz)", "streams": 6},
+    }
+    demo_options = list(demo_stream_map.keys())
+    default_demo = demo_options[0]
+    demo_selected = st.selectbox(
+        "Demo Stream Type",
+        demo_options,
+        index=0,
+        help="Select the demo stream configuration."
+    )
+    #st.info(f"Demo mode selected: {demo_selected} (Streams: {demo_stream_map[demo_selected]['streams']}, Rate: {demo_stream_map[demo_selected]['rate']} Hz)")
+    # Start/Stop buttons for demo mode
+    if 'demo_stream_started' not in st.session_state:
+        st.session_state['demo_stream_started'] = False
+    col1, col2 = st.columns(2)
+    with col1:
+        start_demo = st.button("Start Demo Stream")
+    with col2:
+        stop_demo = st.button("Stop Demo Stream")
+    if start_demo:
+        # Always stop any running stream for clean state
+        try:
+            requests.post(f"{BACKEND_URL}/stop_audio").json()
+        except Exception:
+            pass
+        time.sleep(1)
+        demo_cfg = demo_stream_map[demo_selected]
+        # Octets per frame logic matches quality_map
+        q = QUALITY_MAP[demo_cfg['quality']]
+
+        # Language configs and test files
+        lang_cfgs = [
+            (auracast_config.AuracastBigConfigDeu, 'de'),
+            (auracast_config.AuracastBigConfigEng, 'en'),
+            (auracast_config.AuracastBigConfigFra, 'fr'),
+            (auracast_config.AuracastBigConfigSpa, 'es'),
+            (auracast_config.AuracastBigConfigIta, 'it'),
+            (auracast_config.AuracastBigConfigPol, 'pl'),
+        ]
+        bigs1 = []
+        for i in range(demo_cfg['streams']):
+            cfg_cls, lang = lang_cfgs[i % len(lang_cfgs)]
+            bigs1.append(cfg_cls(
+                audio_source=f'file:../testdata/wave_particle_5min_{lang}_{int(q["rate"]/1000)}kHz_mono.wav',
+                iso_que_len=32,
+                sampling_frequency=q['rate'],
+                octets_per_frame=q['octets'],
+            ))
+
+        # Split bigs into two configs if needed
+        max_per_mc = {48000: 1, 24000: 2, 16000: 3}
+        max_streams = max_per_mc.get(q['rate'], 3)
+        bigs2 = []
+        if len(bigs1) > max_streams:
+            bigs2 = bigs1[max_streams:]
+            bigs1 = bigs1[:max_streams]
+        config1 = auracast_config.AuracastConfigGroup(
+            auracast_sampling_rate_hz=q['rate'],
+            octets_per_frame=q['octets'],
+            transport=TRANSPORT1,
+            bigs=bigs1
+        )
+        config2 = None
+        if bigs2:
+            config2 = auracast_config.AuracastConfigGroup(
+                auracast_sampling_rate_hz=q['rate'],
+                octets_per_frame=q['octets'],
+                transport=TRANSPORT2,
+                bigs=bigs2
+            )
+        # Call /init and /init2
+        try:
+            r1 = requests.post(f"{BACKEND_URL}/init", json=config1.model_dump())
+            if r1.status_code == 200:
+                msg = f"Demo stream started on multicaster 1 ({len(bigs1)} streams)"
+                st.session_state['demo_stream_started'] = True
+                st.success(msg)
+            else:
+                st.session_state['demo_stream_started'] = False
+                st.error(f"Failed to initialize multicaster 1: {r1.text}")
+            if config2:
+                r2 = requests.post(f"{BACKEND_URL}/init2", json=config2.model_dump())
+                if r2.status_code == 200:
+                    st.success(f"Demo stream started on multicaster 2 ({len(bigs2)} streams)")
+                else:
+                    st.error(f"Failed to initialize multicaster 2: {r2.text}")
+        except Exception as e:
+            st.session_state['demo_stream_started'] = False
+            st.error(f"Error: {e}")
+    elif stop_demo:
+        try:
+            r = requests.post(f"{BACKEND_URL}/stop_audio").json()
+            st.session_state['demo_stream_started'] = False
+            if r.get('was_running'):
+                st.info("Demo stream stopped.")
+            else:
+                st.info("Demo stream was not running.")
+        except Exception as e:
+            st.error(f"Error: {e}")
+    elif st.session_state['demo_stream_started']:
+        st.success(f"Demo stream running: {demo_selected}")
+    else:
+        st.info("Demo stream not running.")
+    quality = None  # Not used in demo mode
+else:
+    # Stream quality selection (now enabled)
+
+    quality_options = list(QUALITY_MAP.keys())
+    default_quality = "Medium (24kHz)" if "Medium (24kHz)" in quality_options else quality_options[0]
+    quality = st.selectbox(
+        "Stream Quality (Sampling Rate)",
+        quality_options,
+        index=quality_options.index(default_quality),
+        help="Select the audio sampling rate for the stream. Lower rates may improve compatibility."
+    )
+    default_name = saved_settings.get('channel_names', ["Broadcast0"])[0]
+    default_lang = saved_settings.get('languages', ["deu"])[0]
+    default_input = saved_settings.get('input_device') or 'default'
+    stream_name = st.text_input(
+        "Channel Name",
+        value=default_name,
+        help="The primary name for your broadcast. Like the SSID of a WLAN, it identifies your stream for receivers."
+    )
+    raw_program_info = saved_settings.get('program_info', default_name)
+    if isinstance(raw_program_info, list) and raw_program_info:
+        default_program_info = raw_program_info[0]
+    else:
+        default_program_info = raw_program_info
+    program_info = st.text_input(
+        "Program Info",
+        value=default_program_info,
+        help="Additional details about the broadcast program, such as its content or purpose. Shown to receivers for more context."
+    )
+    language = st.text_input(
+        "Language (ISO 639-3)",
+        value=default_lang,
+        help="Three-letter language code (e.g., 'eng' for English, 'deu' for German). Used by receivers to display the language of the stream. See: https://en.wikipedia.org/wiki/List_of_ISO_639-3_codes"
+    )
+    # Gain slider for Webapp mode
+    if audio_mode == "Webapp":
+        mic_gain = st.slider("Microphone Gain", 0.0, 2.0, 1.0, 0.1, help="Adjust microphone volume sent to Auracast")
+    else:
+        mic_gain = 1.0
+
+    # Input device selection for USB mode
+    if audio_mode == "USB/Network":
+        resp = requests.get(f"{BACKEND_URL}/audio_inputs")
+        device_list = resp.json().get('inputs', [])
+        # Display "name [id]" but use name as value
+        input_options = [f"{d['name']} [{d['id']}]" for d in device_list]
+        option_name_map = {f"{d['name']} [{d['id']}]": d['name'] for d in device_list}
+        device_names = [d['name'] for d in device_list]
+
+        # Determine default input by name
+        default_input_name = saved_settings.get('input_device')
+        if default_input_name not in device_names and device_names:
+            default_input_name = device_names[0]
+        default_input_label = None
+        for label, name in option_name_map.items():
+            if name == default_input_name:
+                default_input_label = label
+                break
+        if not input_options:
+            st.warning("No hardware audio input devices found. Plug in a USB input device and click Refresh.")
+            if st.button("Refresh"):
+                try:
+                    requests.post(f"{BACKEND_URL}/refresh_audio_inputs", timeout=3)
+                except Exception as e:
+                    st.error(f"Failed to refresh devices: {e}")
+                st.rerun()
+            input_device = None
+        else:
+            col1, col2 = st.columns([3, 1], vertical_alignment="bottom")
+            with col1:
+                selected_option = st.selectbox(
+                    "Input Device",
+                    input_options,
+                    index=input_options.index(default_input_label) if default_input_label in input_options else 0
+                )
+            with col2:
+                if st.button("Refresh"):
+                    try:
+                        requests.post(f"{BACKEND_URL}/refresh_audio_inputs", timeout=3)
+                    except Exception as e:
+                        st.error(f"Failed to refresh devices: {e}")
+                    st.rerun()
+            # Send only the device name to backend
+            input_device = option_name_map[selected_option] if selected_option in option_name_map else None
+    else:
+        input_device = None
+    
+    start_stream = st.button("Start Auracast")
+    stop_stream = st.button("Stop Auracast")
+
+    # If gain slider moved while streaming, send update to JS without restarting
+    if audio_mode == "Webapp" and st.session_state.get('stream_started'):
+        update_js = f"""
+        <script>
+            if (window.gainNode) {{ window.gainNode.gain.value = {mic_gain}; }}
+        </script>
+        """
+        st.components.v1.html(update_js, height=0)
+
+    if stop_stream:
+        st.session_state['stream_started'] = False
+        try:
+            r = requests.post(f"{BACKEND_URL}/stop_audio").json()
+            if r['was_running']:
+                st.success("Stream Stopped!")
+            else:
+                st.success("Stream was not running.")
+        except Exception as e:
+            st.error(f"Error: {e}")
+        # Ensure existing WebRTC connection is fully closed so that a fresh
+        # connection is created the next time we start the stream.
