wip

2026-04-16 00:25:31 +00:00 · 2023-05-12 16:25:46 -07:00
parent e6a623db93
commit 7b7ef85b14
9 changed files with 1342 additions and 66 deletions
--- a/apps/speaker/logo.svg
+++ b/apps/speaker/logo.svg
@@ -0,0 +1,42 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?> <!-- Created with Vectornator for iOS (http://vectornator.io/) --><!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<svg height="100%" style="fill-rule:nonzero;clip-rule:evenodd;stroke-linecap:round;stroke-linejoin:round;" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns="http://www.w3.org/2000/svg" xml:space="preserve" width="100%" xmlns:vectornator="http://vectornator.io" version="1.1" viewBox="0 0 745 744.634">
+<metadata>
+<vectornator:setting key="DimensionsVisible" value="1"/>
+<vectornator:setting key="PencilOnly" value="0"/>
+<vectornator:setting key="SnapToPoints" value="0"/>
+<vectornator:setting key="OutlineMode" value="0"/>
+<vectornator:setting key="CMYKEnabledKey" value="0"/>
+<vectornator:setting key="RulersVisible" value="1"/>
+<vectornator:setting key="SnapToEdges" value="0"/>
+<vectornator:setting key="GuidesVisible" value="1"/>
+<vectornator:setting key="DisplayWhiteBackground" value="0"/>
+<vectornator:setting key="doHistoryDisabled" value="0"/>
+<vectornator:setting key="SnapToGuides" value="1"/>
+<vectornator:setting key="TimeLapseWatermarkDisabled" value="0"/>
+<vectornator:setting key="Units" value="Pixels"/>
+<vectornator:setting key="DynamicGuides" value="0"/>
+<vectornator:setting key="IsolateActiveLayer" value="0"/>
+<vectornator:setting key="SnapToGrid" value="0"/>
+</metadata>
+<defs/>
+<g id="Layer 1" vectornator:layerName="Layer 1">
+<path stroke="#000000" stroke-width="18.6464" d="M368.753+729.441L58.8847+550.539L58.8848+192.734L368.753+13.8313L678.621+192.734L678.621+550.539L368.753+729.441Z" fill="#0082fc" stroke-linecap="butt" fill-opacity="0.307489" opacity="1" stroke-linejoin="round"/>
+<g opacity="1">
+<g opacity="1">
+<path stroke="#000000" stroke-width="20" d="M292.873+289.256L442.872+289.256L442.872+539.254L292.873+539.254L292.873+289.256Z" fill="#fcd100" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+<path stroke="#000000" stroke-width="20" d="M292.873+289.256C292.873+247.835+326.452+214.257+367.873+214.257C409.294+214.257+442.872+247.835+442.872+289.256C442.872+330.677+409.294+364.256+367.873+364.256C326.452+364.256+292.873+330.677+292.873+289.256Z" fill="#fcd100" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+<path stroke="#000000" stroke-width="20" d="M292.873+539.254C292.873+497.833+326.452+464.255+367.873+464.255C409.294+464.255+442.872+497.833+442.872+539.254C442.872+580.675+409.294+614.254+367.873+614.254C326.452+614.254+292.873+580.675+292.873+539.254Z" fill="#fcd100" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+<path stroke="#0082fc" stroke-width="0.1" d="M302.873+289.073L432.872+289.073L432.872+539.072L302.873+539.072L302.873+289.073Z" fill="#fcd100" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+</g>
+<path stroke="#000000" stroke-width="0.1" d="M103.161+309.167L226.956+443.903L366.671+309.604L103.161+309.167Z" fill="#0082fc" stroke-linecap="round" opacity="1" stroke-linejoin="round"/>
+<path stroke="#000000" stroke-width="0.1" d="M383.411+307.076L508.887+440.112L650.5+307.507L383.411+307.076Z" fill="#0082fc" stroke-linecap="round" opacity="1" stroke-linejoin="round"/>
+<path stroke="#000000" stroke-width="20" d="M522.045+154.808L229.559+448.882L83.8397+300.104L653.666+302.936L511.759+444.785L223.101+156.114" fill="none" stroke-linecap="round" opacity="1" stroke-linejoin="round"/>
+<path stroke="#000000" stroke-width="61.8698" d="M295.857+418.738L438.9+418.738" fill="none" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+<path stroke="#000000" stroke-width="61.8698" d="M295.857+521.737L438.9+521.737" fill="none" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+<g opacity="1">
+<path stroke="#0082fc" stroke-width="0.1" d="M367.769+667.024L367.821+616.383L403.677+616.336C383.137+626.447+368.263+638.69+367.769+667.024Z" fill="#000000" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+<path stroke="#0082fc" stroke-width="0.1" d="M367.836+667.024L367.784+616.383L331.928+616.336C352.468+626.447+367.341+638.69+367.836+667.024Z" fill="#000000" stroke-linecap="butt" opacity="1" stroke-linejoin="round"/>
+</g>
+</g>
+</g>
+</svg>
--- a/apps/speaker/speaker.css
+++ b/apps/speaker/speaker.css
@@ -0,0 +1,66 @@
+body, h1, h2, h3, h4, h5, h6 {
+    font-family: sans-serif;
+}
+
+#controlsDiv {
+    margin: 6px;
+}
+
+#connectionText {
+    background-color: rgb(239, 89, 75);
+    border: none;
+    border-radius: 4px;
+    padding: 8px;
+    display: inline-block;
+    margin: 4px;
+}
+
+#startButton {
+    padding: 4px;
+    margin: 6px;
+}
+
+#fftCanvas {
+    border-radius: 16px;
+    margin: 6px;
+}
+
+#bandwidthCanvas {
+    border: grey;
+    border-style: solid;
+    border-radius: 8px;
+    margin: 6px;
+}
+
+#streamStateText {
+    background-color: rgb(93, 165, 93);
+    border: none;
+    border-radius: 8px;
+    padding: 10px 20px;
+    display: inline-block;
+    margin: 6px;
+}
+
+#propertiesTable {
+    border: grey;
+    border-style: solid;
+    border-radius: 4px;
+    padding: 4px;
+    margin: 6px;
+}
+
+th, td {
+    padding-left: 8px;
+    padding-right: 8px;
+}
+
+.properties td:nth-child(even) {
+    background-color: #D6EEEE;
+    font-family: monospace;
+}
+
+.properties td:nth-child(odd) {
+    font-weight: bold;
+}
+
+.properties tr td:nth-child(2) { width: 150px; }
--- a/apps/speaker/speaker.html
+++ b/apps/speaker/speaker.html
@@ -0,0 +1,33 @@
+<!DOCTYPE html>
+<html>
+<head>
+  <title>Bumble Speaker</title>
+  <script type="text/javascript" src="speaker.js"></script>
+  <link rel="stylesheet" href="speaker.css">
+</head>
+<body>
+  <h1><img src="logo.svg" width=100 height=100 style="vertical-align:middle" alt=""/>Bumble Virtual Speaker</h1>
+  <div id="connectionText"></div>
+  <div id="speaker">
+    <table><tr>
+      <td>
+        <table id="propertiesTable" class="properties">
+          <tr><td>Codec</td><td><span id="codecText"></span></td></tr>
+          <tr><td>Packets</td><td><span id="packetsReceivedText"></span></td></tr>
+          <tr><td>Bytes</td><td><span id="bytesReceivedText"></span></td></tr>
+        </table>
+      </td>
+      <td>
+        <canvas id="bandwidthCanvas" width="500", height="100">Bandwidth Graph</canvas>
+      </td>
+    </tr></table>
+    <span id="streamStateText">IDLE</span>
+    <div id="controlsDiv">
+        <button id="audioOnButton">Audio On</button>
+        <span id="audioSupportMessageText"></span>
+    </div>
+    <canvas id="fftCanvas" width="1024", height="300">Audio Frequencies Animation</canvas>
+    <audio id="audio"></audio>
+  </div>
+</body>
+</html>
--- a/apps/speaker/speaker.js
+++ b/apps/speaker/speaker.js
@@ -0,0 +1,297 @@
+(function () {
+    'use strict';
+
+const channelUrl = ((window.location.protocol === "https:") ? "wss://" : "ws://") + window.location.host + "/channel";
+let channelSocket;
+let connectionText;
+let codecText;
+let packetsReceivedText;
+let bytesReceivedText;
+let streamStateText;
+let controlsDiv;
+let audioOnButton;
+let mediaSource;
+let sourceBuffer;
+let audioElement;
+let audioContext;
+let audioAnalyzer;
+let audioFrequencyBinCount;
+let audioFrequencyData;
+let packetsReceived = 0;
+let bytesReceived = 0;
+let audioState = "stopped";
+let streamState = "IDLE";
+let audioSupportMessageText;
+let fftCanvas;
+let fftCanvasContext;
+let bandwidthCanvas;
+let bandwidthCanvasContext;
+let bandwidthBinCount;
+let bandwidthBins;
+
+const FFT_WIDTH = 800;
+const FFT_HEIGHT = 256;
+const BANDWIDTH_WIDTH = 500;
+const BANDWIDTH_HEIGHT = 100;
+
+function hexToBytes(hex) {
+    return Uint8Array.from(hex.match(/.{1,2}/g).map((byte) => parseInt(byte, 16)));
+}
+
+function init() {
+    initUI();
+    initMediaSource();
+    initAudioElement();
+    initAnalyzer();
+
+    connect();
+}
+
+function initUI() {
+    controlsDiv = document.getElementById("controlsDiv");
+    controlsDiv.style.visibility = "hidden";
+    connectionText = document.getElementById("connectionText");
+    audioOnButton = document.getElementById("audioOnButton");
