Spaces:
Runtime error
Runtime error
Randriamiandry commited on
Commit ·
aacfb92
1
Parent(s): bc155b3
Ajout du clonage vocal et interface complète
Browse files- --output +0 -0
- -H +0 -0
- -d +0 -0
- app.py +122 -25
- static/index.html +381 -0
--output
ADDED
|
File without changes
|
-H
ADDED
|
File without changes
|
-d
ADDED
|
File without changes
|
app.py
CHANGED
|
@@ -1,54 +1,151 @@
|
|
| 1 |
import io
|
| 2 |
-
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
| 4 |
from pydantic import BaseModel
|
| 5 |
from pocket_tts import TTSModel
|
| 6 |
import scipy.io.wavfile
|
| 7 |
import numpy as np
|
|
|
|
| 8 |
|
| 9 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
print("Chargement du modèle Pocket TTS...")
|
| 11 |
tts_model = TTSModel.load_model()
|
| 12 |
-
|
| 13 |
-
# Sélection d'une voix pré-définie. La liste est disponible ici :
|
| 14 |
-
# https://huggingface.co/kyutai/tts-voices
|
| 15 |
-
# Tu peux aussi utiliser un fichier audio local (.wav)
|
| 16 |
-
VOICE_NAME = "alba" # Choisis parmi 'alba', 'gentle', 'friendly', etc.
|
| 17 |
-
voice_state = tts_model.get_state_for_audio_prompt(VOICE_NAME)
|
| 18 |
print("Modèle chargé et prêt !")
|
| 19 |
-
# --- Fin du chargement ---
|
| 20 |
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
class TTSRequest(BaseModel):
|
| 27 |
text: str
|
|
|
|
| 28 |
|
| 29 |
-
@app.get("/")
|
| 30 |
-
def root():
|
| 31 |
-
return {"message": "Bienvenue sur l'API Pocket TTS. Utilisez l'endpoint /tts pour générer de l'audio."}
|
| 32 |
|
| 33 |
@app.post("/tts")
|
| 34 |
async def generate_speech(request: TTSRequest):
|
| 35 |
"""
|
| 36 |
-
Reçoit du texte et retourne un fichier audio WAV.
|
| 37 |
"""
|
| 38 |
text = request.text
|
| 39 |
if not text:
|
| 40 |
raise HTTPException(status_code=400, detail="Le texte ne peut pas être vide.")
|
| 41 |
|
| 42 |
-
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
-
|
| 45 |
-
audio_tensor = tts_model.generate_audio(voice_state, text)
|
| 46 |
|
| 47 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
audio_numpy = audio_tensor.numpy().astype(np.int16)
|
| 49 |
buffer = io.BytesIO()
|
| 50 |
scipy.io.wavfile.write(buffer, tts_model.sample_rate, audio_numpy)
|
| 51 |
-
buffer.seek(0)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
-
|
| 54 |
-
|
|
|
|
|
|
|
|
|
| 1 |
import io
|
| 2 |
+
import os
|
| 3 |
+
import tempfile
|
| 4 |
+
from fastapi import FastAPI, HTTPException, UploadFile, File, Form
|
| 5 |
+
from fastapi.responses import Response, FileResponse
|
| 6 |
+
from fastapi.staticfiles import StaticFiles
|
| 7 |
from pydantic import BaseModel
|
| 8 |
from pocket_tts import TTSModel
|
| 9 |
import scipy.io.wavfile
|
| 10 |
import numpy as np
|
| 11 |
+
import time
|
| 12 |
|
| 13 |
+
# Créer l'application FastAPI
|
| 14 |
+
app = FastAPI(title="API Pocket TTS",
|
| 15 |
+
description="Une API simple pour générer de la parole à partir de texte avec Pocket TTS, avec clonage de voix.",
|
| 16 |
+
version="2.0")
|
| 17 |
+
|
| 18 |
+
# Charger le modèle (une seule fois au démarrage)
|
| 19 |
print("Chargement du modèle Pocket TTS...")
