Migration vers Kokoro-82M (rapide)

Dockerfile

@@ -1,6 +1,6 @@
 FROM python:3.11-slim
-RUN apt-get update && apt-get install -y ffmpeg && rm -rf /var/lib/apt/lists/*
 WORKDIR /app
+RUN apt-get update && apt-get install -y ffmpeg
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 COPY . .

app.py

@@ -1,55 +1,47 @@
 import io
 import os
 import tempfile
-import time
 import numpy as np
 from fastapi import FastAPI, HTTPException, UploadFile, File, Form
-from fastapi.responses import StreamingResponse, FileResponse
+from fastapi.responses import Response, FileResponse
 from fastapi.staticfiles import StaticFiles
 from pydantic import BaseModel
-import scipy.io.wavfile
-from pydub import AudioSegment
+import soundfile as sf
+import time
 
-# ---------- Configuration ----------
-HF_TOKEN = os.environ.get("HF_TOKEN")
-print("🚀 Démarrage API Pocket TTS (streaming + clonage)")
+print("🚀 Démarrage API Kokoro-82M (ONNX rapide)")
 
-if HF_TOKEN:
-    from huggingface_hub import login
-    login(token=HF_TOKEN)
-    print("✅ Authentifié HF")
+# -------------------------------------------------------------
+# Chargement du modèle Kokoro (version ONNX quantifiée)
+# -------------------------------------------------------------
+from kokoro_onnx import KokoroONNX
 
-# ---------- Chargement du modèle (API correcte) ----------
-from pocket_tts import TTSModel
+# Modèle recommandé (quantifié automatiquement)
+MODEL_NAME = "onnx-community/Kokoro-82M-v1.0-ONNX"
 
-print("📥 Chargement du modèle Pocket TTS...")
-model = TTSModel.load_model()
-print(f"✅ Modèle chargé | Sample rate: {model.sample_rate} Hz")
+# Téléchargement et chargement
+tts = KokoroONNX.from_pretrained(MODEL_NAME, device="cpu")
+print(f"✅ Kokoro-82M chargé | Voices disponibles: {list(tts.voices.keys())}")
 
-# ---------- Dictionnaire des voix ----------
-voices = {}
+# Dictionnaire pour les voix clonées (zero-shot)
+cloned_voices = {}
 
-# Liste des voix intégrées (disponibles dans le modèle)
-builtin_voices = ["alba", "marius", "javert", "jean", "fantine", "cosette", "eponine", "azelma"]
-for v in builtin_voices:
-    try:
-        # Pour les voix intégrées, on passe simplement le nom
-        voices[v] = model.get_state_for_audio_prompt(v)
-        print(f"✓ Voix intégrée: {v}")
-    except Exception as e:
-        print(f"✗ Erreur chargement {v}: {e}")
+# -------------------------------------------------------------
+# Application FastAPI
+# -------------------------------------------------------------
+app = FastAPI(title="Kokoro TTS API", version="2.0")
 
-print(f"{len(voices)} voix disponibles (intégrées).")
-
-# ---------- Application FastAPI ----------
-app = FastAPI(title="Pocket TTS API", version="3.1")
-
-# Dossier statique pour l'interface
+# Servir l'interface HTML (optionnel)
 app.mount("/static", StaticFiles(directory="static"), name="static")
 
 @app.get("/")
 async def root():
-    return {"status": "ok", "model": "Pocket TTS", "endpoints": ["/tts", "/clone-voice", "/voices"]}
+    return {
+        "status": "ok",
+        "model": "Kokoro-82M (ONNX quantifié)",
+        "available_voices": list(tts.voices.keys()),
+        "endpoints": ["/tts", "/clone-voice", "/voices"]
+    }
 
 @app.get("/demo")
 async def demo():
@@ -57,12 +49,9 @@ async def demo():
 
 class TTSRequest(BaseModel):
     text: str
-    voice_name: str = "alba"
+    voice: str = "af_heart"      # voix par défaut (anglais)
     speed: float = 1.0
-
-def generate_wav_chunks(wav_bytes: bytes, chunk_size=8192):
-    for i in range(0, len(wav_bytes), chunk_size):
-        yield wav_bytes[i:i+chunk_size]
+    lang: str = "fr"             # "fr" ou "en"
 