+    codecText = document.getElementById("codecText");
+    packetsReceivedText = document.getElementById("packetsReceivedText");
+    bytesReceivedText = document.getElementById("bytesReceivedText");
+    streamStateText = document.getElementById("streamStateText");
+    audioSupportMessageText = document.getElementById("audioSupportMessageText");
+
+    audioOnButton.onclick = () => startAudio();
+
+    setConnectionText("");
+}
+
+function initMediaSource() {
+    mediaSource = new MediaSource();
+    mediaSource.onsourceopen = onMediaSourceOpen;
+    mediaSource.onsourceclose = onMediaSourceClose;
+    mediaSource.onsourceended = onMediaSourceEnd;
+}
+
+function initAudioElement() {
+    audioElement = document.getElementById("audio");
+    audioElement.src = URL.createObjectURL(mediaSource);
+    // audioElement.controls = true;
+}
+
+function initAnalyzer() {
+    fftCanvas = document.getElementById("fftCanvas");
+    fftCanvas.width = FFT_WIDTH
+    fftCanvas.height = FFT_HEIGHT
+    fftCanvasContext = fftCanvas.getContext('2d');
+    fftCanvasContext.fillStyle = "rgb(0, 0, 0)";
+    fftCanvasContext.fillRect(0, 0, FFT_WIDTH, FFT_HEIGHT);
+
+    bandwidthCanvas = document.getElementById("bandwidthCanvas");
+    bandwidthCanvas.width = BANDWIDTH_WIDTH
+    bandwidthCanvas.height = BANDWIDTH_HEIGHT
+    bandwidthCanvasContext = bandwidthCanvas.getContext('2d');
+    bandwidthCanvasContext.fillStyle = "rgb(255, 255, 255)";
+    bandwidthCanvasContext.fillRect(0, 0, BANDWIDTH_WIDTH, BANDWIDTH_HEIGHT);
+}
+
+function startAnalyzer() {
+    // FFT
+    audioContext = new AudioContext();
+    audioAnalyzer = audioContext.createAnalyser();
+    audioAnalyzer.fftSize = 128;
+    audioFrequencyBinCount = audioAnalyzer.frequencyBinCount;
+    audioFrequencyData = new Uint8Array(audioFrequencyBinCount);
+    const stream = audioElement.captureStream();
+    const source = audioContext.createMediaStreamSource(stream);
+    source.connect(audioAnalyzer);
+
+    // Bandwidth
+    bandwidthBinCount = BANDWIDTH_WIDTH / 2;
+    bandwidthBins = [];
+
+    requestAnimationFrame(onAnimationFrame);
+}
+
+function setConnectionText(message) {
+    connectionText.innerText = message;
+    if (message.length == 0) {
+        connectionText.style.display = "none";
+    } else {
+        connectionText.style.display = "inline-block";
+    }
+}
+
+function onAnimationFrame() {
+    // FFT
+    audioAnalyzer.getByteFrequencyData(audioFrequencyData);
+    fftCanvasContext.fillStyle = "rgb(0, 0, 0)";
+    fftCanvasContext.fillRect(0, 0, FFT_WIDTH, FFT_HEIGHT);
+    const barCount = audioFrequencyBinCount;
+    const barWidth = (FFT_WIDTH / audioFrequencyBinCount) - 1;
+    for (let bar = 0; bar < barCount; bar++) {
+        const barHeight = audioFrequencyData[bar];
+        fftCanvasContext.fillStyle = `rgb(${barHeight / 256 * 200 + 50}, 50, ${50 + 2 * bar})`;
+        fftCanvasContext.fillRect(bar * (barWidth + 1), FFT_HEIGHT - barHeight, barWidth, barHeight);
+    }
+
+    // Bandwidth
+    bandwidthCanvasContext.fillStyle = "rgb(255, 255, 255)";
+    bandwidthCanvasContext.fillRect(0, 0, BANDWIDTH_WIDTH, BANDWIDTH_HEIGHT);
+    bandwidthCanvasContext.fillStyle = `rgb(100, 100, 100)`;
+    for (let t = 0; t < bandwidthBins.length; t++) {
+        const lineHeight = (bandwidthBins[t] / 1000) * BANDWIDTH_HEIGHT;
+        bandwidthCanvasContext.fillRect(t * 2, BANDWIDTH_HEIGHT - lineHeight, 2, lineHeight);
+    }
+
+    // Display again at the next frame
+    requestAnimationFrame(onAnimationFrame);
+}
+
+function onMediaSourceOpen() {
+    console.log(this.readyState);
+    sourceBuffer = mediaSource.addSourceBuffer("audio/aac");
+}
+
+function onMediaSourceClose() {
+    console.log(this.readyState);
+}
+
+function onMediaSourceEnd() {
+    console.log(this.readyState);
+}
+
+async function startAudio() {
+    try {
+        console.log("starting audio...");
+        audioOnButton.disabled = true;
+        audioState = "starting";
+        await audioElement.play();
+        console.log("audio started");
+        audioState = "playing";
+        startAnalyzer();
+    } catch(error) {
+        console.error(`play failed: ${error}`);
+        audioState = "stopped";
+        audioOnButton.disabled = false;
+    }
+}
+
+function onAudioPacket(packet) {
+    if (audioState == "stopped") {
+        // Drop the packet, we're not ready to play.
+        return;
+    }
+
+    // Queue the audio packet.
+    sourceBuffer.appendBuffer(packet);
+
+    packetsReceived += 1;
+    packetsReceivedText.innerText = packetsReceived;
+    bytesReceived += packet.byteLength;
+    bytesReceivedText.innerText = bytesReceived;
+
+    bandwidthBins[bandwidthBins.length] = packet.byteLength;
+    if (bandwidthBins.length > bandwidthBinCount) {
+        bandwidthBins.shift();
+    }
+}
+
+function onChannelOpen() {
+    console.log('channel OPEN');
+    setConnectionText("");
+    controlsDiv.style.visibility = "visible";
+
+    // Handshake with the backend.
+    sendMessage({
+        type: "hello"
+    });
+}
+
+function onChannelClose() {
+    console.log('channel CLOSED');
+    setConnectionText("Connection to CLI app closed, restart it and reload this page.");
+    controlsDiv.style.visibility = "hidden";
+}
+
+function onChannelError(error) {
+    console.log(`channel ERROR: ${error}`);
+    setConnectionText(`Connection to CLI app error ({${error}}), restart it and reload this page.`);
+    controlsDiv.style.visibility = "hidden";
+}
+
+function onChannelMessage(message) {
+    if (typeof message.data === 'string' || message.data instanceof String) {
+        // JSON message.
+        const jsonMessage = JSON.parse(message.data);
+        console.log(`channel MESSAGE: ${message.data}`);
+
+        // Dispatch the message.
+        const handlerName = `on${jsonMessage.type.charAt(0).toUpperCase()}${jsonMessage.type.slice(1)}Message`
+        const handler = messageHandlers[handlerName];
+        if (handler !== undefined) {
+            const params = jsonMessage.params;
+            if (params === undefined) {
+                params = {};
+            }
+            handler(params);
+        } else {
+            console.warn(`unhandled message: ${jsonMessage.type}`)
+        }
+    } else {
+        // BINARY audio data.
+        onAudioPacket(message.data);
+    }
+}
+
+function onHelloMessage(params) {
+    codecText.innerText = params.codec;
+    if (params.codec != "aac") {
+        audioOnButton.disabled = true;
+        audioSupportMessageText.innerText = "Only AAC can be played, audio will be disabled";
+        audioSupportMessageText.style.display = "inline-block";
+    } else {
+        audioSupportMessageText.innerText = "";
+        audioSupportMessageText.style.display = "none";
+    }
+}
+
+function onStartMessage(params) {
+    streamState = "STARTED";
+    streamStateText.innerText = streamState;
+}
+
+function onStopMessage(params) {
+    streamState = "STOPPED";
+    streamStateText.innerText = streamState;
+}
+
+function onSuspendMessage(params) {
+    streamState = "SUSPENDED";
+    streamStateText.innerText = streamState;
+}
+
+function sendMessage(message) {
+    channelSocket.send(JSON.stringify(message));
+}
+
+function connect() {
+    console.log("connecting to CLI app");
+
+    channelSocket = new WebSocket(channelUrl);
+    channelSocket.binaryType = "arraybuffer";
+    channelSocket.onopen = onChannelOpen;
+    channelSocket.onclose = onChannelClose;
+    channelSocket.onerror = onChannelError;
+    channelSocket.onmessage = onChannelMessage;
+}
+
+const messageHandlers = {
+    onHelloMessage,
+    onStartMessage,
+    onStopMessage,
+    onSuspendMessage
+}
+
+window.onload = (event) => {
+    init();
+}
+
+}());
--- a/apps/speaker/speaker.py
+++ b/apps/speaker/speaker.py
@@ -15,19 +15,32 @@
 # -----------------------------------------------------------------------------
 # Imports
 # -----------------------------------------------------------------------------
+from __future__ import annotations
 import asyncio
+import asyncio.subprocess
+from importlib import resources
+import json
 import os
 import logging
+import pathlib
+from typing import Dict, List, Optional
+import weakref

