Spaces:
Sleeping
Sleeping
Delete src/tts_utils.py
Browse files- src/tts_utils.py +0 -96
src/tts_utils.py
DELETED
|
@@ -1,96 +0,0 @@
|
|
| 1 |
-
import os, uuid, asyncio
|
| 2 |
-
from typing import Dict
|
| 3 |
-
from pydub import AudioSegment
|
| 4 |
-
import edge_tts
|
| 5 |
-
|
| 6 |
-
from .config import TMP_DIR
|
| 7 |
-
|
| 8 |
-
EDGE_VOICES: Dict[str, str] = {}
|
| 9 |
-
|
| 10 |
-
async def fetch_edge_voices_async():
|
| 11 |
-
"""Charge dynamiquement toutes les voix FR/NL depuis Edge-TTS."""
|
| 12 |
-
global EDGE_VOICES
|
| 13 |
-
try:
|
| 14 |
-
voices = await edge_tts.list_voices()
|
| 15 |
-
filtered = [v for v in voices if v.get("Locale", "").startswith(("fr", "nl"))]
|
| 16 |
-
filtered.sort(key=lambda v: (v.get("Locale"), v.get("Gender"), v.get("ShortName")))
|
| 17 |
-
EDGE_VOICES = {
|
| 18 |
-
f"{v['ShortName']} - {v['Locale']} ({v['Gender']})": v["ShortName"]
|
| 19 |
-
for v in filtered
|
| 20 |
-
}
|
| 21 |
-
print(f"[Edge-TTS] {len(EDGE_VOICES)} voix FR/NL chargées.")
|
| 22 |
-
except Exception as e:
|
| 23 |
-
print(f"[Edge-TTS] Erreur chargement voix : {e}")
|
| 24 |
-
EDGE_VOICES.update({
|
| 25 |
-
"fr-FR-DeniseNeural - fr-FR (Female)": "fr-FR-DeniseNeural",
|
| 26 |
-
"nl-NL-MaaikeNeural - nl-NL (Female)": "nl-NL-MaaikeNeural",
|
| 27 |
-
})
|
| 28 |
-
|
| 29 |
-
def init_edge_voices():
|
| 30 |
-
"""Démarre le chargement asynchrone sans bloquer Gradio."""
|
| 31 |
-
try:
|
| 32 |
-
loop = asyncio.get_event_loop()
|
| 33 |
-
if loop.is_running():
|
| 34 |
-
import nest_asyncio
|
| 35 |
-
nest_asyncio.apply()
|
| 36 |
-
loop.create_task(fetch_edge_voices_async())
|
| 37 |
-
else:
|
| 38 |
-
loop.run_until_complete(fetch_edge_voices_async())
|
| 39 |
-
except RuntimeError:
|
| 40 |
-
asyncio.run(fetch_edge_voices_async())
|
| 41 |
-
|
| 42 |
-
def get_edge_voices(lang="fr"):
|
| 43 |
-
"""Retourne les voix déjà chargées (selon la langue)."""
|
| 44 |
-
global EDGE_VOICES
|
| 45 |
-
if not EDGE_VOICES:
|
| 46 |
-
init_edge_voices()
|
| 47 |
-
if lang == "fr":
|
| 48 |
-
return [v for k, v in EDGE_VOICES.items() if k.startswith("fr-")]
|
| 49 |
-
elif lang == "nl":
|
| 50 |
-
return [v for k, v in EDGE_VOICES.items() if k.startswith("nl-")]
|
| 51 |
-
return list(EDGE_VOICES.values())
|
| 52 |
-
|
| 53 |
-
async def _edge_tts_async(text, voice, outfile):
|
| 54 |
-
communicate = edge_tts.Communicate(text, voice)
|
| 55 |
-
await communicate.save(outfile)
|
| 56 |
-
return outfile
|
| 57 |
-
|
| 58 |
-
def tts_edge(text: str, voice: str = "fr-FR-DeniseNeural") -> str:
|
| 59 |
-
"""Génère un fichier WAV avec Edge-TTS (et fallback gTTS)."""
|
| 60 |
-
out_mp3 = os.path.join(TMP_DIR, f"edge_{uuid.uuid4().hex}.mp3")
|
| 61 |
-
try:
|
| 62 |
-
try:
|
| 63 |
-
loop = asyncio.get_event_loop()
|
| 64 |
-
if loop.is_running():
|
| 65 |
-
import nest_asyncio
|
| 66 |
-
nest_asyncio.apply()
|
| 67 |
-
except RuntimeError:
|
| 68 |
-
pass
|
| 69 |
-
asyncio.run(_edge_tts_async(text, voice, out_mp3))
|
| 70 |
-
|
| 71 |
-
out_wav = os.path.join(TMP_DIR, f"edge_{uuid.uuid4().hex}.wav")
|
| 72 |
-
AudioSegment.from_file(out_mp3).export(out_wav, format="wav")
|
| 73 |
-
os.remove(out_mp3)
|
| 74 |
-
return out_wav
|
| 75 |
-
except Exception as e:
|
| 76 |
-
print(f"[Edge-TTS] Erreur : {e} → fallback gTTS")
|
| 77 |
-
return tts_gtts(text, lang="fr" if voice.startswith("fr") else "nl")
|
| 78 |
-
|
| 79 |
-
def tts_gtts(text: str, lang: str = "fr") -> str:
|
| 80 |
-
from gtts import gTTS
|
| 81 |
-
out = os.path.join(TMP_DIR, f"gtts_{uuid.uuid4().hex}.mp3")
|
| 82 |
-
gTTS(text=text, lang=lang).save(out)
|
| 83 |
-
out_wav = os.path.join(TMP_DIR, f"gtts_{uuid.uuid4().hex}.wav")
|
| 84 |
-
AudioSegment.from_file(out).export(out_wav, format="wav")
|
| 85 |
-
os.remove(out)
|
| 86 |
-
return out_wav
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
def normalize_audio_to_wav(in_path: str) -> str:
|
| 90 |
-
"""Convertit n'importe quel format en WAV 44.1kHz stéréo."""
|
| 91 |
-
from pydub import AudioSegment
|
| 92 |
-
wav_path = os.path.join(TMP_DIR, f"norm_{uuid.uuid4().hex}.wav")
|
| 93 |
-
snd = AudioSegment.from_file(in_path)
|
| 94 |
-
snd = snd.set_frame_rate(44100).set_channels(2).set_sample_width(2)
|
| 95 |
-
snd.export(wav_path, format="wav")
|
| 96 |
-
return wav_path
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|