 @app.post("/tts")
 async def generate_speech(request: TTSRequest):
@@ -70,39 +59,41 @@ async def generate_speech(request: TTSRequest):
     text = request.text.strip()
     if not text:
         raise HTTPException(400, "Texte vide")
-    if request.voice_name not in voices:
-        available = ", ".join(voices.keys())
-        raise HTTPException(400, f"Voix inconnue. Disponibles: {available}")
 
-    try:
-        voice_state = voices[request.voice_name]
-        # generate_audio attend (state, text) sans paramètre speed dans cette version
-        # On ignore speed si non supporté
-        audio_tensor = model.generate_audio(voice_state, text)
-        audio_np = audio_tensor.numpy()
-        # Conversion float -> int16
-        if audio_np.dtype in (np.float32, np.float64):
-            audio_np = np.clip(audio_np, -1.0, 1.0)
-            audio_np = (audio_np * 32767).astype(np.int16)
-        else:
-            audio_np = audio_np.astype(np.int16)
+    # Choix de la voix : clonée ou intégrée
+    if request.voice in cloned_voices:
+        voice = cloned_voices[request.voice]
+    elif request.voice in tts.voices:
+        voice = request.voice
+    else:
+        # Fallback sur une voix par défaut selon la langue
+        voice = "af_heart" if request.lang == "en" else "fr_001"
+        if voice not in tts.voices:
+            voice = list(tts.voices.keys())[0]
 
-        buffer = io.BytesIO()
-        scipy.io.wavfile.write(buffer, model.sample_rate, audio_np)
-        wav_bytes = buffer.getvalue()
-        print(f"✅ Généré en {time.time()-start:.2f}s | Voix: {request.voice_name}")
-        return StreamingResponse(
-            generate_wav_chunks(wav_bytes),
-            media_type="audio/wav",
-            headers={"Content-Disposition": f"attachment; filename=speech_{request.voice_name}.wav"}
+    try:
+        # Génération audio
+        audio, sample_rate = tts.tts(
+            text=text,
+            voice=voice,
+            speed=request.speed,
+            lang=request.lang
         )
+        # audio est un numpy array (float32)
+        buffer = io.BytesIO()
+        sf.write(buffer, audio, sample_rate, format="WAV")
+        buffer.seek(0)
+
+        print(f"✅ Généré en {time.time()-start:.2f}s | voix={voice} | langue={request.lang}")
+        return Response(content=buffer.read(), media_type="audio/wav")
     except Exception as e:
         raise HTTPException(500, f"Erreur TTS: {str(e)}")
 
 @app.post("/clone-voice")
 async def clone_voice(audio_file: UploadFile = File(...), voice_name: str = Form(...)):
-    if voice_name in voices:
-        raise HTTPException(400, f"La voix '{voice_name}' existe déjà.")
+    if voice_name in cloned_voices or voice_name in tts.voices:
+        raise HTTPException(400, f"La voix '{voice_name}' existe déjà (intégrée ou clonée).")
+
     # Sauvegarde temporaire
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
         content = await audio_file.read()
@@ -110,14 +101,10 @@ async def clone_voice(audio_file: UploadFile = File(...), voice_name: str = Form
         tmp_path = tmp.name
 
     try:
-        # Conversion en WAV standard (mono 24kHz) via pydub
-        audio = AudioSegment.from_file(tmp_path)
-        audio = audio.set_channels(1).set_frame_rate(24000)
-        audio.export(tmp_path, format="wav")
-        # Clonage
-        voice_state = model.get_state_for_audio_prompt(tmp_path)
-        voices[voice_name] = voice_state
-        return {"message": f"Voix '{voice_name}' clonée", "voice_name": voice_name, "total_voices": len(voices)}
+        # Clonage zero-shot (Kokoro-ONNX le permet)
+        cloned_voice = tts.clone_voice(tmp_path)
+        cloned_voices[voice_name] = cloned_voice
+        return {"message": f"Voix '{voice_name}' clonée", "voice_name": voice_name, "total_cloned": len(cloned_voices)}
     except Exception as e:
         raise HTTPException(500, f"Erreur clonage: {str(e)}")
     finally:
@@ -126,7 +113,10 @@ async def clone_voice(audio_file: UploadFile = File(...), voice_name: str = Form
 