|
| 20 |
tts_model = TTSModel.load_model()
|
| 21 |
+
print(f"Taux d'échantillonnage : {tts_model.sample_rate} Hz")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
print("Modèle chargé et prêt !")
|
|
|
|
| 23 |
|
| 24 |
+
# Dictionnaire pour stocker les états de voix (intégrées + clonées)
|
| 25 |
+
voices = {}
|
| 26 |
+
|
| 27 |
+
# Charger les voix intégrées
|
| 28 |
+
builtin_voices = ["alba", "marius", "javert", "jean", "fantine", "cosette", "eponine", "azelma"]
|
| 29 |
+
for voice_name in builtin_voices:
|
| 30 |
+
try:
|
| 31 |
+
voices[voice_name] = tts_model.get_state_for_audio_prompt(voice_name)
|
| 32 |
+
print(f"Voix intégrée chargée : {voice_name}")
|
| 33 |
+
except Exception as e:
|
| 34 |
+
print(f"Erreur lors du chargement de la voix {voice_name} : {e}")
|
| 35 |
+
|
| 36 |
+
print(f"{len(voices)} voix intégrées chargées. Prêtes pour la génération.")
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
# Servir les fichiers statiques (HTML, CSS, JS)
|
| 40 |
+
app.mount("/static", StaticFiles(directory="static"), name="static")
|
| 41 |
+
|
| 42 |
|
| 43 |
+
@app.get("/")
|
| 44 |
+
async def root():
|
| 45 |
+
return {"message": "Bienvenue sur l'API Pocket TTS. Utilisez l'interface /demo ou l'endpoint /tts pour générer de l'audio."}
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
@app.get("/demo")
|
| 49 |
+
async def demo():
|
| 50 |
+
return FileResponse("static/index.html")
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
# Modèle pour les requêtes JSON
|
| 54 |
class TTSRequest(BaseModel):
|
| 55 |
text: str
|
| 56 |
+
voice_name: str = "alba"
|
| 57 |
|
|
|
|
|
|
|
|
|
|
| 58 |
|
| 59 |
@app.post("/tts")
|
| 60 |
async def generate_speech(request: TTSRequest):
|
| 61 |
"""
|
| 62 |
+
Reçoit du texte et un nom de voix, retourne un fichier audio WAV.
|
| 63 |
"""
|
| 64 |
text = request.text
|
| 65 |
if not text:
|
| 66 |
raise HTTPException(status_code=400, detail="Le texte ne peut pas être vide.")
|
| 67 |
|
| 68 |
+
voice_name = request.voice_name
|
| 69 |
+
if voice_name not in voices:
|
| 70 |
+
available = ", ".join(voices.keys())
|
| 71 |
+
raise HTTPException(status_code=400, detail=f"Voix '{voice_name}' inconnue. Voix disponibles : {available}")
|
| 72 |
|
| 73 |
+
print(f"Génération avec la voix '{voice_name}' pour : {text[:50]}...")
|
|
|
|
| 74 |
|
| 75 |
+
# Générer l'audio
|
| 76 |
+
try:
|
| 77 |
+
# Pour une voix clonée, voice_state est déjà le résultat de get_state_for_audio_prompt
|
| 78 |
+
voice_state = voices[voice_name]
|
| 79 |
+
audio_tensor = tts_model.generate_audio(voice_state, text)
|
| 80 |
+
except Exception as e:
|
| 81 |
+
raise HTTPException(status_code=500, detail=f"Erreur de génération audio : {str(e)}")
|
| 82 |
+
|
| 83 |
+
# Convertir en numpy et sauvegarder dans un buffer
|
| 84 |
audio_numpy = audio_tensor.numpy().astype(np.int16)
|
| 85 |
buffer = io.BytesIO()
|
| 86 |
scipy.io.wavfile.write(buffer, tts_model.sample_rate, audio_numpy)
|
| 87 |
+
buffer.seek(0)
|
| 88 |
+
|
| 89 |
+
# Retourner le fichier audio
|
| 90 |
+
return Response(content=buffer.read(), media_type="audio/wav", headers={"Content-Disposition": "attachment; filename=speech.wav"})
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
@app.post("/clone-voice")
|
| 94 |
+
async def clone_voice(audio_file: UploadFile = File(...), voice_name: str = Form(...)):
|
| 95 |
+
"""
|
| 96 |
+
Endpoint pour cloner une voix à partir d'un fichier audio.