 import click
-from bumble.core import BT_BR_EDR_TRANSPORT
+import aiohttp
+from aiohttp import web

+import bumble
+from bumble.colors import color
+from bumble.core import BT_BR_EDR_TRANSPORT
 from bumble.device import Device, DeviceConfiguration
+from bumble.sdp import ServiceAttribute
 from bumble.transport import open_transport
 from bumble.avdtp import (
    AVDTP_AUDIO_MEDIA_TYPE,
    Listener,
    MediaCodecCapabilities,
+    MediaPacket,
    Protocol,
 )
 from bumble.a2dp import (
@@ -42,22 +55,323 @@ from bumble.a2dp import (
    SBC_STEREO_CHANNEL_MODE,
    SBC_JOINT_STEREO_CHANNEL_MODE,
    SbcMediaCodecInformation,
-    AacMediaCodecInformation
+    AacMediaCodecInformation,
 )
 from bumble.utils import AsyncRunner
+from bumble.codecs import AacAudioRtpPacket
+
+
+# -----------------------------------------------------------------------------
+# Constants
+# -----------------------------------------------------------------------------
+DEFAULT_UI_PORT = 7654
+
+# -----------------------------------------------------------------------------
+class AudioExtractor:
+    @staticmethod
+    def create(codec: str):
+        if codec == 'aac':
+            return AacAudioExtractor()
+        if codec == 'sbc':
+            return SbcAudioExtractor()
+
+    def extract_audio(self, packet: MediaPacket) -> bytes:
+        raise NotImplementedError()
+
+
+# -----------------------------------------------------------------------------
+class AacAudioExtractor:
+    def extract_audio(self, packet: MediaPacket) -> bytes:
+        return AacAudioRtpPacket(packet.payload).to_adts()
+
+
+# -----------------------------------------------------------------------------
+class SbcAudioExtractor:
+    def extract_audio(self, packet: MediaPacket) -> bytes:
+        # header = packet.payload[0]
+        # fragmented = header >> 7
+        # start = (header >> 6) & 0x01
+        # last = (header >> 5) & 0x01
+        # number_of_frames = header & 0x0F
+
+        # TODO: support fragmented payloads
+        return packet.payload[1:]
+
+
+# -----------------------------------------------------------------------------
+class Output:
+    async def start(self):
+        pass
+
+    async def stop(self):
+        pass
+
+    async def suspend(self):
+        pass
+
+
+# -----------------------------------------------------------------------------
+class FileOutput(Output):
+    filename: str
+    codec: str
+    extractor: AudioExtractor
+
+    def __init__(self, filename, codec):
+        self.filename = filename
+        self.codec = codec
+        self.file = open(filename, 'wb')
+        self.extractor = AudioExtractor.create(codec)
+
+    def on_rtp_packet(self, packet: MediaPacket) -> None:
+        self.file.write(self.extractor.extract_audio(packet))
+
+
+# -----------------------------------------------------------------------------
+class QueuedOutput(Output):
+    MAX_QUEUE_SIZE = 32768
+
+    packets: asyncio.Queue
+    extractor: AudioExtractor
+    packet_pump_task: Optional[asyncio.Task]
+    started: bool
+
+    def __init__(self, extractor):
+        self.extractor = extractor
+        self.packets = asyncio.Queue()
+        self.packet_pump_task = None
+        self.started = False
+
+    async def start(self):
+        if self.started:
+            return
+
+        self.packet_pump_task = asyncio.create_task(self.pump_packets())
+
+    async def pump_packets(self):
+        while True:
+            packet = await self.packets.get()
+            await self.on_audio_packet(packet)
+
+    async def on_audio_packet(self, packet: bytes) -> None:
+        pass
+
+    def on_rtp_packet(self, packet: MediaPacket) -> None:
+        if self.packets.qsize() > self.MAX_QUEUE_SIZE:
+            print("queue full, dropping")
+            return
+
+        self.packets.put_nowait(self.extractor.extract_audio(packet))
+
+
+# -----------------------------------------------------------------------------
+class WebSocketOutput(QueuedOutput):
+    def __init__(self, codec, send_audio, send_message):
+        super().__init__(AudioExtractor.create(codec))
+        self.send_audio = send_audio
+        self.send_message = send_message
+
+    async def on_audio_packet(self, packet: bytes) -> None:
+        await self.send_audio(packet)
+
+    async def start(self):
+        await super().start()
+        await self.send_message('start')
+
+    async def stop(self):
+        await super().stop()
+        await self.send_message('stop')
+
+    async def suspend(self):
+        await super().suspend()
+        await self.send_message('suspend')
+
+
+# -----------------------------------------------------------------------------
+class FfplayOutput(QueuedOutput):
+    MAX_QUEUE_SIZE = 32768
+
+    subprocess: Optional[asyncio.subprocess.Process]
+    ffplay_task: Optional[asyncio.Task]
+
+    def __init__(self) -> None:
+        super().__init__(AacAudioExtractor())
+        self.subprocess = None
+        self.ffplay_task = None
+
+    async def start(self):
+        if self.started:
+            return
+
+        super().start()
+
+        self.subprocess = await asyncio.create_subprocess_shell(
+            'ffplay -acodec aac pipe:0',
+            stdin=asyncio.subprocess.PIPE,
+            stdout=asyncio.subprocess.PIPE,
+            stderr=asyncio.subprocess.PIPE,
+        )
+
+        self.ffplay_task = asyncio.create_task(self.monitor_ffplay())
+
+    async def stop(self):
+        # TODO
+        pass
+
+    async def suspend(self):
+        # TODO
+        pass
+
+    async def monitor_ffplay(self):
+        async def read_stream(name, stream):
+            while True:
+                data = await stream.read()
+                print(f'{name}:', data)
+
+        await asyncio.wait(
+            [
+                asyncio.create_task(
+                    read_stream('[ffplay stdout]', self.subprocess.stdout)
+                ),
+                asyncio.create_task(
+                    read_stream('[ffplay stderr]', self.subprocess.stderr)
+                ),
+                asyncio.create_task(self.subprocess.wait()),
+            ]
+        )
+        print("FFPLAY done")
+
+    async def on_audio_packet(self, packet):
+        try:
+            self.subprocess.stdin.write(packet)
+        except Exception:
+            print('!!!! exception while sending audio to ffplay pipe')
+
+
+# -----------------------------------------------------------------------------
+class UiServer:
+    speaker: Speaker
+    port: int
+
+    def __init__(self, speaker: Speaker, port: int) -> None:
+        self.speaker = weakref.ref(speaker)
+        self.port = port
+        self.channel_socket = None
+
+    async def start_http(self) -> None:
+        """Start the UI HTTP server."""
+
+        app = web.Application()
+        app.add_routes(
+            [
+                web.get('/', self.get_static),
+                web.get('/speaker.html', self.get_static),
+                web.get('/speaker.js', self.get_static),
+                web.get('/speaker.css', self.get_static),
+                web.get('/logo.svg', self.get_static),
+                web.get('/channel', self.get_channel),
+            ]
+        )
+
+        runner = web.AppRunner(app)
+        await runner.setup()
+        site = web.TCPSite(runner, 'localhost', self.port)
+        print('UI HTTP server at ' + color(f'http://127.0.0.1:{self.port}', 'green'))
+        await site.start()
+
+    async def get_static(self, request):
+        path = request.path
+        if path == '/':
+            path = '/speaker.html'
+        if path.endswith('.html'):
+            content_type = 'text/html'
+        elif path.endswith('.js'):
+            content_type = 'text/javascript'
+        elif path.endswith('.css'):
+            content_type = 'text/css'
+        elif path.endswith('.svg'):
+            content_type = 'image/svg+xml'
+        else:
+            content_type = 'text/plain'
+        text = (
+            resources.files("bumble.apps.speaker")
+            .joinpath(pathlib.Path(path).relative_to('/'))
+            .read_text(encoding="utf-8")
+        )
+        return aiohttp.web.Response(text=text, content_type=content_type)
+
+    async def get_channel(self, request):
+        ws = web.WebSocketResponse()
+        await ws.prepare(request)
+
+        # Process messages until the socket is closed.
+        self.channel_socket = ws
+        async for message in ws:
+            if message.type == aiohttp.WSMsgType.TEXT:
+                print(f'<<< received message: {message.data}')
+                await self.on_message(message.data)
+            elif message.type == aiohttp.WSMsgType.ERROR:
+                print(f'channel connection closed with exception {ws.exception()}')
+
+        self.channel_socket = None
+        print('--- channel connection closed')
+
+        return ws
+
+    async def on_message(self, message_str: str):
+        # Parse the message as JSON
+        message = json.loads(message_str)
+
+        # Dispatch the message
+        message_type = message['type']
+        message_params = message.get('params', {})
+        handler = getattr(self, f'on_{message_type}_message')
+        if handler:
+            await handler(**message_params)
+
+    async def on_hello_message(self):
+        print('HELLO')
+        await self.send_message(
+            'hello', bumble_version=bumble.__version__, codec=self.speaker().codec
+        )
+
+    async def send_message(self, message_type: str, **kwargs) -> None:
+        if self.channel_socket is None:
+            return
+
+        message = {'type': message_type, 'params': kwargs}
+        await self.channel_socket.send_json(message)
+
+    async def send_audio(self, data: bytes) -> None:
+        if self.channel_socket is None:
+            return
+
+        await self.channel_socket.send_bytes(data)


