Files
auracast-translator/multilang_translator/main.py
2024-12-19 14:01:32 +01:00

109 lines
3.4 KiB
Python

# -*- coding: utf-8 -*-
"""
list prompt example
"""
from __future__ import print_function, unicode_literals
from pprint import pprint
from PyInquirer import prompt, Separator
from examples import custom_style_2
import os
from copy import copy
import time
import logging as log
from .translator import llm_translator
from .text_to_speech import text_to_speech, resample
from .backend_controller.broadcaster_config import broadcaster_config
from .backend_controller.broadcaster_play_once import broadcaster_play_file
from .backend_controller.broadcaster_copy_files import copy_to_broadcaster
from .encode import encode_lc3
ANNOUNCEMENT_DIR = os.path.join(os.path.dirname(__file__), 'announcements')
SAMPLING_RATE = int(16e3)
FRAME_DUR_MS = 10
BPS = int(32e3) # TODO test 16khz 16kbps
CONFIG = {
"de": {
"file": f"{ANNOUNCEMENT_DIR}/announcement_{SAMPLING_RATE//1000}_{FRAME_DUR_MS}_{BPS//1000}_de",
"tts": 'de_DE-kerstin-low',
},
"en": {
"file": f"{ANNOUNCEMENT_DIR}/announcement_{SAMPLING_RATE//1000}_{FRAME_DUR_MS}_{BPS//1000}_en",
"tts": 'en_US-lessac-medium'
},
"fr": {
"file": f"{ANNOUNCEMENT_DIR}/announcement_{SAMPLING_RATE//1000}_{FRAME_DUR_MS}_{BPS//1000}_fr",
"tts": 'fr_FR-siwis-medium'
},
"es": {
"file": f"{ANNOUNCEMENT_DIR}/announcement_{SAMPLING_RATE//1000}_{FRAME_DUR_MS}_{BPS//1000}_es",
"tts": 'es_ES-sharvard-medium'
},
"it": {
"file": f"{ANNOUNCEMENT_DIR}/announcement_{SAMPLING_RATE//1000}_{FRAME_DUR_MS}_{BPS//1000}_it",
"tts": 'it_IT-paola-medium'
}
}
os.makedirs(ANNOUNCEMENT_DIR, exist_ok=True)
def synthesize_resample_encode(text, tts_model, output_file):
audio_dur = text_to_speech.synthesize(text, tts_model, output_file)
resample.resample(output_file, output_file, target_rate=SAMPLING_RATE)
encode_lc3.encode_lc3(output_file, bps=BPS, frame_dur_ms=FRAME_DUR_MS)
return audio_dur
def translate_from_german_and_encode(text_de):
config = copy(CONFIG)
base_lang = "de"
file = config[base_lang]["file"]
audio_dur_s [base_lang] = synthesize_resample_encode(text_de, config['de']["tts"], f'{file}.wav')
del config[base_lang]
audio_dur_s = {}
for key, val in config.items():
text = llm_translator.translate_de_to_x(key, text_de)
file = val['file']
audio_dur_s[key] = synthesize_resample_encode(text, val['tts'], f'{file}.wav')
return audio_dur_s
def announcement_from_german_text(text_de):
audio_durs = translate_from_german_and_encode(text_de)
# Transfer the files to broadcaster memory
start = time.time()
for val in CONFIG.values():
copy_to_broadcaster(f'{val["file"]}.lc3')
log.info("Transfering files to broadcaster took %s s", round(time.time() - start, 3))
# Instruct the broadcaster to stream the files
for i, d in enumerate(list(CONFIG.items())):
key, val = d
broadcaster_play_file(i, f'{os.path.basename(val["file"])}.lc3')
time.sleep(audio_durs[key] + 1)
log.info("Starting all broadcasts %s s", round(time.time() - start, 3))
# questions = [
# {
# 'type': 'list',
# 'name': 'theme',
# 'message': 'What type of annoucement would you like to make?',
# 'choices': [
# 'predefined',
# 'custom',
# 'audio'
# ]
# },
# ]
# answers = prompt(questions, style=custom_style_2)
# pprint(answers)