+        if audio_mode == "Webapp":
+            cleanup_js = """
+            <script>
+                if (window.webrtc_pc) {
+                    window.webrtc_pc.getSenders().forEach(s => s.track.stop());
+                    window.webrtc_pc.close();
+                    window.webrtc_pc = null;
+                }
+                window.webrtc_started = false;
+            </script>
+            """
+            st.components.v1.html(cleanup_js, height=0)
+
+    if start_stream:
+        # Always send stop to ensure backend is in a clean state, regardless of current status
+        r = requests.post(f"{BACKEND_URL}/stop_audio").json()
+        if r['was_running']:
+            st.success("Stream Stopped!")
+
+        # Small pause lets backend fully release audio devices before re-init
+        time.sleep(1)
+        # Prepare config using the model (do NOT send qos_config, only relevant fields)
+        q = QUALITY_MAP[quality]
+        config = auracast_config.AuracastConfigGroup(
+            auracast_sampling_rate_hz=q['rate'],
+            octets_per_frame=q['octets'],
+            transport=TRANSPORT1, # transport for raspberry pi gpio header
+            bigs = [
+                auracast_config.AuracastBigConfig(
+                    name=stream_name,
+                    program_info=program_info,
+                    language=language,
+                    audio_source=(
+                        f"device:{input_device}" if audio_mode == "USB/Network" else (
+                            "webrtc" if audio_mode == "Webapp" else "network"
+                        )
+                    ),
+                    input_format=(f"int16le,{q['rate']},1" if audio_mode == "USB/Network" else "auto"),
+                    iso_que_len=1,
+                    sampling_frequency=q['rate'],
+                    octets_per_frame=q['octets'],
+                ),
+            ]
+        )
+
+        try:
+            r = requests.post(f"{BACKEND_URL}/init", json=config.model_dump())
+            if r.status_code == 200:
+                st.success("Stream Started!")
+            else:
+                st.error(f"Failed to initialize: {r.text}")
+        except Exception as e:
+            st.error(f"Error: {e}")
+
+    # Render / maintain WebRTC component
+    if audio_mode == "Webapp" and (start_stream or st.session_state.get('stream_started')):
+        st.markdown("Starting microphone; allow access if prompted and speak.")
+        component = f"""
+        <script>
+            (async () => {{
+                 // Clean up any previous WebRTC connection before starting a new one
+                 if (window.webrtc_pc) {{
+                     window.webrtc_pc.getSenders().forEach(s => s.track.stop());
+                     window.webrtc_pc.close();
+                 }}
+                const GAIN_VALUE = {mic_gain};
+                const pc = new RTCPeerConnection(); // No STUN needed for localhost
+                window.webrtc_pc = pc;
+                window.webrtc_started = true;
+                const micStream = await navigator.mediaDevices.getUserMedia({{audio:true}});
+                // Create Web Audio gain processing
+                const audioCtx = new (window.AudioContext || window.webkitAudioContext)();
+                const source = audioCtx.createMediaStreamSource(micStream);
+                const gainNode = audioCtx.createGain();
+                gainNode.gain.value = GAIN_VALUE;
+                // Expose for later adjustments
+                window.gainNode = gainNode;
+                const dest = audioCtx.createMediaStreamDestination();
+                source.connect(gainNode).connect(dest);
+                // Add processed tracks to WebRTC
+                dest.stream.getTracks().forEach(t => pc.addTrack(t, dest.stream));
+                // --- WebRTC offer/answer exchange ---
+                const offer = await pc.createOffer();
+                // Patch SDP offer to include a=ptime using global PTIME
+                let sdp = offer.sdp;
+                const ptime_line = 'a=ptime:{PTIME}';
+                const maxptime_line = 'a=maxptime:{PTIME}';
+                if (sdp.includes('a=sendrecv')) {{
+                    sdp = sdp.replace('a=sendrecv', 'a=sendrecv\\n' + ptime_line + '\\n' + maxptime_line);
+                }} else {{
+                    sdp += '\\n' + ptime_line + '\\n' + maxptime_line;
+                }}
+                const patched_offer = new RTCSessionDescription({{sdp, type: offer.type}});
+                await pc.setLocalDescription(patched_offer);
+                // Send offer to backend
+                const response = await fetch(
+                    "{BACKEND_URL}/offer",
+                    {{
+                        method: 'POST',
+                        headers: {{'Content-Type':'application/json'}},
+                        body: JSON.stringify({{sdp: pc.localDescription.sdp, type: pc.localDescription.type}})
+                    }}
+                );
+                const answer = await response.json();
+                await pc.setRemoteDescription(new RTCSessionDescription({{sdp: answer.sdp, type: answer.type}}));
+            }})();
+        </script>
+        """
+        st.components.v1.html(component, height=0)
+        st.session_state['stream_started'] = True
+#else:
+#    st.header("Advertised Streams (Cloud Announcements)")
+#    st.info("This feature requires backend support to list advertised streams.")
+    # Placeholder for future implementation
+    # Example: r = requests.get(f"{BACKEND_URL}/advertised_streams")
+    # if r.status_code == 200:
+    #     streams = r.json()
+    #     for s in streams:
+    #         st.write(s)
+    # else:
+    #     st.error("Could not fetch advertised streams.")
+
+log.basicConfig(
+    level=os.environ.get('LOG_LEVEL', log.DEBUG),
+    format='%(module)s.py:%(lineno)d %(levelname)s: %(message)s'
+)
--- a/src/auracast/server/multicast_server.py
+++ b/src/auracast/server/multicast_server.py
@@ -0,0 +1,421 @@
+import glob
+import os
+import logging as log
+import uuid
+import json
+import sys
+from datetime import datetime
+import asyncio
+import numpy as np
+from pydantic import BaseModel
+from fastapi import FastAPI, HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from auracast import multicast_control, auracast_config
+from aiortc import RTCPeerConnection, RTCSessionDescription, MediaStreamTrack
+import av
+import av.audio.layout
+import sounddevice as sd  # type: ignore
+from typing import Set, List, Dict, Any
+import traceback
+
+
+PTIME = 40 # TODO: seems to have no effect at all
+pcs: Set[RTCPeerConnection] = set()      # keep refs so they don’t GC early
+AUDIO_INPUT_DEVICES_CACHE: List[Dict[str, Any]] = []
+
+class Offer(BaseModel):
+    sdp: str
+    type: str
+
+def get_device_index_by_name(name: str):
+    """Return the device index for a given device name, or None if not found."""
+    for d in AUDIO_INPUT_DEVICES_CACHE:
+        if d["name"] == name:
+            return d["id"]
+    return None
+
+
+# Path to persist stream settings
+STREAM_SETTINGS_FILE = os.path.join(os.path.dirname(__file__), 'stream_settings.json')
+
+def load_stream_settings() -> dict:
+    """Load persisted stream settings if available."""
+    if os.path.exists(STREAM_SETTINGS_FILE):
+        try:
+            with open(STREAM_SETTINGS_FILE, 'r', encoding='utf-8') as f:
+                return json.load(f)
+        except Exception:
+            return {}
+    return {}
+
+def save_stream_settings(settings: dict):
+    """Save stream settings to disk."""
+    try:
+        with open(STREAM_SETTINGS_FILE, 'w', encoding='utf-8') as f:
+            json.dump(settings, f, indent=2)
+    except Exception as e:
+        log.error('Unable to persist stream settings: %s', e)
+
+
+app = FastAPI()
+
+# Allow CORS for frontend on localhost
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],  # You can restrict this to ["http://localhost:8501"] if you want
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Initialize global configuration
+global_config_group = auracast_config.AuracastConfigGroup()
+
+# Create multicast controller
+multicaster1: multicast_control.Multicaster | None = None
+multicaster2: multicast_control.Multicaster | None = None
+
+@app.post("/init")
+async def initialize(conf: auracast_config.AuracastConfigGroup):
+    """Initializes the primary broadcaster (multicaster1)."""
+    global global_config_group
+    global multicaster1
+    try:
+        if conf.transport == 'auto':
+            serial_devices = glob.glob('/dev/serial/by-id/*')
+            log.info('Found serial devices: %s', serial_devices)
+            for device in serial_devices:
+                if 'usb-ZEPHYR_Zephyr_HCI_UART_sample' in device:
+                    log.info('Using: %s', device)
+                    conf.transport = f'serial:{device},115200,rtscts'
+                    break
+        if conf.transport == 'auto':
+            raise HTTPException(status_code=500, detail='No suitable transport found.')
+        # Derive audio_mode and input_device from first BIG audio_source
+        first_source = conf.bigs[0].audio_source if conf.bigs else ''
+        if first_source.startswith('device:'):
+            audio_mode_persist = 'USB'
+            input_device_name = first_source.split(':', 1)[1] if ':' in first_source else None
+            # Map device name to current index for use with sounddevice
+            device_index = get_device_index_by_name(input_device_name) if input_device_name else None
+            # Patch config to use index for sounddevice (but persist name)
+            if device_index is not None:
+                for big in conf.bigs:
+                    if big.audio_source.startswith('device:'):
+                        big.audio_source = f'device:{device_index}'
+            else:
+                log.error(f"Device name '{input_device_name}' not found in current device list.")