 # -----------------------------------------------------------------------------
 class Speaker:
-    def __init__(self, transport, discover):
+    def __init__(self, transport, codec, discover, outputs, ui_port):
        self.transport = transport
+        self.codec = codec
        self.discover = discover
+        self.ui_port = ui_port
        self.device = None
        self.listener = None
-        self.output_filename = 'speaker_output.sbc'
-        self.output = None
+        self.packets_received = 0
+        self.bytes_received = 0
+        self.outputs = []
+        for output in outputs:
+            if output == '@ffplay':
+                self.outputs.append(FfplayOutput())
+                continue

-    def sdp_records(self):
+            # Default to FileOutput
+            self.outputs.append(FileOutput(output, codec))
+
+        # Create an HTTP server for the UI
+        self.ui_server = UiServer(speaker=self, port=ui_port)
+
+    def sdp_records(self) -> Dict[int, List[ServiceAttribute]]:
        service_record_handle = 0x00010001
        return {
            service_record_handle: make_audio_sink_service_sdp_records(
@@ -65,23 +379,29 @@ class Speaker:
            )
        }

-    def codec_capabilities(self):
-        return self.aac_codec_capabilities()
+    def codec_capabilities(self) -> MediaCodecCapabilities:
+        if self.codec == 'aac':
+            return self.aac_codec_capabilities()

-    def aac_codec_capabilities(self):
+        if self.codec == 'sbc':
+            return self.sbc_codec_capabilities()
+
+        raise RuntimeError('unsupported codec')
+
+    def aac_codec_capabilities(self) -> MediaCodecCapabilities:
        return MediaCodecCapabilities(
            media_type=AVDTP_AUDIO_MEDIA_TYPE,
            media_codec_type=A2DP_MPEG_2_4_AAC_CODEC_TYPE,
            media_codec_information=AacMediaCodecInformation.from_lists(
                object_types=[MPEG_2_AAC_LC_OBJECT_TYPE],
                sampling_frequencies=[48000, 44100],
-                channels=[1,2],
+                channels=[1, 2],
                vbr=1,
-                bitrate=256000
-            )
+                bitrate=256000,
+            ),
        )

-    def sbc_codec_capabilities(self):
+    def sbc_codec_capabilities(self) -> MediaCodecCapabilities:
        return MediaCodecCapabilities(
            media_type=AVDTP_AUDIO_MEDIA_TYPE,
            media_codec_type=A2DP_SBC_CODEC_TYPE,
@@ -137,12 +457,18 @@ class Speaker:

    def on_sink_start(self):
        print("Sink Start")
+        for output in self.outputs:
+            AsyncRunner.spawn(output.start())

    def on_sink_stop(self):
        print("Sink Stop")
+        for output in self.outputs:
+            AsyncRunner.spawn(output.stop())

    def on_sink_suspend(self):
        print("Sink Suspend")
+        for output in self.outputs:
+            AsyncRunner.spawn(output.suspend())

    def on_sink_configuration(self, config):
        print("Sink Configuration:")
@@ -155,21 +481,15 @@ class Speaker:
        print("RTP Channel Closed")

    def on_rtp_packet(self, packet):
-        # header = packet.payload[0]
-        # fragmented = header >> 7
-        # # start = (header >> 6) & 0x01
-        # # last = (header >> 5) & 0x01
-        # number_of_frames = header & 0x0F
+        self.packets_received += 1
+        self.bytes_received += len(packet.payload)
+        print(
+            f'[{self.bytes_received} bytes in {self.packets_received} packets] {packet}',
+            end='\r',
+        )

-        # payload = packet.payload[1:]
-        # payload_size = len(payload)
-        # if fragmented:
-        #     print(f'RTP: fragment {payload_size} bytes in {number_of_frames} frames')
-        # else:
-        #     print(f'RTP: {payload_size} bytes in {number_of_frames} frames')
-        print(packet.payload.hex())
-
-        self.output.write(packet.payload)
+        for output in self.outputs:
+            output.on_rtp_packet(packet)

    async def advertise(self):
        await self.device.set_discoverable(True)
@@ -178,9 +498,7 @@ class Speaker:
    async def connect(self, address):
        # Connect to the source
        print(f'=== Connecting to {address}...')
-        connection = await self.device.connect(
-            address, transport=BT_BR_EDR_TRANSPORT
-        )
+        connection = await self.device.connect(address, transport=BT_BR_EDR_TRANSPORT)
        print(f'=== Connected to {connection.peer_address}')
        self.on_bluetooth_connection(connection)

@@ -205,71 +523,117 @@ class Speaker:
            print('@@@', endpoint)

    async def run(self, connect_address):
+        print(f'Speaker ready to play, codec={color(self.codec, "cyan")}')
+        await self.ui_server.start_http()
+        self.outputs.append(
+            WebSocketOutput(
+                self.codec, self.ui_server.send_audio, self.ui_server.send_message
+            )
+        )
+
        async with await open_transport(self.transport) as (hci_source, hci_sink):
-            with open(self.output_filename, 'wb') as sbc_file:
-                self.output = sbc_file
+            # Create a device
+            device_config = DeviceConfiguration()
+            device_config.name = "Bumble Speaker"
+            device_config.class_of_device = 0x240404
+            device_config.keystore = "JsonKeyStore"
+            device_config.classic_enabled = True
+            device_config.le_enabled = False
+            self.device = Device.from_config_with_hci(
+                device_config, hci_source, hci_sink
+            )

-                # Create a device
-                device_config = DeviceConfiguration()
-                device_config.name = "Bumble Speaker"
-                device_config.class_of_device = 2360324
-                device_config.keystore = "JsonKeyStore"
-                device_config.classic_enabled = True
-                device_config.le_enabled = False
-                self.device = Device.from_config_with_hci(
-                    device_config, hci_source, hci_sink
-                )
+            # Setup the SDP to expose the sink service
+            self.device.sdp_service_records = self.sdp_records()

-                # Setup the SDP to expose the sink service
-                self.device.sdp_service_records = self.sdp_records()
+            # Start the controller
+            await self.device.power_on()

-                # Start the controller
-                await self.device.power_on()
+            # Listen for Bluetooth connections
+            self.device.on('connection', self.on_bluetooth_connection)

-                # Listen for Bluetooth connections
-                self.device.on('connection', self.on_bluetooth_connection);
+            # Create a listener to wait for AVDTP connections
+            self.listener = Listener(Listener.create_registrar(self.device))
+            self.listener.on('connection', self.on_avdtp_connection)