 @app.get("/voices")
 async def list_voices():
-    return {"voices": list(voices.keys())}
+    return {
+        "builtin": list(tts.voices.keys()),
+        "cloned": list(cloned_voices.keys())
+    }
 
 if __name__ == "__main__":
     import uvicorn

requirements.txt

@@ -1,7 +1,6 @@
 fastapi
 uvicorn[standard]
-pocket-tts
-scipy
-pydub
-huggingface-hub
+soundfile
 numpy
+kokoro-onnx>=0.3.0
+huggingface-hub

static/index.html

@@ -1,409 +1,77 @@
 <!DOCTYPE html>
-<html lang="fr">
+<html>
 <head>
-    <meta charset="UTF-8">
-    <meta name="viewport" content="width=device-width, initial-scale=1.0">
-    <title>Pocket TTS Studio – Streaming</title>
+    <title>Kokoro TTS - rapide FR/EN</title>
     <style>
-        * {
-            margin: 0;
-            padding: 0;
-            box-sizing: border-box;
-        }
-
-        body {
-            font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
-            background: linear-gradient(135deg, #1a1a2e 0%, #16213e 100%);
-            min-height: 100vh;
-            padding: 20px;
-            color: #eee;
-        }
-
-        .container {
-            max-width: 900px;
-            margin: 0 auto;
-        }
-
-        .card {
-            background: rgba(255,255,255,0.1);
-            backdrop-filter: blur(10px);
-            border-radius: 20px;
-            padding: 25px;
-            margin-bottom: 20px;
-            box-shadow: 0 8px 32px rgba(0,0,0,0.2);
-            border: 1px solid rgba(255,255,255,0.2);
-        }
-
-        h1, h2 {
-            margin-bottom: 15px;
-            color: #fff;
-        }
-
-        textarea, select, input {
-            width: 100%;
-            padding: 12px;
-            margin: 10px 0;
-            border: none;
-            border-radius: 10px;
-            font-size: 16px;
-            background: rgba(255,255,255,0.9);
-            transition: all 0.3s;
-        }
-
-        textarea:focus, select:focus, input:focus {
-            outline: none;
-            background: white;
-            box-shadow: 0 0 0 2px #00a8ff;
-        }
-
-        button {
-            background: linear-gradient(135deg, #00a8ff 0%, #0077ff 100%);
-            color: white;
-            border: none;
-            padding: 12px 24px;
-            border-radius: 30px;
-            font-size: 16px;
-            cursor: pointer;
-            transition: transform 0.2s, box-shadow 0.2s;
-            margin-top: 10px;
-            margin-right: 10px;
-        }
-
-        button:hover {
-            transform: translateY(-2px);
-            box-shadow: 0 10px 20px rgba(0,0,0,0.2);
-        }
-
-        button:active {
-            transform: translateY(0);
-        }
-
-        .voice-list {
-            display: flex;
-            flex-wrap: wrap;
-            gap: 10px;
-            margin: 15px 0;
-        }
-
-        .voice-tag {
-            background: rgba(255,255,255,0.2);
-            padding: 6px 14px;
-            border-radius: 30px;
-            font-size: 14px;
-            cursor: pointer;
-            transition: all 0.2s;
-        }
-
-        .voice-tag:hover, .voice-tag.active {
-            background: #00a8ff;
-            color: white;
-        }
-
-        audio {
-            width: 100%;
-            margin-top: 20px;
-            border-radius: 30px;
-        }
-
-        .status {
-            margin-top: 15px;
-            padding: 10px;
-            border-radius: 10px;
-            font-size: 14px;
-        }
-
-        .status.success {
-            background: rgba(0,255,0,0.2);
-            color: #aaffaa;
-        }
-
-        .status.error {
-            background: rgba(255,0,0,0.2);
-            color: #ffaaaa;
-        }
-
-        .status.info {
-            background: rgba(0,168,255,0.2);
-            color: #aae0ff;
-        }
-
-        .file-upload {
-            border: 2px dashed #00a8ff;
-            background: rgba(0,168,255,0.1);
-            text-align: center;
-            padding: 20px;
-            border-radius: 15px;
-            cursor: pointer;
-            margin: 15px 0;
-        }
-
-        .file-upload input {
-            display: none;
-        }
-
-        .row {
-            display: flex;
-            gap: 15px;
-            align-items: center;
-        }
-
-        .speed-control {
-            flex: 1;
-        }
-
-        hr {
-            margin: 20px 0;
-            border-color: rgba(255,255,255,0.2);
-        }