|
| 97 |
+
"""
|
| 98 |
+
# Vérifier que le nom de la voix n'existe pas déjà
|
| 99 |
+
if voice_name in voices:
|
| 100 |
+
raise HTTPException(status_code=400, detail=f"Une voix nommée '{voice_name}' existe déjà. Veuillez choisir un autre nom.")
|
| 101 |
+
|
| 102 |
+
# Vérifier le type de fichier
|
| 103 |
+
if not audio_file.filename.endswith(('.wav', '.mp3', '.m4a')):
|
| 104 |
+
raise HTTPException(status_code=400, detail="Format de fichier non supporté. Utilisez WAV, MP3 ou M4A.")
|
| 105 |
+
|
| 106 |
+
# Sauvegarder le fichier temporairement
|
| 107 |
+
try:
|
| 108 |
+
# Créer un fichier temporaire
|
| 109 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
|
| 110 |
+
content = await audio_file.read()
|
| 111 |
+
tmp_file.write(content)
|
| 112 |
+
tmp_path = tmp_file.name
|
| 113 |
+
|
| 114 |
+
# Si ce n'est pas déjà un WAV, convertir avec pydub (optionnel)
|
| 115 |
+
if not audio_file.filename.endswith('.wav'):
|
| 116 |
+
# Ici, tu pourrais ajouter une conversion avec pydub si nécessaire
|
| 117 |
+
# Pour simplifier, on va supposer que le modèle peut gérer les MP3
|
| 118 |
+
pass
|
| 119 |
+
|
| 120 |
+
# Cloner la voix
|
| 121 |
+
voice_state = tts_model.get_state_for_audio_prompt(tmp_path)
|
| 122 |
+
voices[voice_name] = voice_state
|
| 123 |
+
|
| 124 |
+
# Nettoyer
|
| 125 |
+
os.unlink(tmp_path)
|
| 126 |
+
|
| 127 |
+
print(f"Voix clonée '{voice_name}' ajoutée avec succès. Nombre total de voix : {len(voices)}")
|
| 128 |
+
return {"message": f"Voix '{voice_name}' clonée avec succès.", "voice_name": voice_name, "total_voices": len(voices)}
|
| 129 |
+
|
| 130 |
+
except Exception as e:
|
| 131 |
+
# Nettoyer en cas d'erreur
|
| 132 |
+
if 'tmp_path' in locals():
|
| 133 |
+
try:
|
| 134 |
+
os.unlink(tmp_path)
|
| 135 |
+
except:
|
| 136 |
+
pass
|
| 137 |
+
raise HTTPException(status_code=500, detail=f"Erreur lors du clonage : {str(e)}")
|
| 138 |
+
|
| 139 |
+
|
| 140 |
+
@app.get("/voices")
|
| 141 |
+
async def list_voices():
|
| 142 |
+
"""
|
| 143 |
+
Retourne la liste des voix disponibles.