-                # Create a listener to wait for AVDTP connections
-                self.listener = Listener(Listener.create_registrar(self.device))
-                self.listener.on('connection', self.on_avdtp_connection)
+            if connect_address:
+                # Connect to the source
+                await self.connect(connect_address)
+            else:
+                # Start being discoverable and connectable
+                print("Waiting for connection...")
+                await self.advertise()

-                if connect_address:
-                    # Connect to the source
-                    await self.connect(connect_address)
-                else:
-                    # Start being discoverable and connectable
-                    await self.advertise()
+            await hci_source.wait_for_termination()

-                await hci_source.wait_for_termination()
+        for output in self.outputs:
+            await output.stop()


 # -----------------------------------------------------------------------------
@click.group()
@click.option('--device-config', metavar='FILENAME', help='Device configuration file')
@click.pass_context
-def speaker(ctx, device_config):
+def speaker_cli(ctx, device_config):
    ctx.ensure_object(dict)
    ctx.obj['device_config'] = device_config


-@speaker.command()
+@speaker_cli.command()
@click.argument('transport')
+@click.option(
+    '--codec', type=click.Choice(['sbc', 'aac']), default='aac', show_default=True
+)
@click.option(
    '--connect',
    'connect_address',
    metavar='ADDRESS_OR_NAME',
    help='Address or name to connect to',
 )
-@click.option('--discover', is_flag=True)
+@click.option(
+    '--discover', is_flag=True, help='Discover remote endpoints once connected'
+)
+@click.option(
+    '--output',
+    multiple=True,
+    metavar='NAME',
+    help=(
+        'Send audio to this named output '
+        '(may be used more than once for multiple outputs)'
+    ),
+)
+@click.option(
+    '--ui-port',
+    'ui_port',
+    metavar='HTTP_PORT',
+    default=DEFAULT_UI_PORT,
+    show_default=True,
+    help='HTTP port for the UI server',
+)
@click.pass_context
-def play(ctx, transport, connect_address, discover):
-    asyncio.run(Speaker(transport, discover).run(connect_address))
+def play(ctx, transport, codec, connect_address, discover, output, ui_port):
+    """Run the speaker in playback mode."""
+
+    # ffplay only works with AAC for now
+    if codec != 'aac' and '@ffplay' in output:
+        print(
+            color(
+                f'{codec} not supported with @ffplay output, '
+                '@ffplay output will be skipped',
+                'yellow',
+            )
+        )
+        output = list(filter(lambda x: x != '@ffplay', output))
+
+    asyncio.run(
+        Speaker(transport, codec, discover, output, ui_port).run(connect_address)
+    )


 # -----------------------------------------------------------------------------
 def main():
    logging.basicConfig(level=os.environ.get('BUMBLE_LOGLEVEL', 'INFO').upper())
-    speaker()
+    speaker_cli()