+        body { font-family: sans-serif; max-width: 600px; margin: auto; padding: 2rem; }
+        textarea, select, input, button { width: 100%; margin: 0.5rem 0; padding: 0.5rem; }
+        audio { width: 100%; margin-top: 1rem; }
+        .status { margin-top: 1rem; color: gray; }
     </style>
 </head>
 <body>
-<div class="container">
-    <div class="card">
-        <h1>🎤 Pocket TTS Studio – Streaming Audio</h1>
-        <p>Synthèse vocale en temps réel (flux chunked) + clonage de voix</p>
-    </div>
-
-    <!-- Section Synthèse -->
-    <div class="card">
-        <h2>📝 1. Synthèse vocale</h2>
-        <textarea id="text" rows="4" placeholder="Saisissez le texte à synthétiser...">Bonjour, ceci est un test en streaming avec Pocket TTS. La voix est claire et le son se déclenche rapidement.</textarea>
-
-        <div class="row">
-            <select id="voice" style="flex:2"></select>
-            <div class="speed-control">
-                <label>🎚️ Vitesse : <span id="speedValue">1.0</span></label>
-                <input type="range" id="speed" min="0.5" max="1.5" step="0.05" value="1.0">
-            </div>
-        </div>
-
-        <button id="generateBtn">🔊 Générer & lire (streaming)</button>
-        <audio id="audioPlayer" controls style="display: none;"></audio>
-        <div id="ttsStatus" class="status"></div>
-    </div>
-
-    <!-- Section Clonage -->
-    <div class="card">
-        <h2>🎙️ 2. Cloner une nouvelle voix</h2>
-        <div class="file-upload" id="uploadArea">
-            <input type="file" id="audioFile" accept=".wav,.mp3,.m4a,.ogg">
-            <p>📁 Cliquez ou glissez un fichier audio (WAV, MP3, M4A, OGG)</p>
-            <p style="font-size: 12px;">5-15 secondes, voix claire, peu de bruit</p>
-        </div>
-        <input type="text" id="newVoiceName" placeholder="Nom de la nouvelle voix (ex: ma_voix)">
-        <button id="cloneBtn">✨ Cloner cette voix</button>
-        <div id="cloneStatus" class="status"></div>
-    </div>
-
-    <!-- Section Liste des voix -->
-    <div class="card">
-        <h2>🎧 3. Voix disponibles</h2>
-        <div id="voiceList" class="voice-list">
-            <p>Chargement...</p>
-        </div>
-        <button id="refreshVoicesBtn" style="background: #6c757d;">🔄 Actualiser</button>
-    </div>
-</div>
-
-<script>
-    // Éléments
-    const textarea = document.getElementById('text');
-    const voiceSelect = document.getElementById('voice');
-    const speedSlider = document.getElementById('speed');
-    const speedValue = document.getElementById('speedValue');
-    const generateBtn = document.getElementById('generateBtn');
-    const audioPlayer = document.getElementById('audioPlayer');
-    const ttsStatus = document.getElementById('ttsStatus');
-    const cloneBtn = document.getElementById('cloneBtn');
-    const cloneStatus = document.getElementById('cloneStatus');
-    const refreshBtn = document.getElementById('refreshVoicesBtn');
-    const voiceListDiv = document.getElementById('voiceList');
-    const fileInput = document.getElementById('audioFile');
-    const uploadArea = document.getElementById('uploadArea');
-    const newVoiceNameInput = document.getElementById('newVoiceName');
-
-    let mediaSource = null;
-    let sourceBuffer = null;
-    let audioQueue = [];
-    let isUpdating = false;
-
-    // Met à jour l'affichage de la vitesse
-    speedSlider.addEventListener('input', () => {
-        speedValue.textContent = speedSlider.value;
-    });
-
-    // Charger les voix
-    async function loadVoices() {
-        try {
-            const response = await fetch('/voices');
-            if (response.ok) {
-                const data = await response.json();
-                const voices = data.voices;
-                // Mettre à jour le select
-                voiceSelect.innerHTML = '';
-                voices.forEach(v => {
-                    const option = document.createElement('option');
-                    option.value = v;
-                    option.textContent = v;
-                    voiceSelect.appendChild(option);
-                });
-                // Mettre à jour les tags
-                voiceListDiv.innerHTML = '';
-                voices.forEach(v => {
-                    const tag = document.createElement('div');
-                    tag.className = 'voice-tag';
-                    tag.textContent = v;
-                    tag.onclick = () => {