|
| 144 |
+
"""
|
| 145 |
+
return {"voices": list(voices.keys())}
|
| 146 |
+
|
| 147 |
|
| 148 |
+
# Démarrer le serveur si exécuté directement
|
| 149 |
+
if __name__ == "__main__":
|
| 150 |
+
import uvicorn
|
| 151 |
+
uvicorn.run(app, host="0.0.0.0", port=7860)
|
static/index.html
ADDED
|
@@ -0,0 +1,381 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<!DOCTYPE html>
|
| 2 |
+
<html lang="fr">
|
| 3 |
+
<head>
|
| 4 |
+
<meta charset="UTF-8">
|
| 5 |
+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
| 6 |
+
<title>Pocket TTS - Synthèse vocale et clonage de voix</title>
|
| 7 |
+
<style>
|
| 8 |
+
* {
|
| 9 |
+
margin: 0;
|
| 10 |
+
padding: 0;
|
| 11 |
+
box-sizing: border-box;
|
| 12 |
+
}
|
| 13 |
+
|
| 14 |
+
body {
|
| 15 |
+
font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
|
| 16 |
+
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
|
| 17 |
+
min-height: 100vh;
|
| 18 |
+
padding: 20px;
|
| 19 |
+
}
|
| 20 |
+
|
| 21 |
+
.container {
|
| 22 |
+
max-width: 800px;
|
| 23 |
+
margin: 0 auto;
|
| 24 |
+
}
|
| 25 |
+
|
| 26 |
+
.card {
|
| 27 |
+
background: white;
|
| 28 |
+
border-radius: 20px;
|
| 29 |
+
padding: 30px;
|
| 30 |
+
margin-bottom: 20px;
|
| 31 |
+
box-shadow: 0 10px 40px rgba(0,0,0,0.1);
|
| 32 |
+
}
|
| 33 |
+
|
| 34 |
+
h1 {
|
| 35 |
+
color: #667eea;
|
| 36 |
+
margin-bottom: 10px;
|
| 37 |
+
}
|
| 38 |
+
|
| 39 |
+
h2 {
|
| 40 |
+
color: #764ba2;
|
| 41 |
+
margin: 20px 0 15px 0;
|
| 42 |
+
font-size: 1.3em;
|
| 43 |
+
}
|
| 44 |
+
|
| 45 |
+
textarea, select, input {
|
| 46 |
+
width: 100%;
|
| 47 |
+
padding: 12px;
|
| 48 |
+
margin: 10px 0;
|
| 49 |
+
border: 2px solid #e0e0e0;
|
| 50 |
+
border-radius: 10px;
|
| 51 |
+
font-size: 16px;
|
| 52 |
+
transition: all 0.3s;
|
| 53 |
+
}
|
| 54 |
+
|
| 55 |
+
textarea:focus, select:focus, input:focus {
|
| 56 |
+
outline: none;
|
| 57 |
+
border-color: #667eea;
|
| 58 |
+
}
|
| 59 |
+
|
| 60 |
+
button {
|
| 61 |
+
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
|
| 62 |
+
color: white;
|
| 63 |
+
border: none;
|
| 64 |
+
padding: 12px 30px;
|
| 65 |
+
border-radius: 25px;
|
| 66 |
+
font-size: 16px;
|
| 67 |
+
cursor: pointer;
|
| 68 |
+
transition: transform 0.2s;
|
| 69 |
+
margin-top: 10px;
|
| 70 |
+
}
|
| 71 |
+
|
| 72 |
+
button:hover {
|
| 73 |
+
transform: translateY(-2px);
|
| 74 |
+
}
|
| 75 |
+
|
| 76 |
+
button:active {
|
| 77 |
+
transform: translateY(0);
|
| 78 |
+
}
|
| 79 |
+
|
| 80 |
+
audio {
|
| 81 |
+
width: 100%;
|
| 82 |
+
margin-top: 20px;
|
| 83 |
+
}
|
| 84 |
+
|
| 85 |
+
.status {
|
| 86 |
+
margin-top: 15px;
|
| 87 |
+
padding: 10px;
|
| 88 |
+
border-radius: 10px;
|
| 89 |
+
font-size: 14px;
|
| 90 |
+
}
|
| 91 |
+
|
| 92 |
+
.status.success {