 # -----------------------------------------------------------------------------
--- a/bumble/codecs.py
+++ b/bumble/codecs.py
@@ -0,0 +1,381 @@
+# Copyright 2023 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# -----------------------------------------------------------------------------
+# Imports
+# -----------------------------------------------------------------------------
+from __future__ import annotations
+from dataclasses import dataclass
+
+
+# -----------------------------------------------------------------------------
+class BitReader:
+    """Simple but not optimized bit stream reader."""
+
+    data: bytes
+    bytes_position: int
+    bit_position: int
+    cache: int
+    bits_cached: int
+
+    def __init__(self, data: bytes):
+        self.data = data
+        self.byte_position = 0
+        self.bit_position = 0
+        self.cache = 0
+        self.bits_cached = 0
+
+    def read(self, bits: int) -> int:
+        """ "Read up to 32 bits."""
+
+        if bits > 32:
+            raise ValueError('maximum read size is 32')
+
+        if self.bits_cached >= bits:
+            # We have enough bits.
+            self.bits_cached -= bits
+            self.bit_position += bits
+            return (self.cache >> self.bits_cached) & ((1 << bits) - 1)
+
+        # Read more cache, up to 32 bits
+        feed_bytes = self.data[self.byte_position : self.byte_position + 4]
+        feed_size = len(feed_bytes)
+        feed_int = int.from_bytes(feed_bytes, byteorder='big')
+        if 8 * feed_size + self.bits_cached < bits:
+            raise ValueError('trying to read past the data')
+        self.byte_position += feed_size
+
+        # Combine the new cache and the old cache
+        cache = self.cache & ((1 << self.bits_cached) - 1)
+        new_bits = bits - self.bits_cached
+        self.bits_cached = 8 * feed_size - new_bits
+        result = (feed_int >> self.bits_cached) | (cache << new_bits)
+        self.cache = feed_int
+
+        self.bit_position += bits
+        return result
+
+    def read_bytes(self, count: int):
+        if self.bit_position + 8 * count > 8 * len(self.data):
+            raise ValueError('not enough data')
+
+        if self.bit_position % 8:
+            # Not byte aligned
+            result = bytearray(count)
+            for i in range(count):
+                result[i] = self.read(8)
+            return bytes(result)
+
+        # Byte aligned
+        self.byte_position = self.bit_position // 8
+        self.bits_cached = 0
+        self.cache = 0
+        offset = self.bit_position // 8
+        self.bit_position += 8 * count
+        return self.data[offset : offset + count]
+
+    def bits_left(self) -> int:
+        return (8 * len(self.data)) - self.bit_position
+
+    def skip(self, bits: int) -> None:
+        # Slow, but simple...
+        while bits:
+            if bits > 32:
+                self.read(32)
+                bits -= 32
+            else:
+                self.read(bits)
+                break
+
+
+# -----------------------------------------------------------------------------
+class AacAudioRtpPacket:
+    """AAC payload encapsulated in an RTP packet payload"""
+
+    @staticmethod
+    def latm_value(reader: BitReader) -> int:
+        bytes_for_value = reader.read(2)
+        value = 0
+        for _ in range(bytes_for_value + 1):
+            value = value * 256 + reader.read(8)
+        return value
+
+    @staticmethod
+    def program_config_element(reader: BitReader):
+        raise ValueError('program_config_element not supported')
+
+    @dataclass
+    class GASpecificConfig:
+        def __init__(
+            self, reader: BitReader, channel_configuration: int, audio_object_type: int
+        ) -> None:
+            # GASpecificConfig - ISO/EIC 14496-3 Table 4.1
+            frame_length_flag = reader.read(1)
+            depends_on_core_coder = reader.read(1)
+            if depends_on_core_coder:
+                self.core_coder_delay = reader.read(14)
+            extension_flag = reader.read(1)
+            if not channel_configuration:
+                AacAudioRtpPacket.program_config_element(reader)
+            if audio_object_type in (6, 20):
+                self.layer_nr = reader.read(3)
+            if extension_flag:
+                if audio_object_type == 22:
+                    num_of_sub_frame = reader.read(5)
+                layer_length = reader.read(11)
+                if audio_object_type in (17, 19, 20, 23):
+                    aac_section_data_resilience_flags = reader.read(1)
+                    aac_scale_factor_data_resilience_flags = reader.read(1)
+                    aac_spectral_data_resilience_flags = reader.read(1)
+                extension_flag_3 = reader.read(1)
+                if extension_flag_3 == 1:
+                    raise ValueError('extensionFlag3 == 1 not supported')
+
+    @staticmethod
+    def audio_object_type(reader: BitReader):
+        # GetAudioObjectType - ISO/EIC 14496-3 Table 1.16
+        audio_object_type = reader.read(5)
+        if audio_object_type == 31:
+            audio_object_type = 32 + reader.read(6)
+
+        return audio_object_type
+
+    @dataclass
+    class AudioSpecificConfig:
+        audio_object_type: int
+        sampling_frequency_index: int
+        sampling_frequency: int
+        channel_configuration: int
+        sbr_present_flag: int
+        ps_present_flag: int
+        extension_audio_object_type: int
+        extension_sampling_frequency_index: int
+        extension_sampling_frequency: int
+        extension_channel_configuration: int
+
+        SAMPLING_FREQUENCIES = [
+            96000,
+            88200,
+            64000,
+            48000,
+            44100,
+            32000,
+            24000,
+            22050,
+            16000,
+            12000,
+            11025,
+            8000,
+            7350,
+        ]
+
+        def __init__(self, reader: BitReader) -> None:
+            # AudioSpecificConfig - ISO/EIC 14496-3 Table 1.15
+            self.audio_object_type = AacAudioRtpPacket.audio_object_type(reader)
+            self.sampling_frequency_index = reader.read(4)
+            if self.sampling_frequency_index == 0xF:
+                self.sampling_frequency = reader.read(24)
+            else:
+                self.sampling_frequency = self.SAMPLING_FREQUENCIES[
+                    self.sampling_frequency_index
+                ]
+            self.channel_configuration = reader.read(4)
+            self.sbr_present_flag = -1
+            self.ps_present_flag = -1
+            if self.audio_object_type in (5, 29):
+                self.extension_audio_object_type = 5
+                self.sbc_present_flag = 1
+                if self.audio_object_type == 29:
+                    self.ps_present_flag = 1
+                self.extension_sampling_frequency_index = reader.read(4)
+                if self.extension_sampling_frequency_index == 0xF:
+                    self.extension_sampling_frequency = reader.read(24)
+                else:
+                    self.extension_sampling_frequency = self.SAMPLING_FREQUENCIES[
+                        self.extension_sampling_frequency_index
+                    ]
+                self.audio_object_type = AacAudioRtpPacket.audio_object_type(reader)
+                if self.audio_object_type == 22:
+                    self.extension_channel_configuration = reader.read(4)
+            else:
+                self.extension_audio_object_type = 0
+
+            if self.audio_object_type in (1, 2, 3, 4, 6, 7, 17, 19, 20, 21, 22, 23):
+                ga_specific_config = AacAudioRtpPacket.GASpecificConfig(
+                    reader, self.channel_configuration, self.audio_object_type
+                )
+            else:
+                raise ValueError(
+                    f'audioObjectType {self.audio_object_type} not supported'
+                )
+
+            # if self.extension_audio_object_type != 5 and bits_to_decode >= 16:
+            #     sync_extension_type = reader.read(11)