-                        voiceSelect.value = v;
-                    };
-                    voiceListDiv.appendChild(tag);
+    <h1>🗣️ Kokoro TTS (ultra rapide)</h1>
+    <textarea id="text" rows="4">Bonjour, ceci est un test de Kokoro. La synthèse est très rapide et naturelle.</textarea>
+    <select id="voice"></select>
+    <select id="lang">
+        <option value="fr">Français</option>
+        <option value="en">Anglais</option>
+    </select>
+    <input type="range" id="speed" min="0.5" max="1.5" step="0.05" value="1.0">
+    <span>Vitesse: <span id="speedVal">1.0</span></span>
+    <button id="generate">🎤 Générer</button>
+    <audio id="audio" controls></audio>
+    <div id="status" class="status"></div>
+
+    <script>
+        const voiceSelect = document.getElementById('voice');
+        const textarea = document.getElementById('text');
+        const langSelect = document.getElementById('lang');
+        const speedSlider = document.getElementById('speed');
+        const speedVal = document.getElementById('speedVal');
+        const generateBtn = document.getElementById('generate');
+        const audioPlayer = document.getElementById('audio');
+        const statusDiv = document.getElementById('status');
+
+        speedSlider.oninput = () => speedVal.innerText = speedSlider.value;
+
+        async function loadVoices() {
+            const res = await fetch('/voices');
+            const data = await res.json();
+            const allVoices = [...data.builtin, ...data.cloned];
+            voiceSelect.innerHTML = allVoices.map(v => `<option value="${v}">${v}</option>`).join('');
+        }
+
+        generateBtn.onclick = async () => {
+            const text = textarea.value.trim();
+            const voice = voiceSelect.value;
+            const lang = langSelect.value;
+            const speed = parseFloat(speedSlider.value);
+            if (!text) return;
+
+            statusDiv.innerText = '⏳ Génération en cours...';
+            try {
+                const response = await fetch('/tts', {
+                    method: 'POST',
+                    headers: { 'Content-Type': 'application/json' },
+                    body: JSON.stringify({ text, voice, speed, lang })
                 });
-                if (voices.length === 0) voiceListDiv.innerHTML = '<p>Aucune voix.</p>';
-            } else {
-                throw new Error('Erreur chargement');
-            }
-        } catch (error) {
-            console.error(error);
-            voiceListDiv.innerHTML = '<p>Erreur chargement voix</p>';
-        }
-    }
-
-    // Génération avec streaming réel (MediaSource)
-    async function generateAndPlayStreaming() {
-        const text = textarea.value.trim();
-        const voiceName = voiceSelect.value;
-        const speed = parseFloat(speedSlider.value);
-
-        if (!text) {
-            ttsStatus.className = 'status error';
-            ttsStatus.textContent = 'Veuillez saisir du texte.';
-            return;
-        }
-        if (!voiceName) {
-            ttsStatus.className = 'status error';
-            ttsStatus.textContent = 'Veuillez sélectionner une voix.';
-            return;
-        }
-
-        ttsStatus.className = 'status info';
-        ttsStatus.textContent = '⏳ Génération en cours (streaming)...';
-        audioPlayer.style.display = 'none';
-
-        try {
-            const response = await fetch('/tts', {
-                method: 'POST',
-                headers: { 'Content-Type': 'application/json' },
-                body: JSON.stringify({ text, voice_name: voiceName, speed })
-            });
-
-            if (!response.ok) {
-                const errorText = await response.text();
-                throw new Error(errorText);
-            }
-
-            // Récupérer le blob complet (le streaming côté client se fait via MediaSource, mais on peut utiliser blob pour simplifier)
-            // Pour un vrai streaming avec lecture progressive, on utilise MediaSource.
-            // Ici, on va simplement récupérer le blob et le lire (la réponse est déjà en chunked, mais le navigateur attend la fin).
-            // Pour démonstration, on garde cette méthode simple.
-            const blob = await response.blob();
-            const url = URL.createObjectURL(blob);
-            audioPlayer.src = url;
-            audioPlayer.style.display = 'block';
-            audioPlayer.play();