|
| 93 |
+
background-color: #d4edda;
|
| 94 |
+
color: #155724;
|
| 95 |
+
}
|
| 96 |
+
|
| 97 |
+
.status.error {
|
| 98 |
+
background-color: #f8d7da;
|
| 99 |
+
color: #721c24;
|
| 100 |
+
}
|
| 101 |
+
|
| 102 |
+
.status.info {
|
| 103 |
+
background-color: #d1ecf1;
|
| 104 |
+
color: #0c5460;
|
| 105 |
+
}
|
| 106 |
+
|
| 107 |
+
.voice-list {
|
| 108 |
+
display: flex;
|
| 109 |
+
flex-wrap: wrap;
|
| 110 |
+
gap: 10px;
|
| 111 |
+
margin: 15px 0;
|
| 112 |
+
}
|
| 113 |
+
|
| 114 |
+
.voice-tag {
|
| 115 |
+
background: #f0f0f0;
|
| 116 |
+
padding: 5px 12px;
|
| 117 |
+
border-radius: 20px;
|
| 118 |
+
font-size: 14px;
|
| 119 |
+
cursor: pointer;
|
| 120 |
+
transition: all 0.2s;
|
| 121 |
+
}
|
| 122 |
+
|
| 123 |
+
.voice-tag:hover {
|
| 124 |
+
background: #667eea;
|
| 125 |
+
color: white;
|
| 126 |
+
}
|
| 127 |
+
|
| 128 |
+
.file-upload {
|
| 129 |
+
border: 2px dashed #667eea;
|
| 130 |
+
background: #f8f9ff;
|
| 131 |
+
text-align: center;
|
| 132 |
+
padding: 20px;
|
| 133 |
+
border-radius: 10px;
|
| 134 |
+
cursor: pointer;
|
| 135 |
+
}
|
| 136 |
+
|
| 137 |
+
.file-upload input {
|
| 138 |
+
display: none;
|
| 139 |
+
}
|
| 140 |
+
|
| 141 |
+
hr {
|
| 142 |
+
margin: 20px 0;
|
| 143 |
+
border: none;
|
| 144 |
+
border-top: 2px solid #e0e0e0;
|
| 145 |
+
}
|
| 146 |
+
</style>
|
| 147 |
+
</head>
|
| 148 |
+
<body>
|
| 149 |
+
<div class="container">
|
| 150 |
+
<div class="card">
|
| 151 |
+
<h1>🎤 Pocket TTS Studio</h1>
|
| 152 |
+
<p>Synthèse vocale et clonage de voix - Interface complète</p>
|
| 153 |
+
</div>
|
| 154 |
+
|
| 155 |
+
<!-- Section Synthèse vocale -->
|
| 156 |
+
<div class="card">
|
| 157 |
+
<h2>📝 1. Synthèse vocale</h2>
|
| 158 |
+
<textarea id="text" rows="4" placeholder="Saisissez le texte à synthétiser...">Bonjour, ceci est un test de mon API vocale avec clonage de voix.</textarea>
|
| 159 |
+
|
| 160 |
+
<label for="voice">Choisir une voix :</label>
|
| 161 |
+
<select id="voice">
|
| 162 |
+
<option value="">Chargement des voix...</option>
|
| 163 |
+
</select>
|
| 164 |
+
|
| 165 |
+
<button id="generateBtn">🎵 Générer l'audio</button>
|
| 166 |
+
<audio id="audioPlayer" controls style="display: none;"></audio>
|
| 167 |
+
<div id="ttsStatus" class="status"></div>
|
| 168 |
+
</div>
|
| 169 |
+
|
| 170 |
+
<!-- Section Clonage de voix -->
|
| 171 |
+
<div class="card">
|
| 172 |
+
<h2>🎙️ 2. Cloner une nouvelle voix</h2>
|
| 173 |
+
<div class="file-upload" id="uploadArea">
|
| 174 |
+
<input type="file" id="audioFile" accept=".wav,.mp3,.m4a">
|
| 175 |
+
<p>📁 Cliquez ou glissez un fichier audio (WAV, MP3, M4A)</p>
|
| 176 |
+
<p style="font-size: 12px; color: #666;">5-15 secondes pour un meilleur résultat</p>
|