+            #     if sync_extension_type == 0x2B7:
+            #         self.extension_audio_object_type = AacAudioRtpPacket.audio_object_type(reader)
+            #         if self.extension_audio_object_type == 5:
+            #             self.sbr_present_flag = reader.read(1)
+            #             if self.sbr_present_flag:
+            #                 self.extension_sampling_frequency_index = reader.read(4)
+            #                 if self.extension_sampling_frequency_index == 0xF:
+            #                     self.extension_sampling_frequency = reader.read(24)
+            #                 else:
+            #                     self.extension_sampling_frequency = self.SAMPLING_FREQUENCIES[self.extension_sampling_frequency_index]
+            #                 if bits_to_decode >= 12:
+            #                     sync_extension_type = reader.read(11)
+            #                     if sync_extension_type == 0x548:
+            #                         self.ps_present_flag = reader.read(1)
+            #         elif self.extension_audio_object_type == 22:
+            #             self.sbr_present_flag = reader.read(1)
+            #             if self.sbr_present_flag:
+            #                 self.extension_sampling_frequency_index = reader.read(4)
+            #                 if self.extension_sampling_frequency_index == 0xF:
+            #                     self.extension_sampling_frequency = reader.read(24)
+            #                 else:
+            #                     self.extension_sampling_frequency = self.SAMPLING_FREQUENCIES[self.extension_sampling_frequency_index]
+            #             self.extension_channel_configuration = reader.read(4)
+
+    @dataclass
+    class StreamMuxConfig:
+        other_data_present: int
+        other_data_len_bits: int
+        audio_specific_config: AacAudioRtpPacket.AudioSpecificConfig
+
+        def __init__(self, reader: BitReader) -> None:
+            # StreamMuxConfig - ISO/EIC 14496-3 Table 1.42
+            audio_mux_version = reader.read(1)
+            if audio_mux_version == 1:
+                audio_mux_version_a = reader.read(1)
+            else:
+                audio_mux_version_a = 0
+            if audio_mux_version_a != 0:
+                raise ValueError('audioMuxVersionA != 0 not supported')
+            if audio_mux_version == 1:
+                tara_buffer_fullness = AacAudioRtpPacket.latm_value(reader)
+            stream_cnt = 0
+            all_streams_same_time_framing = reader.read(1)
+            num_sub_frames = reader.read(6)
+            num_program = reader.read(4)
+            if num_program != 0:
+                raise ValueError('num_program != 0 not supported')
+            num_layer = reader.read(3)
+            if num_layer != 0:
+                raise ValueError('num_layer != 0 not supported')
+            if audio_mux_version == 0:
+                self.audio_specific_config = AacAudioRtpPacket.AudioSpecificConfig(
+                    reader
+                )
+            else:
+                asc_len = AacAudioRtpPacket.latm_value(reader)
+                marker = reader.bit_position
+                self.audio_specific_config = AacAudioRtpPacket.AudioSpecificConfig(
+                    reader
+                )
+                audio_specific_config_len = reader.bit_position - marker
+                if asc_len < audio_specific_config_len:
+                    raise ValueError('audio_specific_config_len > asc_len')
+                asc_len -= audio_specific_config_len
+                reader.skip(asc_len)
+            frame_length_type = reader.read(3)
+            if frame_length_type == 0:
+                latm_buffer_fullness = reader.read(8)
+            elif frame_length_type == 1:
+                frame_length = reader.read(9)
+            else:
+                raise ValueError(f'frame_length_type {frame_length_type} not supported')
+
+            self.other_data_present = reader.read(1)
+            if self.other_data_present:
+                if audio_mux_version == 1:
+                    self.other_data_len_bits = AacAudioRtpPacket.latm_value(reader)
+                else:
+                    self.other_data_len_bits = 0
+                    while True:
+                        self.other_data_len_bits *= 256
+                        other_data_len_esc = reader.read(1)
+                        self.other_data_len_bits += reader.read(8)
+                        if other_data_len_esc == 0:
+                            break
+            crc_check_present = reader.read(1)
+            if crc_check_present:
+                crc_checksum = reader.read(8)
+
+    @dataclass
+    class AudioMuxElement:
+        payload: bytes
+        stream_mux_config: AacAudioRtpPacket.StreamMuxConfig
+
+        def __init__(self, reader: BitReader, mux_config_present: int):
+            if mux_config_present == 0:
+                raise ValueError('muxConfigPresent == 0 not supported')
+
+            # AudioMuxElement - ISO/EIC 14496-3 Table 1.41
+            use_same_stream_mux = reader.read(1)
+            if use_same_stream_mux:
+                raise ValueError('useSameStreamMux == 1 not supported')
+            self.stream_mux_config = AacAudioRtpPacket.StreamMuxConfig(reader)
+
+            # We only support:
+            # allStreamsSameTimeFraming == 1
+            # audioMuxVersionA == 0,
+            # numProgram == 0
+            # numSubFrames == 0
+            # numLayer == 0
+
+            mux_slot_length_bytes = 0
+            while True:
+                tmp = reader.read(8)
+                mux_slot_length_bytes += tmp
+                if tmp != 255:
+                    break
+
+            self.payload = reader.read_bytes(mux_slot_length_bytes)
+
+            if self.stream_mux_config.other_data_present:
+                reader.skip(self.stream_mux_config.other_data_len_bits)
+
+            # ByteAlign
+            while reader.bit_position % 8:
+                reader.read(1)
+
+    def __init__(self, data: bytes) -> None:
+        # Parse the bit stream
+        reader = BitReader(data)
+        self.audio_mux_element = self.AudioMuxElement(reader, mux_config_present=1)
+
+    def to_adts(self):
+        # pylint: disable=line-too-long
+        sampling_frequency_index = (
+            self.audio_mux_element.stream_mux_config.audio_specific_config.sampling_frequency_index
+        )
+        channel_configuration = (
+            self.audio_mux_element.stream_mux_config.audio_specific_config.channel_configuration
+        )
+        frame_size = len(self.audio_mux_element.payload)
+        return (
+            bytes(
+                [
+                    0xFF,
+                    0xF1,  # 0xF9 (MPEG2)
+                    0x40
+                    | (sampling_frequency_index << 2)
+                    | (channel_configuration >> 2),
+                    ((channel_configuration & 0x3) << 6) | ((frame_size + 7) >> 11),
+                    ((frame_size + 7) >> 3) & 0xFF,
+                    (((frame_size + 7) << 5) & 0xFF) | 0x1F,
+                    0xFC,
+                ]
+            )
+            + self.audio_mux_element.payload
+        )
--- a/setup.cfg
+++ b/setup.cfg
@@ -30,21 +30,22 @@ package_dir =
    bumble.apps = apps
 include-package-data = True
 install_requires =
+    aiohttp >= 22.1.0; platform_system!='Emscripten'
    appdirs >= 1.4
    click >= 7.1.2; platform_system!='Emscripten'
    cryptography == 35; platform_system!='Emscripten'
    grpcio >= 1.46; platform_system!='Emscripten'
+    humanize >= 4.6.0
    libusb1 >= 2.0.1; platform_system!='Emscripten'
    libusb-package == 1.0.26.1; platform_system!='Emscripten'
    prompt_toolkit >= 3.0.16; platform_system!='Emscripten'
+    prettytable >= 3.6.0
    protobuf >= 3.12.4
    pyee >= 8.2.2
    pyserial-asyncio >= 0.5; platform_system!='Emscripten'
    pyserial >= 3.5; platform_system!='Emscripten'
    pyusb >= 1.2; platform_system!='Emscripten'
    websockets >= 8.1; platform_system!='Emscripten'
-    prettytable >= 3.6.0
-    humanize >= 4.6.0