+                raise HTTPException(status_code=400, detail=f"Audio device '{input_device_name}' not found.")
+        elif first_source == 'webrtc':
+            audio_mode_persist = 'Webapp'
+            input_device_name = None
+        elif first_source.startswith('file:'):
+            audio_mode_persist = 'Demo'
+            input_device_name = None
+        else:
+            audio_mode_persist = 'Network'
+            input_device_name = None
+        save_stream_settings({
+            'channel_names': [big.name for big in conf.bigs],
+            'languages': [big.language for big in conf.bigs],
+            'audio_mode': audio_mode_persist,
+            'input_device': input_device_name,
+            'program_info': [getattr(big, 'program_info', None) for big in conf.bigs],
+            'gain': [getattr(big, 'input_gain', 1.0) for big in conf.bigs],
+            'timestamp': datetime.utcnow().isoformat()
+        })
+        global_config_group = conf
+        if multicaster1 is not None:
+            try:
+                await multicaster1.shutdown()
+            except Exception:
+                log.warning("Failed to shutdown previous multicaster", exc_info=True)
+        log.info('Initializing multicaster1 with config:\n %s', conf.model_dump_json(indent=2))
+        multicaster1 = multicast_control.Multicaster(conf, conf.bigs)
+        await multicaster1.init_broadcast()
+        if any(big.audio_source.startswith("device:") or big.audio_source.startswith("file:") for big in conf.bigs):
+            log.info("Auto-starting streaming on multicaster1")
+            await multicaster1.start_streaming()
+    except Exception as e:
+        log.error("Exception in /init: %s", traceback.format_exc())
+        raise HTTPException(status_code=500, detail=str(e))
+
+@app.post("/init2")
+async def initialize2(conf: auracast_config.AuracastConfigGroup):
+    """Initializes the secondary broadcaster (multicaster2). Does NOT persist stream settings."""
+    global multicaster2
+    try:
+        if conf.transport == 'auto':
+            serial_devices = glob.glob('/dev/serial/by-id/*')
+            log.info('Found serial devices: %s', serial_devices)
+            for device in serial_devices:
+                if 'usb-ZEPHYR_Zephyr_HCI_UART_sample' in device:
+                    log.info('Using: %s', device)
+                    conf.transport = f'serial:{device},115200,rtscts'
+                    break
+        if conf.transport == 'auto':
+            raise HTTPException(status_code=500, detail='No suitable transport found.')
+        # Patch device name to index for sounddevice
+        for big in conf.bigs:
+            if big.audio_source.startswith('device:'):
+                device_name = big.audio_source.split(':', 1)[1]
+                device_index = get_device_index_by_name(device_name)
+                if device_index is not None:
+                    big.audio_source = f'device:{device_index}'
+                else:
+                    log.error(f"Device name '{device_name}' not found in current device list.")
+                    raise HTTPException(status_code=400, detail=f"Audio device '{device_name}' not found.")
+        log.info('Initializing multicaster2 with config:\n %s', conf.model_dump_json(indent=2))
+        multicaster2 = multicast_control.Multicaster(conf, conf.bigs)
+        await multicaster2.init_broadcast()
+        if any(big.audio_source.startswith("device:") or big.audio_source.startswith("file:") for big in conf.bigs):
+            log.info("Auto-starting streaming on multicaster2")
+            await multicaster2.start_streaming()
+    except Exception as e:
+        log.error("Exception in /init2: %s", traceback.format_exc())
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.post("/stream_lc3")
+async def send_audio(audio_data: dict[str, str]):
+    """Sends a block of pre-coded LC3 audio."""
+    if multicaster1 is None:
+        raise HTTPException(status_code=500, detail='Auracast endpoint was never intialized')
+    try:
+        for big in global_config_group.bigs:    
+            assert big.language in audio_data, HTTPException(status_code=500, detail='language len missmatch')
+            log.info('Received a send audio request for %s', big.language)
+            big.audio_source = audio_data[big.language].encode('latin-1') # TODO: use base64 encoding
+
+        multicaster1.big_conf = global_config_group.bigs
+        await multicaster1.start_streaming()
+        return {"status": "audio_sent"}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.post("/stop_audio")
+async def stop_audio():
+    """Stops streaming on both multicaster1 and multicaster2."""
+    try:
+        # First close any active WebRTC peer connections so their track loops finish cleanly
+        close_tasks = [pc.close() for pc in list(pcs)]
+        pcs.clear()
+        if close_tasks:
+            await asyncio.gather(*close_tasks, return_exceptions=True)
+
+        # Now shut down both multicasters and release audio devices
+        running = False
+        if multicaster1 is not None:
+            await multicaster1.stop_streaming()
+            await multicaster1.reset()  # Fully reset controller and advertising
+            running = True
+        if multicaster2 is not None:
+            await multicaster2.stop_streaming()
+            await multicaster2.reset()  # Fully reset controller and advertising
+            running = True
+
+        return {"status": "stopped", "was_running": running}
+    except Exception as e:
+        log.error("Exception in /stop_audio: %s", traceback.format_exc())
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+@app.get("/status")
+async def get_status():
+    """Gets the current status of the multicaster together with persisted stream info."""
+    status = multicaster1.get_status() if multicaster1 else {
+        'is_initialized': False,
+        'is_streaming': False,
+    }
+    status.update(load_stream_settings())
+    return status
+
+
+async def scan_audio_devices():
+    """Scans for available audio devices and updates the cache."""
+    global AUDIO_INPUT_DEVICES_CACHE
+    log.info("Scanning for audio input devices...")
+    try:
+        if sys.platform == 'linux':
+            log.info("Re-initializing sounddevice to scan for new devices")
+            sd._terminate()
+            sd._initialize()
+
+        devs = sd.query_devices()
+        inputs = [
+            dict(d, id=idx)
+            for idx, d in enumerate(devs)
+            if d.get("max_input_channels", 0) > 0
+        ]
+        log.info('Found %d audio input devices: %s', len(inputs), inputs)
+        AUDIO_INPUT_DEVICES_CACHE = inputs
+    except Exception:
+        log.error("Exception while scanning audio devices:", exc_info=True)
+        # Do not clear cache on error, keep the last known good list
+
+
+@app.on_event("startup")
+async def startup_event():
+    """Pre-scans audio devices on startup."""
+    await scan_audio_devices()
+
+
+@app.get("/audio_inputs")
+async def list_audio_inputs():
+    """Return available hardware audio input devices from cache (by name, for selection)."""
+    # Only expose name and id for frontend
+    return {"inputs": AUDIO_INPUT_DEVICES_CACHE}
+
+
+@app.post("/refresh_audio_inputs")
+async def refresh_audio_inputs():
+    """Triggers a re-scan of audio devices."""
+    await scan_audio_devices()
+    return {"status": "ok", "inputs": AUDIO_INPUT_DEVICES_CACHE}
+
+
+@app.post("/offer")
+async def offer(offer: Offer):
+    log.info("/offer endpoint called")
+
+    # If a previous PeerConnection is still alive, close it so we only ever keep one active.
+    if pcs:
+        log.info("Closing %d existing PeerConnection(s) before creating a new one", len(pcs))
+        close_tasks = [p.close() for p in list(pcs)]
+        await asyncio.gather(*close_tasks, return_exceptions=True)
+        pcs.clear()
+
+    pc = RTCPeerConnection()  # No STUN needed for localhost
+    pcs.add(pc)
+    id_ = uuid.uuid4().hex[:8]
+    log.info(f"{id_}: new PeerConnection")
+
+    # create directory for records - only for testing
+    os.makedirs("./records", exist_ok=True)
+
+    # Do NOT start the streamer yet – we'll start it lazily once we actually
+    # receive the first audio frame, ensuring WebRTCAudioInput is ready and
+    # avoiding race-conditions on restarts.
+    @pc.on("track")
+    async def on_track(track: MediaStreamTrack):
+        log.info(f"{id_}: track {track.kind} received")
+        try:
+            first = True
+            while True:
+                frame: av.audio.frame.AudioFrame = await track.recv()          # RTP audio frame (already decrypted)
+                if first:
+                    log.info(f"{id_}: frame layout={frame.layout}")
+                    log.info(f"{id_}: frame format={frame.format}")
+                    log.info(
+                        f"{id_}: frame sample_rate={frame.sample_rate}, samples_per_channel={frame.samples}, planes={frame.planes}"
+                    )
+                    # Lazily start the streamer now that we know a track exists.
+                    if multicaster1.streamer is None:
+                        await multicaster1.start_streaming()
+                        # Yield control so the Streamer coroutine has a chance to
+                        # create the WebRTCAudioInput before we push samples.
+                        await asyncio.sleep(0)
+                    first = False
+                    # in stereo case this is interleaved data format
+                    frame_array = frame.to_ndarray()
+                    log.info(f"array.shape{frame_array.shape}")
+                    log.info(f"array.dtype{frame_array.dtype}")
+                    log.info(f"frame.to_ndarray(){frame_array}")
+
+                    samples = frame_array.reshape(-1)
+                    log.info(f"samples.shape: {samples.shape}")
+
+                    if frame.layout.name == 'stereo':
+                        # Interleaved stereo: [L0, R0, L1, R1, ...]