| 177 |
+
</div>
|
| 178 |
+
|
| 179 |
+
<input type="text" id="newVoiceName" placeholder="Nom de la nouvelle voix (ex: ma_voix)" style="margin-top: 15px;">
|
| 180 |
+
<button id="cloneBtn">✨ Cloner cette voix</button>
|
| 181 |
+
<div id="cloneStatus" class="status"></div>
|
| 182 |
+
</div>
|
| 183 |
+
|
| 184 |
+
<!-- Section Liste des voix -->
|
| 185 |
+
<div class="card">
|
| 186 |
+
<h2>🎧 3. Voix disponibles</h2>
|
| 187 |
+
<div id="voiceList" class="voice-list">
|
| 188 |
+
<p>Chargement des voix...</p>
|
| 189 |
+
</div>
|
| 190 |
+
<button id="refreshVoicesBtn" style="background: #6c757d;">🔄 Actualiser la liste</button>
|
| 191 |
+
</div>
|
| 192 |
+
</div>
|
| 193 |
+
|
| 194 |
+
<script>
|
| 195 |
+
let currentAudioBlob = null;
|
| 196 |
+
|
| 197 |
+
// Fonction pour charger la liste des voix
|
| 198 |
+
async function loadVoices() {
|
| 199 |
+
try {
|
| 200 |
+
const response = await fetch('/voices');
|
| 201 |
+
if (response.ok) {
|
| 202 |
+
const data = await response.json();
|
| 203 |
+
const voiceSelect = document.getElementById('voice');
|
| 204 |
+
const voiceListDiv = document.getElementById('voiceList');
|
| 205 |
+
|
| 206 |
+
// Mettre à jour le select
|
| 207 |
+
voiceSelect.innerHTML = '';
|
| 208 |
+
data.voices.forEach(voice => {
|
| 209 |
+
const option = document.createElement('option');
|
| 210 |
+
option.value = voice;
|
| 211 |
+
option.textContent = voice;
|
| 212 |
+
voiceSelect.appendChild(option);
|
| 213 |
+
});
|
| 214 |
+
|
| 215 |
+
// Mettre à jour l'affichage des tags
|
| 216 |
+
voiceListDiv.innerHTML = '';
|
| 217 |
+
data.voices.forEach(voice => {
|
| 218 |
+
const tag = document.createElement('div');
|
| 219 |
+
tag.className = 'voice-tag';
|
| 220 |
+
tag.textContent = voice;
|
| 221 |
+
tag.onclick = () => {
|
| 222 |
+
document.getElementById('voice').value = voice;
|
| 223 |
+
};
|
| 224 |
+
voiceListDiv.appendChild(tag);
|
| 225 |
+
});
|
| 226 |
+
|
| 227 |
+
if (data.voices.length === 0) {
|
| 228 |
+
voiceListDiv.innerHTML = '<p>Aucune voix disponible.</p>';
|
| 229 |
+
}
|
| 230 |
+
}
|
| 231 |
+
} catch (error) {
|
| 232 |
+
console.error('Erreur chargement voix:', error);
|
| 233 |
+
}
|
| 234 |
+
}
|
| 235 |
+
|
| 236 |
+
// Génération audio
|
| 237 |
+
document.getElementById('generateBtn').onclick = async () => {
|
| 238 |
+
const text = document.getElementById('text').value.trim();
|
| 239 |
+
const voiceName = document.getElementById('voice').value;
|
| 240 |
+
const statusDiv = document.getElementById('ttsStatus');
|
| 241 |
+
const audioPlayer = document.getElementById('audioPlayer');
|
| 242 |
+
|
| 243 |
+
if (!text) {
|
| 244 |
+
statusDiv.className = 'status error';
|
| 245 |
+