 [options.entry_points]
 console_scripts =
--- a/speaker.html
+++ b/speaker.html
@@ -0,0 +1,28 @@
+<!DOCTYPE html>
+<html>
+  <head>
+    <title>Audio WAV Player</title>
+  </head>
+  <body>
+    <h1>Audio WAV Player</h1>
+    <audio id="audioPlayer" controls>
+      <source src="" type="audio/wav">
+    </audio>
+
+    <script>
+      const audioPlayer = document.getElementById('audioPlayer');
+      const ws = new WebSocket('ws://localhost:8080');
+
+      let mediaSource = new MediaSource();
+      audioPlayer.src = URL.createObjectURL(mediaSource);
+
+      mediaSource.addEventListener('sourceopen', function(event) {
+        const sourceBuffer = mediaSource.addSourceBuffer('audio/wav');
+
+        ws.onmessage = function(event) {
+          sourceBuffer.appendBuffer(event.data);
+        };
+      });
+    </script>
+  </body>
+</html>
--- a/tests/codecs_test.py
+++ b/tests/codecs_test.py
@@ -0,0 +1,64 @@
+# Copyright 2021-2023 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#      https://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# -----------------------------------------------------------------------------
+# Imports
+# -----------------------------------------------------------------------------
+import pytest
+from bumble.codecs import AacAudioRtpPacket, BitReader
+
+
+# -----------------------------------------------------------------------------
+def test_reader():
+    reader = BitReader(b'')
+    with pytest.raises(ValueError):
+        reader.read(1)
+
+    reader = BitReader(b'hello')
+    with pytest.raises(ValueError):
+        reader.read(40)
+
+    reader = BitReader(bytes([0xFF]))
+    assert reader.read(1) == 1
+    with pytest.raises(ValueError):
+        reader.read(10)
+
+    reader = BitReader(bytes([0x78]))
+    value = 0
+    for _ in range(8):
+        value = (value << 1) | reader.read(1)
+    assert value == 0x78
+
+
+    data = bytes([x & 0xFF for x in range(66 * 100)])
+    reader = BitReader(data)
+    value = 0
+    for _ in range(100):
+        for bits in range(1, 33):
+            value = value << bits | reader.read(bits)
+    assert value == int.from_bytes(data, byteorder='big')
+
+
+def test_aac_rtp():
+    # pylint: disable=line-too-long
+    packet_data = bytes.fromhex('47fc0000b090800300202066000198000de120000000000000000000000000000000000000000000001c')
+    packet = AacAudioRtpPacket(packet_data)
+    adts = packet.to_adts()
+    assert adts == bytes.fromhex('fff1508004fffc2066000198000de120000000000000000000000000000000000000000000001c')
+
+
+# -----------------------------------------------------------------------------
+if __name__ == '__main__':
+    test_reader()
+    test_aac_rtp()