+                        mono_array = samples[::2]  # Take left channel
+                    else:
+                        mono_array = samples
+
+                    log.info(f"mono_array.shape: {mono_array.shape}")
+
+
+                frame_array = frame.to_ndarray()
+
+                # Flatten in case it's (1, N) or (N,)
+                samples = frame_array.reshape(-1)
+
+                if frame.layout.name == 'stereo':
+                    # Interleaved stereo: [L0, R0, L1, R1, ...]
+                    mono_array = samples[::2]  # Take left channel
+                else:
+                    mono_array = samples
+
+                # Get current WebRTC audio input (streamer may have been restarted)
+                big0 = list(multicaster1.bigs.values())[0]
+                audio_input = big0.get('audio_input')
+                # Wait until the streamer has instantiated the WebRTCAudioInput
+                if audio_input is None or getattr(audio_input, 'closed', False):
+                    continue
+                # Feed mono PCM samples to the global WebRTC audio input
+                await audio_input.put_samples(mono_array.astype(np.int16))
+
+                # Save to WAV file - only for testing
+                # if not hasattr(pc, 'wav_writer'):
+                #     import wave
+                #     wav_path = f"./records/auracast_{id_}.wav"
+                #     pc.wav_writer = wave.open(wav_path, "wb")
+                #     pc.wav_writer.setnchannels(1)  # mono
+                #     pc.wav_writer.setsampwidth(2)  # 16-bit PCM
+                #     pc.wav_writer.setframerate(frame.sample_rate)
+
+                # pcm_data = mono_array.astype(np.int16).tobytes()
+                # pc.wav_writer.writeframes(pcm_data)
+
+
+        except Exception as e:
+            log.error(f"{id_}: Exception in on_track: {e}")
+        finally:
+            # Always close the wav file when the track ends or on error
+            if hasattr(pc, 'wav_writer'):
+                try:
+                    pc.wav_writer.close()
+                except Exception:
+                    pass
+                del pc.wav_writer
+
+    # --- SDP negotiation ---
+    log.info(f"{id_}: setting remote description")
+    await pc.setRemoteDescription(RTCSessionDescription(**offer.model_dump()))
+
+    log.info(f"{id_}: creating answer")
+    answer = await pc.createAnswer()
+    sdp = answer.sdp
+    # Insert a=ptime using the global PTIME variable
+    ptime_line = f"a=ptime:{PTIME}"
+    if "a=sendrecv" in sdp:
+        sdp = sdp.replace("a=sendrecv", f"a=sendrecv\n{ptime_line}")
+    else:
+        sdp += f"\n{ptime_line}"
+    new_answer = RTCSessionDescription(sdp=sdp, type=answer.type)
+    await pc.setLocalDescription(new_answer)
+    log.info(f"{id_}: sending answer with {ptime_line}")
+    return {"sdp": pc.localDescription.sdp,
+            "type": pc.localDescription.type}
+
+
+@app.post("/shutdown")
+async def shutdown():
+    """Stops broadcasting and releases all audio/Bluetooth resources."""
+    try:
+        await multicaster1.shutdown()
+        return {"status": "stopped"}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+if __name__ == '__main__':
+    import os
+    os.chdir(os.path.dirname(__file__))
+    import uvicorn
+    log.basicConfig( # for debug log level export LOG_LEVEL=DEBUG
+        level=os.environ.get('LOG_LEVEL', log.INFO),
+        format='%(module)s.py:%(lineno)d %(levelname)s: %(message)s'
+    )
+    # Bind to localhost only for security: prevents network access, only frontend on same machine can connect
+    uvicorn.run(app, host="127.0.0.1", port=5000)
--- a/src/auracast/server/provision_domain_hostname.sh
+++ b/src/auracast/server/provision_domain_hostname.sh
@@ -0,0 +1,89 @@
+#!/bin/bash
+# change_domain_hostname.sh
+# Safely change the system hostname and Avahi mDNS domain name, update /etc/hosts, restart Avahi,
+# and generate a per-device certificate signed by the CA.
+# Usage: sudo ./change_domain_hostname.sh <new_hostname> <new_domain> [--force]
+
+set -e
+
+if [ "$EUID" -ne 0 ]; then
+    echo "Please run as root."
+    exit 1
+fi
+
+if [ $# -lt 2 ]; then
+    echo "Usage: sudo $0 <new_hostname> <new_domain> [--force]"
+    exit 1
+fi
+
+NEW_HOSTNAME="$1"
+NEW_DOMAIN="$2"
+FORCE=0
+if [ "$3" == "--force" ]; then
+    FORCE=1
+fi
+
+# Validate hostname: single label, no dots
+if [[ "$NEW_HOSTNAME" == *.* ]]; then
+    echo "ERROR: Hostname must not contain dots."
+    exit 1
+fi
+
+# Set system hostname
+hostnamectl set-hostname "$NEW_HOSTNAME"
+echo "/etc/hostname set to $NEW_HOSTNAME."
+
+# Update /etc/hosts
+if grep -q '^127.0.1.1' /etc/hosts; then
+    sed -i "s/^127.0.1.1.*/127.0.1.1   $NEW_HOSTNAME/" /etc/hosts
+else
+    echo "127.0.1.1   $NEW_HOSTNAME" >> /etc/hosts
+fi
+echo "/etc/hosts updated."
+
+# Set Avahi domain name
+AVAHI_CONF="/etc/avahi/avahi-daemon.conf"
+sed -i "/^\[server\]/,/^\s*\[/{s/^\s*domain-name\s*=.*/domain-name=$NEW_DOMAIN/}" "$AVAHI_CONF"
+echo "Set Avahi domain name to $NEW_DOMAIN."
+
+# Restart Avahi
+echo "Restarting avahi-daemon..."
+systemctl restart avahi-daemon
+
+echo "Done. Hostname: $NEW_HOSTNAME, Avahi domain: $NEW_DOMAIN"
+
+# --- Per-device certificate logic ---
+CA_DIR="$(dirname "$0")/certs/ca"
+PER_DEVICE_DIR="$(dirname "$0")/certs/per_device/$NEW_HOSTNAME.$NEW_DOMAIN"
+mkdir -p "$PER_DEVICE_DIR"
+CA_CERT="$CA_DIR/ca_cert.pem"
+CA_KEY="$CA_DIR/ca_key.pem"
+DEVICE_CERT="$PER_DEVICE_DIR/device_cert.pem"
+DEVICE_KEY="$PER_DEVICE_DIR/device_key.pem"
+DEVICE_CSR="$PER_DEVICE_DIR/device.csr"
+SAN_CNF="$PER_DEVICE_DIR/san.cnf"
+
+if [ -f "$DEVICE_CERT" ] && [ $FORCE -eq 0 ]; then
+    echo "Per-device certificate already exists at $DEVICE_CERT. Use --force to regenerate."
+else
+    echo "Generating per-device key/cert for $NEW_HOSTNAME.$NEW_DOMAIN..."
+    openssl genrsa -out "$DEVICE_KEY" 4096
+    cat > "$SAN_CNF" <<EOF
+[req]
+distinguished_name = req_distinguished_name
+req_extensions = v3_req
+prompt = no
+
+[req_distinguished_name]
+CN = $NEW_HOSTNAME.$NEW_DOMAIN
+
+[v3_req]
+subjectAltName = @alt_names
+
+[alt_names]
+DNS.1 = $NEW_HOSTNAME.$NEW_DOMAIN
+EOF
+    openssl req -new -key "$DEVICE_KEY" -out "$DEVICE_CSR" -config "$SAN_CNF"
+    openssl x509 -req -in "$DEVICE_CSR" -CA "$CA_CERT" -CAkey "$CA_KEY" -CAcreateserial -out "$DEVICE_CERT" -days 7300 -extensions v3_req -extfile "$SAN_CNF"
+    echo "Per-device certificate generated at $DEVICE_CERT."
+fi
--- a/src/auracast/server/start_frontend_http.sh
+++ b/src/auracast/server/start_frontend_http.sh
@@ -0,0 +1,2 @@
+# Start Streamlit HTTP server (port 8500)
+poetry run streamlit run multicast_frontend.py --server.port 8500 --server.enableCORS false --server.enableXsrfProtection false --server.headless true --browser.gatherUsageStats false
--- a/src/auracast/server/start_frontend_https.sh
+++ b/src/auracast/server/start_frontend_https.sh
@@ -0,0 +1,37 @@
+#!/bin/bash
+# to bind this to 443, root privileges are required. start this like 
+# sudo -E PATH="$PATH" bash ./start_frontend_https.sh
+# Unified startup script: generates certs if needed, starts HTTPS Streamlit and HTTP->HTTPS redirector
+
+# Dynamically select per-device cert and key based on hostname and Avahi domain
+DEVICE_HOSTNAME=$(hostname)
+AVAHI_CONF="/etc/avahi/avahi-daemon.conf"
+AVAHI_DOMAIN=$(awk -F= '/^\s*domain-name\s*=/{gsub(/ /, "", $2); print $2}' "$AVAHI_CONF")
+if [ -z "$AVAHI_DOMAIN" ]; then
+  AVAHI_DOMAIN=local
+fi
+CERT_DIR="certs/per_device/${DEVICE_HOSTNAME}.${AVAHI_DOMAIN}"
+CERT="$CERT_DIR/device_cert.pem"
+KEY="$CERT_DIR/device_key.pem"
+CA_CERT="certs/ca/ca_cert.pem"
+
+if [ ! -f "$CERT" ] || [ ! -f "$KEY" ]; then
+  echo "ERROR: Device certificate or key not found in $CERT_DIR. Run provision_domain_hostname.sh first."