statusDiv.textContent = 'Veuillez saisir du texte.';
|
| 246 |
+
return;
|
| 247 |
+
}
|
| 248 |
+
|
| 249 |
+
if (!voiceName) {
|
| 250 |
+
statusDiv.className = 'status error';
|
| 251 |
+
statusDiv.textContent = 'Veuillez sélectionner une voix.';
|
| 252 |
+
return;
|
| 253 |
+
}
|
| 254 |
+
|
| 255 |
+
statusDiv.className = 'status info';
|
| 256 |
+
statusDiv.textContent = 'Génération en cours...';
|
| 257 |
+
audioPlayer.style.display = 'none';
|
| 258 |
+
|
| 259 |
+
try {
|
| 260 |
+
const response = await fetch('/tts', {
|
| 261 |
+
method: 'POST',
|
| 262 |
+
headers: { 'Content-Type': 'application/json' },
|
| 263 |
+
body: JSON.stringify({ text: text, voice_name: voiceName })
|
| 264 |
+
});
|
| 265 |
+
|
| 266 |
+
if (response.ok) {
|
| 267 |
+
currentAudioBlob = await response.blob();
|
| 268 |
+
const audioUrl = URL.createObjectURL(currentAudioBlob);
|
| 269 |
+
audioPlayer.src = audioUrl;
|
| 270 |
+
audioPlayer.style.display = 'block';
|
| 271 |
+
audioPlayer.play();
|
| 272 |
+
statusDiv.className = 'status success';
|
| 273 |
+
statusDiv.textContent = '✅ Audio généré avec succès !';
|
| 274 |
+
} else {
|
| 275 |
+
const errorText = await response.text();
|
| 276 |
+
statusDiv.className = 'status error';
|
| 277 |
+
statusDiv.textContent = `Erreur : ${errorText}`;
|
| 278 |
+
}
|
| 279 |
+
} catch (error) {
|
| 280 |
+
statusDiv.className = 'status error';
|
| 281 |
+
statusDiv.textContent = `Erreur réseau : ${error.message}`;
|
| 282 |
+
}
|
| 283 |
+
};
|
| 284 |
+
|
| 285 |
+
// Clonage de voix
|
| 286 |
+
document.getElementById('cloneBtn').onclick = async () => {
|
| 287 |
+
const fileInput = document.getElementById('audioFile');
|
| 288 |
+
const voiceName = document.getElementById('newVoiceName').value.trim();
|
| 289 |
+
const statusDiv = document.getElementById('cloneStatus');
|
| 290 |
+
|
| 291 |
+
if (!fileInput.files || fileInput.files.length === 0) {
|
| 292 |
+
statusDiv.className = 'status error';
|
| 293 |
+
statusDiv.textContent = 'Veuillez sélectionner un fichier audio.';
|
| 294 |
+
return;
|
| 295 |
+
}
|
| 296 |
+
|
| 297 |
+
if (!voiceName) {
|
| 298 |
+
statusDiv.className = 'status error';
|
| 299 |
+
statusDiv.textContent = 'Veuillez donner un nom à cette voix.';
|
| 300 |
+
return;
|
| 301 |
+
}
|
| 302 |
+
|
| 303 |
+
// Vérification simple du nom (pas d'espaces, caractères spéciaux)
|
| 304 |
+
if (!/^[a-zA-Z0-9_-]+$/.test(voiceName)) {
|
| 305 |
+
statusDiv.className = 'status error';
|
| 306 |
+
statusDiv.textContent = 'Le nom ne doit contenir que des lettres, chiffres, tirets ou underscores.';
|
| 307 |
+
return;
|
| 308 |
+
}
|
| 309 |
+
|
| 310 |
+
const formData = new FormData();
|
| 311 |
+
formData.append('audio_file', fileInput.files[0]);