-            ttsStatus.className = 'status success';
-            ttsStatus.textContent = '✅ Audio généré et en cours de lecture.';
-        } catch (error) {
-            ttsStatus.className = 'status error';
-            ttsStatus.textContent = `❌ Erreur : ${error.message}`;
-        }
-    }
-
-    // Version alternative avec MediaSource (streaming plus fin)
-    // On garde la version simple blob car elle fonctionne très bien avec le streaming HTTP chunked
-    // le navigateur commence à jouer dès que les premiers paquets sont reçus si le format le permet (non pour WAV ?)
-    // En réalité WAV nécessite l'en-tête complet. Donc blob est suffisant.
-
-    // Clonage de voix
-    cloneBtn.onclick = async () => {
-        if (!fileInput.files || fileInput.files.length === 0) {
-            cloneStatus.className = 'status error';
-            cloneStatus.textContent = 'Veuillez sélectionner un fichier audio.';
-            return;
-        }
-        const voiceName = newVoiceNameInput.value.trim();
-        if (!voiceName) {
-            cloneStatus.className = 'status error';
-            cloneStatus.textContent = 'Veuillez donner un nom à la voix.';
-            return;
-        }
-        if (!/^[a-zA-Z0-9_-]+$/.test(voiceName)) {
-            cloneStatus.className = 'status error';
-            cloneStatus.textContent = 'Nom invalide (lettres, chiffres, tirets, underscores).';
-            return;
-        }
-
-        const formData = new FormData();
-        formData.append('audio_file', fileInput.files[0]);
-        formData.append('voice_name', voiceName);
-
-        cloneStatus.className = 'status info';
-        cloneStatus.textContent = '⏳ Clonage en cours (cela peut prendre quelques secondes)...';
-
-        try {
-            const response = await fetch('/clone-voice', {
-                method: 'POST',
-                body: formData
-            });
-            if (response.ok) {
-                const result = await response.json();
-                cloneStatus.className = 'status success';
-                cloneStatus.textContent = `✅ Voix "${voiceName}" clonée avec succès ! (Total: ${result.total_voices} voix)`;
-                fileInput.value = '';
-                newVoiceNameInput.value = '';
-                await loadVoices();
-            } else {
-                const error = await response.text();
-                throw new Error(error);
+                if (response.ok) {
+                    const blob = await response.blob();
+                    const url = URL.createObjectURL(blob);
+                    audioPlayer.src = url;
+                    audioPlayer.play();
+                    statusDiv.innerText = '✅ Audio prêt';
+                } else {
+                    const err = await response.text();
+                    statusDiv.innerText = `❌ Erreur: ${err}`;
+                }
+            } catch(e) {
+                statusDiv.innerText = `❌ ${e.message}`;
             }
-        } catch (error) {
-            cloneStatus.className = 'status error';
-            cloneStatus.textContent = `❌ Erreur : ${error.message}`;
-        }
-    };
-
-    // Rafraîchir les voix
-    refreshBtn.onclick = loadVoices;
-
-    // Drag & drop pour upload
-    uploadArea.addEventListener('click', () => fileInput.click());
-    uploadArea.addEventListener('dragover', (e) => {
-        e.preventDefault();
-        uploadArea.style.borderColor = '#fff';
-        uploadArea.style.background = 'rgba(0,168,255,0.2)';
-    });
-    uploadArea.addEventListener('dragleave', () => {
-        uploadArea.style.borderColor = '#00a8ff';
-        uploadArea.style.background = 'rgba(0,168,255,0.1)';
-    });
-    uploadArea.addEventListener('drop', (e) => {
-        e.preventDefault();
-        uploadArea.style.borderColor = '#00a8ff';
-        uploadArea.style.background = 'rgba(0,168,255,0.1)';
-        const files = e.dataTransfer.files;
-        if (files.length > 0 && files[0].type.includes('audio')) {
-            fileInput.files = files;
-            cloneStatus.className = 'status info';
-            cloneStatus.textContent = `Fichier sélectionné : ${files[0].name}`;
-        } else {
-            alert('Déposez un fichier audio valide.');
-        }
-    });
-
-    // Génération au clic
-    generateBtn.onclick = generateAndPlayStreaming;
+        };
 
-    // Initialisation
-    loadVoices();
-</script>
+        loadVoices();
+    </script>
 </body>
 </html>