+  exit 1
+fi
+
+if [ ! -f "$CA_CERT" ]; then
+  echo "WARNING: CA certificate not found at $CA_CERT. HTTPS will work, but clients may not be able to import the CA."
+fi
+
+echo "CA cert: $CA_CERT"
+echo "Device cert: $CERT"
+echo "Device key: $KEY"
+echo "Using hostname: $DEVICE_HOSTNAME"
+echo "Using Avahi domain: $AVAHI_DOMAIN"
+
+# Path to poetry binary
+POETRY_BIN="/home/caster/.local/bin/poetry"
+
+# Start Streamlit HTTPS server (port 443)
+$POETRY_BIN run streamlit run multicast_frontend.py --server.port 443 --server.enableCORS false --server.enableXsrfProtection false --server.headless true --server.sslCertFile "$CERT" --server.sslKeyFile "$KEY" --browser.gatherUsageStats false
--- a/src/auracast/server/start_mdns.sh
+++ b/src/auracast/server/start_mdns.sh
@@ -0,0 +1,26 @@
+#!/bin/bash
+
+# Script to advertise the local device via mDNS for an HTTPS service.
+# This allows other clients on the network to discover this device
+# using its mDNS hostname (e.g., your-hostname.local) on the specified port.
+
+# Update: Advertise HTTPS service on port 443 (default)
+SERVICE_NAME="Auracast HTTPS Service"   # You can customize this name
+SERVICE_TYPE="_https._tcp"              # Standard type for HTTPS services
+SERVICE_PORT="443"                      # Port must match your HTTPS server (default 443)
+
+echo "Starting mDNS advertisement..."
+echo "Command: avahi-publish-service -v \"$SERVICE_NAME\" \"$SERVICE_TYPE\" \"$SERVICE_PORT\""
+
+avahi-publish-service -v "$SERVICE_NAME" "$SERVICE_TYPE" "$SERVICE_PORT"
+EXIT_STATUS=$?
+
+# This part will be reached if avahi-publish-service exits.
+if [ $EXIT_STATUS -eq 0 ]; then
+    echo "mDNS advertisement command finished with status 0."
+    echo "This might indicate an issue connecting to the avahi-daemon or a configuration problem."
+    echo "Please check for any messages above from avahi-publish-service itself."
+else
+    echo "mDNS advertisement command exited with status $EXIT_STATUS."
+    echo "This might be due to an error, or if you pressed Ctrl+C (which typically results in a non-zero status from signal termination)."
+fi
--- a/src/auracast/testdata/wave_particle_5min_de.wav
+++ b/src/auracast/testdata/wave_particle_5min_de.wav
--- a/src/auracast/testdata/wave_particle_5min_de_16kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_de_16kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_de_24kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_de_24kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_de_48kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_de_48kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_en.wav
+++ b/src/auracast/testdata/wave_particle_5min_en.wav
--- a/src/auracast/testdata/wave_particle_5min_en_16kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_en_16kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_en_24kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_en_24kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_en_48kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_en_48kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_es.wav
+++ b/src/auracast/testdata/wave_particle_5min_es.wav
--- a/src/auracast/testdata/wave_particle_5min_es_16kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_es_16kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_es_24kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_es_24kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_es_48kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_es_48kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_fr.wav
+++ b/src/auracast/testdata/wave_particle_5min_fr.wav
--- a/src/auracast/testdata/wave_particle_5min_fr_16kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_fr_16kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_fr_24kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_fr_24kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_fr_48kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_fr_48kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_it.wav
+++ b/src/auracast/testdata/wave_particle_5min_it.wav
--- a/src/auracast/testdata/wave_particle_5min_it_16kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_it_16kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_it_24kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_it_24kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_it_48kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_it_48kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_pl_16kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_pl_16kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_pl_24kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_pl_24kHz_mono.wav
--- a/src/auracast/testdata/wave_particle_5min_pl_48kHz_mono.wav
+++ b/src/auracast/testdata/wave_particle_5min_pl_48kHz_mono.wav
--- a/src/auracast/utils/network_audio_receiver.py
+++ b/src/auracast/utils/network_audio_receiver.py
@@ -0,0 +1,67 @@
+import asyncio
+import socket
+import logging
+import numpy as np
+from typing import AsyncGenerator
+
+class NetworkAudioReceiverUncoded:
+    """
+    Receives PCM audio over UDP and provides an async generator interface for uncoded PCM frames.
+    Combines network receiving and input logic for use with Auracast streamer.
+    """
+    def __init__(self, port: int = 50007, samplerate: int = 16000, channels: int = 1, chunk_size: int = 1024):
+        self.port = port
+        self.samplerate = samplerate
+        self.channels = channels
+        self.chunk_size = chunk_size
+        self.sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
+        self.sock.bind(('0.0.0.0', self.port))
+        self.sock.setblocking(False)
+        self._running = False
+        # Reduce queue size for lower latency (less buffering)
+        self._queue = asyncio.Queue(maxsize=2)  # Was 20
+
+    async def receive(self):
+        self._running = True
+        logging.info(f"NetworkAudioReceiver listening on UDP port {self.port}")
+        try:
+            while self._running:
+                try:
+                    data, _ = await asyncio.get_event_loop().sock_recvfrom(self.sock, self.chunk_size * 2)
+                    await self._queue.put(data)
+                except Exception:
+                    await asyncio.sleep(0.01)
+        finally:
+            self.sock.close()
+            logging.info("NetworkAudioReceiver stopped.")
+
+    def stop(self):
+        self._running = False
+
+    async def open(self):
+        # Dummy PCM format object
+        class PCMFormat:
+            channels = self.channels
+            sample_type = 'int16'
+            sample_rate = self.samplerate
+        return PCMFormat()
+
+    def rewind(self):
+        pass  # Not supported for live network input
+
+    async def frames(self, samples_per_frame: int) -> AsyncGenerator[np.ndarray, None]:
+        bytes_per_frame = samples_per_frame * 2 * self.channels  # 2 bytes for int16
+        buf = bytearray()
+        while True:
+            data = await self._queue.get()
+            # Optional: log queue size for latency debugging
+            # logging.debug(f'NetworkAudioReceiver queue size: {self._queue.qsize()}')
+            if data is None:
+                break
+            buf.extend(data)
+            while len(buf) >= bytes_per_frame:
+                frame = np.frombuffer(buf[:bytes_per_frame], dtype=np.int16).reshape(-1, self.channels)
+                # Optional: log when a frame is yielded
+                # logging.debug(f'Yielding frame of shape {frame.shape}')
+                yield frame
+                buf = buf[bytes_per_frame:]
--- a/src/auracast/utils/sounddevice_utils.py
+++ b/src/auracast/utils/sounddevice_utils.py
@@ -0,0 +1,141 @@
+import sounddevice as sd
+import os, re, json, subprocess
+
+def devices_by_backend(backend_name: str):
+    hostapis = sd.query_hostapis()          # list of host APIs
+    # find the host API index by (case-insensitive) name match
+    try:
+        hostapi_idx = next(
+            i for i, ha in enumerate(hostapis)
+            if backend_name.lower() in ha['name'].lower()
+        )
+    except StopIteration:
+        raise ValueError(f"No host API matching {backend_name!r}. "
+                         f"Available: {[ha['name'] for ha in hostapis]}")
+    # return (global_index, device_dict) pairs filtered by that host API
+    return [(i, d) for i, d in enumerate(sd.query_devices())
+            if d['hostapi'] == hostapi_idx]
+
+def _pa_like_hostapi_index():
+    for i, ha in enumerate(sd.query_hostapis()):
+        if any(k in ha["name"] for k in ("PipeWire", "PulseAudio")):
+            return i
+    raise RuntimeError("PipeWire/PulseAudio host API not present in PortAudio.")
+
+def _pw_dump():
+    return json.loads(subprocess.check_output(["pw-dump"]))
+
+def _sd_refresh():
+    """Force PortAudio to re-enumerate devices on next query.
+
+    sounddevice/PortAudio keeps a static device list after initialization.
+    Terminating here ensures that subsequent sd.query_* calls re-initialize
+    and see newly added devices (e.g., AES67 nodes created after start).
+    """
+    sd._terminate()  # private API, acceptable for runtime refresh
+    sd._initialize()
+
+def _sd_matches_from_names(pa_idx, names):
+    names_l = {n.lower() for n in names if n}
+    out = []
+    for i, d in enumerate(sd.query_devices()):
+        if d["hostapi"] != pa_idx or d["max_input_channels"] <= 0:
+            continue
+        dn = d["name"].lower()
+        if any(n in dn for n in names_l):
+            out.append((i, d))
+    return out
+
+def list_usb_pw_inputs():
+    """
+    Return [(device_index, device_dict), ...] for PipeWire **input** nodes
+    backed by **USB** devices (excludes monitor sources).