|
| 312 |
+
formData.append('voice_name', voiceName);
|
| 313 |
+
|
| 314 |
+
statusDiv.className = 'status info';
|
| 315 |
+
statusDiv.textContent = 'Clonage en cours (cela peut prendre quelques secondes)...';
|
| 316 |
+
|
| 317 |
+
try {
|
| 318 |
+
const response = await fetch('/clone-voice', {
|
| 319 |
+
method: 'POST',
|
| 320 |
+
body: formData
|
| 321 |
+
});
|
| 322 |
+
|
| 323 |
+
if (response.ok) {
|
| 324 |
+
const result = await response.json();
|
| 325 |
+
statusDiv.className = 'status success';
|
| 326 |
+
statusDiv.textContent = `✅ Voix "${voiceName}" clonée avec succès ! (Total: ${result.total_voices} voix)`;
|
| 327 |
+
// Réinitialiser les champs
|
| 328 |
+
fileInput.value = '';
|
| 329 |
+
document.getElementById('newVoiceName').value = '';
|
| 330 |
+
// Recharger la liste des voix
|
| 331 |
+
await loadVoices();
|
| 332 |
+
} else {
|
| 333 |
+
const errorText = await response.text();
|
| 334 |
+
statusDiv.className = 'status error';
|
| 335 |
+
statusDiv.textContent = `Erreur : ${errorText}`;
|
| 336 |
+
}
|
| 337 |
+
} catch (error) {
|
| 338 |
+
statusDiv.className = 'status error';
|
| 339 |
+
statusDiv.textContent = `Erreur réseau : ${error.message}`;
|
| 340 |
+
}
|
| 341 |
+
};
|
| 342 |
+
|
| 343 |
+
// Rafraîchir la liste des voix
|
| 344 |
+
document.getElementById('refreshVoicesBtn').onclick = loadVoices;
|
| 345 |
+
|
| 346 |
+
// Upload zone - drag & drop
|
| 347 |
+
const uploadArea = document.getElementById('uploadArea');
|
| 348 |
+
const fileInput = document.getElementById('audioFile');
|
| 349 |
+
|
| 350 |
+
uploadArea.onclick = () => fileInput.click();
|
| 351 |
+
|
| 352 |
+
uploadArea.ondragover = (e) => {
|
| 353 |
+
e.preventDefault();
|
| 354 |
+
uploadArea.style.borderColor = '#764ba2';
|
| 355 |
+
uploadArea.style.background = '#f0e6ff';
|
| 356 |
+
};
|
| 357 |
+
|
| 358 |
+
uploadArea.ondragleave = (e) => {
|
| 359 |
+
e.preventDefault();
|
| 360 |
+
uploadArea.style.borderColor = '#667eea';
|
| 361 |
+
uploadArea.style.background = '#f8f9ff';
|
| 362 |
+
};
|
| 363 |
+
|
| 364 |
+
uploadArea.ondrop = (e) => {
|
| 365 |
+
e.preventDefault();
|
| 366 |
+
uploadArea.style.borderColor = '#667eea';
|
| 367 |
+
uploadArea.style.background = '#f8f9ff';
|
| 368 |
+
const files = e.dataTransfer.files;
|
| 369 |
+
if (files.length > 0 && (files[0].type.includes('audio'))) {
|
| 370 |
+
fileInput.files = files;
|
| 371 |
+
statusDiv.textContent = `Fichier sélectionné : ${files[0].name}`;
|
| 372 |
+
} else {
|
| 373 |
+
alert('Veuillez déposer un fichier audio valide.');
|
| 374 |
+
}
|
| 375 |
+
};
|
| 376 |
+
|
| 377 |
+
// Initialiser
|
| 378 |
+
loadVoices();
|
| 379 |
+
</script>
|
| 380 |
+
</body>
|
| 381 |
+
</html>
|