+    """
+    # Refresh PortAudio so we see newly added nodes before mapping
+    _sd_refresh()
+    pa_idx = _pa_like_hostapi_index()
+    pw = _pw_dump()
+
+    # Map device.id -> device.bus ("usb"/"pci"/"platform"/"network"/...)
+    device_bus = {}
+    for obj in pw:
+        if obj.get("type") == "PipeWire:Interface:Device":
+            props = (obj.get("info") or {}).get("props") or {}
+            device_bus[obj["id"]] = (props.get("device.bus") or "").lower()
+
+    # Collect names/descriptions of USB input nodes
+    usb_input_names = set()
+    for obj in pw:
+        if obj.get("type") != "PipeWire:Interface:Node":
+            continue
+        props = (obj.get("info") or {}).get("props") or {}
+        media = (props.get("media.class") or "").lower()
+        if "source" not in media and "stream/input" not in media:
+            continue
+        # skip monitor sources ("Monitor of ..." or *.monitor)
+        nname = (props.get("node.name") or "").lower()
+        ndesc = (props.get("node.description") or "").lower()
+        if ".monitor" in nname or "monitor" in ndesc:
+            continue
+        bus = (props.get("device.bus") or device_bus.get(props.get("device.id")) or "").lower()
+        if bus == "usb":
+            usb_input_names.add(props.get("node.description") or props.get("node.name"))
+
+    # Map to sounddevice devices on PipeWire host API
+    return _sd_matches_from_names(pa_idx, usb_input_names)
+
+def list_network_pw_inputs():
+    """
+    Return [(device_index, device_dict), ...] for PipeWire **input** nodes that
+    look like network/AES67/RTP sources (excludes monitor sources).
+    """
+    # Refresh PortAudio so we see newly added nodes before mapping
+    _sd_refresh()
+    pa_idx = _pa_like_hostapi_index()
+    pw = _pw_dump()
+
+    network_input_names = set()
+    for obj in pw:
+        if obj.get("type") != "PipeWire:Interface:Node":
+            continue
+        props = (obj.get("info") or {}).get("props") or {}
+        media = (props.get("media.class") or "").lower()
+        if "source" not in media and "stream/input" not in media:
+            continue
+        nname = (props.get("node.name") or "")
+        ndesc = (props.get("node.description") or "")
+        # skip monitor sources
+        if ".monitor" in nname.lower() or "monitor" in ndesc.lower():
+            continue
+
+        # Heuristics for network/AES67/RTP
+        text = (nname + " " + ndesc).lower()
+        media_name = (props.get("media.name") or "").lower()
+        node_group = (props.get("node.group") or "").lower()
+        # Presence flags/keys that strongly indicate network RTP/AES67 sources
+        node_network_flag = bool(props.get("node.network"))
+        has_rtp_keys = any(k in props for k in (
+            "rtp.session", "rtp.source.ip", "rtp.source.port", "rtp.fmtp", "rtp.rate"
+        ))
+        has_sess_keys = any(k in props for k in (
+            "sess.name", "sess.media", "sess.latency.msec"
+        ))
+        is_network = (
+            (props.get("device.bus") or "").lower() == "network" or
+            node_network_flag or
+            "rtp" in media_name or
+            any(k in text for k in ("rtp", "sap", "aes67", "network", "raop", "airplay")) or
+            has_rtp_keys or
+            has_sess_keys or
+            ("pipewire.ptp" in node_group)
+        )
+        if is_network:
+            network_input_names.add(ndesc or nname)
+
+    return _sd_matches_from_names(pa_idx, network_input_names)
+
+# Example usage:
+# for i, d in list_usb_pw_inputs():
+#     print(f"USB IN  {i}: {d['name']}  in={d['max_input_channels']}")
+# for i, d in list_network_pw_inputs():
+#     print(f"NET IN  {i}: {d['name']}  in={d['max_input_channels']}")
--- a/src/auracast/utils/webrtc_audio_input.py
+++ b/src/auracast/utils/webrtc_audio_input.py
@@ -0,0 +1,42 @@
+import asyncio
+import numpy as np
+import logging
+
+class WebRTCAudioInput:
+    """
+    Buffer PCM samples from WebRTC and provide an async generator interface for chunked frames.
+    """
+    def __init__(self):
+        self.buffer = np.array([], dtype=np.int16)
+        self.lock = asyncio.Lock()
+        self.data_available = asyncio.Event()
+        self.closed = False
+
+    async def frames(self, frame_size: int):
+        """
+        Async generator yielding exactly frame_size samples as numpy arrays.
+        """
+        while not self.closed:
+            async with self.lock:
+                if len(self.buffer) >= frame_size:
+                    chunk = self.buffer[:frame_size]
+                    self.buffer = self.buffer[frame_size:]
+                    logging.debug(f"WebRTCAudioInput: Yielding {frame_size} samples, buffer now has {len(self.buffer)} samples remaining.")
+                    yield chunk
+                else:
+                    self.data_available.clear()
+            await self.data_available.wait()
+
+    async def put_samples(self, samples: np.ndarray):
+        """
+        Add new PCM samples (1D np.int16 array, mono) to the buffer.
+        """
+        async with self.lock:
+            self.buffer = np.concatenate([self.buffer, samples])
+            logging.debug(f"WebRTCAudioInput: Added {len(samples)} samples, buffer now has {len(self.buffer)} samples.")
+            self.data_available.set()
+
+    async def close(self):
+        """Mark the input closed so frames() stops yielding."""
+        self.closed = True
+        self.data_available.set()
--- a/src/scripts/list_pw_nodes.py
+++ b/src/scripts/list_pw_nodes.py
@@ -0,0 +1,24 @@
+import sounddevice as sd, pprint
+from auracast.utils.sounddevice_utils import devices_by_backend, list_usb_pw_inputs, list_network_pw_inputs
+
+
+print("PortAudio library:", sd._libname)
+print("PortAudio version:", sd.get_portaudio_version())
+print("\nHost APIs:")
+pprint.pprint(sd.query_hostapis())
+print("\nDevices:")
+pprint.pprint(sd.query_devices())
+
+# Example: only PulseAudio devices on Linux
+print("\nOnly PulseAudio devices:")
+for i, d in devices_by_backend("PulseAudio"):
+    print(f"{i}: {d['name']}  in={d['max_input_channels']} out={d['max_output_channels']}")
+
+
+print("Network pw inputs:")
+for i, d in list_network_pw_inputs():
+    print(f"{i}: {d['name']}  in={d['max_input_channels']}")
+
+print("USB pw inputs:")
+for i, d in list_usb_pw_inputs():
+    print(f"{i}: {d['name']}  in={d['max_input_channels']}")
--- a/src/service/aes67/90-pipewire-aes67-ptp.rules
+++ b/src/service/aes67/90-pipewire-aes67-ptp.rules
@@ -0,0 +1,6 @@
+# This file was installed by PipeWire project for its pipewire-aes67
+#
+# This is used to give readonly access to the PTP hardware clock.
+# PipeWire uses this to follow PTP grandmaster time. It should be synced by another service
+#
+KERNEL=="ptp[0-9]*", MODE="0644"
--- a/src/service/aes67/pipewire-aes67.conf
+++ b/src/service/aes67/pipewire-aes67.conf
@@ -0,0 +1,114 @@
+# AES67 config file for PipeWire version "1.2.7" #
+#
+# Copy and edit this file in /etc/pipewire for system-wide changes
+# or in ~/.config/pipewire for local changes.
+#
+# It is also possible to place a file with an updated section in
+# /etc/pipewire/pipewire-aes67.conf.d/ for system-wide changes or in
+# ~/.config/pipewire/pipewire-aes67.conf.d/ for local changes.
+#
+
+context.properties = {
+    ## Configure properties in the system.
+    default.clock.rate          = 48000
+    default.clock.allowed-rates = [ 48000 ]
+    # Enforce 3ms quantum on this AES67 PipeWire instance
+    clock.force-quantum         = 144
+    default.clock.quantum       = 144
+    #mem.warn-mlock  = false
+    #mem.allow-mlock = true
+    #mem.mlock-all   = false
+    #log.level       = 2
+
+    #default.clock.quantum-limit = 8192
+}
+
+context.spa-libs = {
+    support.*       = support/libspa-support
+}
+
+context.objects = [
+    # An example clock reading from /dev/ptp0. You can also specify the network interface name,
+    # pipewire will query the interface for the current active PHC index. Another option is to
+    # sync the ptp clock to CLOCK_TAI and then set clock.id = tai, keep in mind that tai may
+    # also be synced by a NTP client.
+    # The precedence is: device, interface, id
+    { factory = spa-node-factory
+        args = {
+            factory.name    = support.node.driver
+            node.name       = PTP0-Driver
+            node.group      = pipewire.ptp0
+            # This driver should only be used for network nodes marked with group
+            priority.driver = 100000
+            clock.name      = "clock.system.ptp0"
+            ### Please select the PTP hardware clock here
+            # Interface name is the preferred method of specifying the PHC
+            clock.interface = "eth0"
+            #clock.device    = "/dev/ptp0"
+            #clock.id        = tai
+            # Lower this in case of periodic out-of-sync
+            resync.ms       = 1.5
+            object.export   = true
+        }
+    }
+]
+
+context.modules = [
+    { name = libpipewire-module-rt
+        args = {
+            nice.level   = -11
+            #rt.prio      = 83
+            #rt.time.soft = -1
+            #rt.time.hard = -1
+        }
+        flags = [ ifexists nofail ]
+    }
+    { name = libpipewire-module-protocol-native }
+    { name = libpipewire-module-client-node }
+    { name = libpipewire-module-spa-node-factory }
+    { name = libpipewire-module-adapter }
+    { name = libpipewire-module-rtp-sap
+        args = {
+            ### Please select the interface here
+            local.ifname = eth0
+            sap.ip = 239.255.255.255
+            sap.port = 9875
+            net.ttl = 32
+            net.loop = false
+            # If you use another PTPv2 daemon supporting management
+            # messages over a UNIX socket, specify its path here
+            ptp.management-socket = "/var/run/ptp4lro"
+
+            stream.rules = [
+                {
+                    matches = [
+                        {
+                            rtp.session = "~.*"
+                        }
+                    ]
+                    actions = {
+                        create-stream = {
+                            node.virtual = false
+                            media.class = "Audio/Source"
+                            device.api = aes67
+                            # You can adjust the latency buffering here. Use integer values only
+                            sess.latency.msec = 6
+                            node.latency = "144/48000"
+                            node.group = pipewire.ptp0
+                        }
+                    }
+                },
+                {
+                    matches = [
+                        {
+                            sess.sap.announce = true
+                        }
+                    ]
+                    actions = {
+                        announce-stream = {}
+                    }
+               }
+            ]
+        }
+    },
+]
--- a/src/service/aes67/ptp_aes67_1.conf
+++ b/src/service/aes67/ptp_aes67_1.conf
@@ -0,0 +1,19 @@
+[global]
+priority1 255
+priority2 254
+# Lower = more likely to become Grandmaster. Keep the same on both for "either can be master".
+domainNumber 0          
+# Default domain
+logSyncInterval -3      
+# AES67 profile: Sync messages every 125ms
+logAnnounceInterval 1   
+# Announce messages every 2s (AES67 default)
+logMinDelayReqInterval 0
+dscp_event 46           
+# QoS for event messages
+dscp_general 0          
+# QoS for general messages
+step_threshold 1        
+# Fast convergence on time jumps
+
+tx_timestamp_timeout 20
--- a/src/service/auracast-frontend.service
+++ b/src/service/auracast-frontend.service
@@ -0,0 +1,15 @@
+[Unit]
+Description=Auracast Frontend HTTPS Server
+# Ensure backend is running as a user service before starting frontend
+After=auracast-server.service network.target
+Wants=auracast-server.service
+
+[Service]
+Type=simple
+WorkingDirectory=/home/caster/bumble-auracast/src/auracast/server
+ExecStart=/home/caster/bumble-auracast/src/auracast/server/start_frontend_https.sh
+Restart=on-failure
+Environment=LOG_LEVEL=INFO
+
+[Install]
+WantedBy=multi-user.target
--- a/src/service/auracast-script.service
+++ b/src/service/auracast-script.service
@@ -0,0 +1,14 @@
+[Unit]
+Description=Auracast Multicast Script
+After=network.target
+
+[Service]
+Type=simple
+WorkingDirectory=/home/caster/bumble-auracast
+ExecStart=/home/caster/.local/bin/poetry run python src/auracast/multicast_script.py
+Restart=on-failure
+Environment=PYTHONUNBUFFERED=1
+Environment=LOG_LEVEL=INFO
+
+[Install]
+WantedBy=default.target
--- a/src/service/auracast-server.service
+++ b/src/service/auracast-server.service
@@ -0,0 +1,14 @@
+[Unit]
+Description=Auracast Backend Server
+After=network.target
+
+[Service]
+Type=simple
+WorkingDirectory=/home/caster/bumble-auracast
+ExecStart=/home/caster/.local/bin/poetry run python src/auracast/server/multicast_server.py
+Restart=on-failure
+Environment=PYTHONUNBUFFERED=1
+Environment=LOG_LEVEL=INFO
+
+[Install]
+WantedBy=default.target
--- a/src/service/pipewire-aes67.service
+++ b/src/service/pipewire-aes67.service
@@ -0,0 +1,11 @@
+[Unit]
+Description=PipeWire AES67 Service
+After=network.target
+
+[Service]
+Type=simple
+ExecStart=/usr/bin/pipewire-aes67  -c /home/caster/bumble-auracast/src/service/aes67/pipewire-aes67.conf
+Restart=on-failure
+
+[Install]
+WantedBy=default.target
--- a/src/service/pipewire/99-lowlatency.conf
+++ b/src/service/pipewire/99-lowlatency.conf
@@ -0,0 +1,13 @@
+context.properties = {
+    default.clock.rate          = 48000
+    default.clock.allowed-rates = [ 48000 ]
+    default.clock.quantum       = 144        # 144/48000 = 3.0 ms
+    default.clock.min-quantum   = 32
+    default.clock.max-quantum   = 256
+}
+
+stream.properties = {
+    # Prefer to let specific nodes (e.g. AES67) or clients set node.latency.
+    node.latency     = "144/48000"
+    resample.quality = 0
+}
--- a/src/service/ptp_aes67.service
+++ b/src/service/ptp_aes67.service
@@ -0,0 +1,13 @@
+[Unit]
+Description=PTP AES67 Service
+After=network.target
+
+[Service]
+Type=simple
+ExecStart=/usr/sbin/ptp4l -i eth0 -f /home/caster/bumble-auracast/src/service/aes67/ptp_aes67_1.conf
+Restart=on-failure
+StandardOutput=journal
+StandardError=journal
+
+[Install]
+WantedBy=multi-user.target
--- a/src/service/status_auracast_services.sh
+++ b/src/service/status_auracast_services.sh
@@ -0,0 +1,84 @@
+#!/bin/bash
+set -euo pipefail
+
+# Print status of relevant services and warn if mutually exclusive ones are running.
+# Utilities (network audio):
+#   - ptp_aes67.service (system)
+#   - pipewire-aes67.service (user)
+# App services (mutually exclusive groups):
+#   - auracast-script.service (user)
+#   - auracast-server.service (prefer user; also check system)
+#   - auracast-frontend.service (system)
+
+print_status() {
+  local scope="$1"  # "system" or "user"
+  local unit="$2"
+  local act="unknown"
+  local ena="unknown"
+  if [[ "$scope" == "user" ]]; then
+    act=$(systemctl --user is-active "$unit" 2>/dev/null || true)
+    ena=$(systemctl --user is-enabled "$unit" 2>/dev/null || true)
+  else
+    act=$(systemctl is-active "$unit" 2>/dev/null || true)
+    ena=$(systemctl is-enabled "$unit" 2>/dev/null || true)
+  fi
+  act="${act:-unknown}"
+  ena="${ena:-unknown}"
+  printf " - %-24s [%s] active=%-10s enabled=%s\n" "$unit" "$scope" "$act" "$ena"
+}
+
+is_active() {
+  local scope="$1" unit="$2"
+  if [[ "$scope" == "user" ]]; then
+    systemctl --user is-active "$unit" &>/dev/null && echo yes || echo no
+  else
+    systemctl is-active "$unit" &>/dev/null && echo yes || echo no
+  fi
+}
+
+hr() { printf '\n%s\n' "----------------------------------------"; }
+
+printf "AURACAST SERVICE STATUS\n"
+hr
+
+printf "Utilities (required for network audio)\n"
+print_status system ptp_aes67.service
+print_status user   pipewire-aes67.service
+
+PTP_ACTIVE=$(is_active system ptp_aes67.service)
+PW_ACTIVE=$(is_active user   pipewire-aes67.service)
+
+if [[ "$PTP_ACTIVE" == "yes" && "$PW_ACTIVE" == "yes" ]]; then
+  echo " ✓ Utilities ready for AES67/network audio"
+else
+  echo " ! Utilities not fully active (AES67/network audio may not work)"
+fi
+
+hr
+printf "Application services (mutually exclusive)\n"
+print_status user   auracast-script.service
+print_status user   auracast-server.service
+print_status system auracast-server.service
+print_status system auracast-frontend.service
+
+SCRIPT_ACTIVE=$(is_active user   auracast-script.service)
+SERVER_USER_ACTIVE=$(is_active user   auracast-server.service)
+SERVER_SYS_ACTIVE=$(is_active system auracast-server.service)
+FRONT_ACTIVE=$(is_active system auracast-frontend.service)
+
+# Consider server active if either user or system instance is active
+if [[ "$SERVER_USER_ACTIVE" == "yes" || "$SERVER_SYS_ACTIVE" == "yes" ]]; then
+  SERVER_ACTIVE=yes
+else
+  SERVER_ACTIVE=no
+fi
+
+if [[ "$SCRIPT_ACTIVE" == "yes" && ( "$SERVER_ACTIVE" == "yes" || "$FRONT_ACTIVE" == "yes" ) ]]; then
+  echo " ! WARNING: 'auracast-script' and 'server/frontend' are running together (mutually exclusive)."
+fi
+
+hr
+printf "Hints\n"
+echo " - Follow logs (user): journalctl --user -u auracast-script.service -f"
+echo " - Follow logs (server): journalctl --user -u auracast-server.service -f || journalctl -u auracast-server.service -f"
+echo " - Frontend logs: journalctl -u auracast-frontend.service -f"
--- a/src/service/stop_aes67.sh
+++ b/src/service/stop_aes67.sh
@@ -0,0 +1,21 @@
+#!/bin/bash
+set -e
+
+# This script stops and disables the AES67 services
+# Requires sudo privileges
+
+# Stop services
+sudo systemctl stop ptp_aes67.service
+systemctl --user stop pipewire-aes67.service
+
+# Disable services from starting on boot
+sudo systemctl disable ptp_aes67.service
+systemctl --user disable pipewire-aes67.service
+
+echo "\n--- ptp_aes67.service status ---"
+sudo systemctl status ptp_aes67.service --no-pager
+
+echo "\n--- pipewire-aes67.service status (user) ---"
+systemctl --user status pipewire-aes67.service --no-pager
+
+echo "AES67 services stopped, disabled, and status printed successfully."
--- a/src/service/stop_auracast_script.sh
+++ b/src/service/stop_auracast_script.sh
@@ -0,0 +1,15 @@
+#!/bin/bash
+set -e
+
+# This script stops and disables the auracast-script user service
+
+# Stop service
+systemctl --user stop auracast-script.service || true
+
+# Disable service from starting on login
+systemctl --user disable auracast-script.service || true
+
+echo "\n--- auracast-script.service status (user) ---"
+systemctl --user status auracast-script.service --no-pager || true
+
+echo "auracast-script service stopped, disabled, and status printed successfully."
--- a/src/service/stop_server_and_frontend.sh
+++ b/src/service/stop_server_and_frontend.sh
@@ -0,0 +1,22 @@
+# This script stops and disables the auracast-server and auracast-frontend services
+# Requires sudo privileges
+
+echo "Stopping auracast-server.service..."
+systemctl --user stop auracast-server.service
+
+echo "Disabling auracast-server.service (user)..."
+systemctl --user disable auracast-server.service
+
+echo "Stopping auracast-frontend.service ..."
+sudo systemctl stop auracast-frontend.service
+
+echo "Disabling auracast-frontend.service ..."
+sudo systemctl disable auracast-frontend.service
+
+echo "\n--- auracast-server.service status ---"
+systemctl --user status auracast-server.service --no-pager
+
+echo "\n--- auracast-frontend.service status ---"
+sudo systemctl status auracast-frontend.service --no-pager
+
+echo "auracast-server and auracast-frontend services stopped, disabled, and status printed successfully."
--- a/src/service/update_and_run_auracast_script.sh
+++ b/src/service/update_and_run_auracast_script.sh
@@ -0,0 +1,23 @@
+#!/bin/bash
+set -e
+
+# This script installs, enables, and restarts the auracast-script user service
+# No sudo is required as it is a user service
+
+# Copy user service file for auracast-script
+mkdir -p /home/caster/.config/systemd/user
+cp /home/caster/bumble-auracast/src/service/auracast-script.service /home/caster/.config/systemd/user/auracast-script.service
+
+# Reload systemd to recognize new/updated services
+systemctl --user daemon-reload
+
+# Enable service to start on user login
+systemctl --user enable auracast-script.service
+
+# Restart service
+systemctl --user restart auracast-script.service
+
+echo "\n--- auracast-script.service status (user) ---"
+systemctl --user status auracast-script.service --no-pager
+
+echo "auracast-script service updated, enabled, restarted, and status printed successfully."
--- a/src/service/update_and_run_pw_aes67.sh
+++ b/src/service/update_and_run_pw_aes67.sh
@@ -0,0 +1,41 @@
+#!/bin/bash
+set -e
+
+# This script installs, enables, and restarts the AES67 services
+# Requires sudo privileges
+
+# Copy system service file for ptp_aes67
+sudo cp /home/caster/bumble-auracast/src/service/ptp_aes67.service /etc/systemd/system/ptp_aes67.service
+
+# Copy user service file for pipewire-aes67
+mkdir -p /home/caster/.config/systemd/user
+cp /home/caster/bumble-auracast/src/service/pipewire-aes67.service /home/caster/.config/systemd/user/pipewire-aes67.service
+
+# Install PipeWire user config to persist 3ms@48kHz (default.clock.quantum=144)
+mkdir -p /home/caster/.config/pipewire/pipewire.conf.d
+cp /home/caster/bumble-auracast/src/service/pipewire/99-lowlatency.conf /home/caster/.config/pipewire/pipewire.conf.d/99-lowlatency.conf
+
+# Reload systemd to recognize new/updated services
+sudo systemctl daemon-reload
+systemctl --user daemon-reload
+
+# Enable services to start on boot
+sudo systemctl enable ptp_aes67.service
+systemctl --user enable pipewire-aes67.service
+
+# Restart services
+systemctl --user restart pipewire.service pipewire-pulse.service
+sudo systemctl restart ptp_aes67.service
+systemctl --user restart pipewire-aes67.service
+
+echo "\n--- pipewire.service status (user) ---"
+systemctl --user status pipewire.service --no-pager
+
+echo "\n--- ptp_aes67.service status ---"
+sudo systemctl status ptp_aes67.service --no-pager
+
+echo "\n--- pipewire-aes67.service status (user) ---"
+systemctl --user status pipewire-aes67.service --no-pager
+
+
+echo "AES67 services updated, enabled, restarted, and status printed successfully."
--- a/src/service/update_and_run_server_and_frontend.sh
+++ b/src/service/update_and_run_server_and_frontend.sh
@@ -0,0 +1,33 @@
+#!/bin/bash
+set -e
+
+# This script installs, enables, and restarts the auracast-server and auracast-frontend services
+# Requires sudo privileges
+
+# Copy system service file for frontend
+sudo cp /home/caster/bumble-auracast/src/service/auracast-frontend.service /etc/systemd/system/auracast-frontend.service
+
+# Copy user service file for backend (now using WantedBy=default.target)
+mkdir -p /home/caster/.config/systemd/user
+cp /home/caster/bumble-auracast/src/service/auracast-server.service /home/caster/.config/systemd/user/auracast-server.service
+
+# Reload systemd for frontend
+sudo systemctl daemon-reload
+# Reload user systemd for server
+systemctl --user daemon-reload
+
+# Enable frontend to start on boot (system)
+sudo systemctl enable auracast-frontend.service
+# Enable server to start on boot (user)
+systemctl --user enable auracast-server.service
+
+# Restart both
+sudo systemctl restart auracast-frontend.service
+systemctl --user restart auracast-server.service
+
+echo "\n--- auracast-frontend.service status ---"
+sudo systemctl status auracast-frontend.service --no-pager
+
+echo "\n--- auracast-server.service status---"
+systemctl --user status auracast-server.service --no-pager
+echo "auracast-server and auracast-frontend services updated, enabled, restarted, and status printed successfully."
--- a/tests/test_audio_device_io.py
+++ b/tests/test_audio_device_io.py
@@ -0,0 +1,44 @@
+"""Utility to diagnose Bumble SoundDeviceAudioInput compatibility.
+
+Run inside the project venv:
+    python -m tests.usb_audio_diag [rate]
+It enumerates all PortAudio input devices and tries to open each with Bumble's
+create_audio_input using the URI pattern `device:<index>` with an explicit input_format of `int16le,<rate>,1`.
+"""
+from __future__ import annotations
+import asyncio
+import sys
+
+import sounddevice as sd  # type: ignore
+from bumble.audio import io as audio_io  # type: ignore
+
+RATE = int(sys.argv[1]) if len(sys.argv) > 1 else 48000
+
+
+aSYNC = asyncio.run
+
+
+async def try_device(index: int, rate: int = 48000) -> None:
+    input_uri = f"device:{index}"
+    try:
+        audio_input = await audio_io.create_audio_input(input_uri, f"int16le,{rate},1")
+        fmt = await audio_input.open()
+        print(f"\033[32m✔︎ {input_uri} -> {fmt.channels}ch @ {fmt.sample_rate}Hz\033[0m")
+        if hasattr(audio_input, "aclose"):
+            await audio_input.aclose()
+    except Exception as exc:  # pylint: disable=broad-except
+        print(f"\033[31m✗ {input_uri}: {exc}\033[0m")
+
+
+async def main() -> None:
+    print(f"Trying PortAudio input devices with rate {RATE} Hz\n")
+    for idx, dev in enumerate(sd.query_devices()):
+        if dev["max_input_channels"] > 0 and "(hw:" in dev["name"].lower():
+            name = dev["name"]
+            print(f"[{idx}] {name}")
+            await try_device(idx, RATE)
+            print()
+
+
+if __name__ == "__main__":
+    aSYNC(main())