Deploy TTS service with failover support

app.py

@@ -0,0 +1,188 @@
+from fastapi import FastAPI, HTTPException, Form, BackgroundTasks
+from fastapi.responses import FileResponse
+from kokoro_onnx import Kokoro
+import tempfile
+import os
+from datetime import datetime
+import soundfile as sf
+
+# ============== CONFIG ==============
+MAX_CHARS = 4500  # ~5 minutes of audio (speaking rate: ~900 chars/min)
+MIN_CHARS = 5
+MAX_AUDIO_DURATION = 300  # 5 minutes of audio
+
+# ============== KOKORO TTS MODEL ==============
+print("🎤 Loading Kokoro TTS model...")
+try:
+    kokoro = Kokoro("kokoro-v0_19.onnx", "voices")
+    print("✅ Kokoro TTS loaded successfully!")
+except Exception as e:
+    print(f"⚠️  Kokoro not found locally. Will download on first use.")
+    kokoro = None
+
+app = FastAPI(
+    title="Kokoro TTS API - Fast & Simple",
+    description="High-speed text-to-speech with emotional voices",
+    version="2.0"
+)
+
+@app.on_event("startup")
+def startup():
+    global kokoro
+    if kokoro is None:
+        import urllib.request
+        
+        print("📥 Downloading Kokoro TTS model files...")
+        
+        # Create directory for voices
+        os.makedirs("voices", exist_ok=True)
+        
+        # Download voices file
+        voices_file = "voices/voices.bin"
+        if not os.path.exists(voices_file):
+            print("Downloading voices.bin...")
+            urllib.request.urlretrieve(
+                "https://github.com/thewh1teagle/kokoro-onnx/releases/download/model-files/voices.bin",
+                voices_file
+            )
+            print("✅ Voices downloaded!")
+        
+        # Download ONNX model
+        model_file = "kokoro-v0_19.onnx"
+        if not os.path.exists(model_file):
+            print("Downloading kokoro-v0_19.onnx...")
+            urllib.request.urlretrieve(
+                "https://github.com/thewh1teagle/kokoro-onnx/releases/download/model-files/kokoro-v0_19.onnx",
+                model_file
+            )
+            print("✅ Model downloaded!")
+        
+        print("🎤 Initializing Kokoro TTS...")
+        kokoro = Kokoro(model_file, voices_file)
+        print("✅ Kokoro TTS loaded!")
+
+# ============== HELPERS ==============
+def cleanup_file(path: str):
+    """Delete temporary file after response is sent"""
+    try:
+        if os.path.exists(path):
+            os.unlink(path)
+    except:
+        pass
+
+def generate_speech(text: str, voice: str = "bf_isabella", speed: float = 1.0) -> str:
+    """
+    Generate speech using Kokoro TTS
+    Available voices: af_heart, af_bella, am_adam, am_michael, bf_emma, bf_isabella
+    """
+    if len(text) < MIN_CHARS:
+        raise ValueError(f"Text too short. Minimum {MIN_CHARS} characters.")
+    if len(text) > MAX_CHARS:
+        raise ValueError(f"Text too long. Maximum {MAX_CHARS} characters (~5 min audio).")
+    
+    # Generate audio samples
+    samples, sample_rate = kokoro.create(
+        text=text,
+        voice=voice,
+        speed=speed,
+        lang="en-us"
+    )
+    
+    # Save to temporary file
+    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
+        sf.write(tmp.name, samples, sample_rate)
+        return tmp.name
+
+# ============== API ENDPOINTS ==============
+@app.get("/")
+def root():
+    return {
+        "service": "Kokoro TTS API",
+        "status": "running",
+        "model": "Kokoro-82M",
+        "version": "2.0",
+        "features": {
+            "speed": "10x faster than XTTS",
+            "voices": 6,
+            "max_chars": MAX_CHARS,
+            "emotional": True
+        },
+        "endpoints": {
+            "health": "/health",
+            "generate": "/api/generate (POST)",
+            "docs": "/docs"
+        }
+    }
+
+@app.get("/health")
+def health():
+    return {
+        "status": "healthy",
+        "model": "Kokoro TTS 82M",
+        "speed": "10x faster than XTTS",
+        "max_chars": MAX_CHARS,
+        "voices": ["af_heart", "af_bella", "am_adam", "am_michael", "bf_emma", "bf_isabella"]
+    }
+
+@app.post("/api/generate")
+async def generate_tts(
+    background_tasks: BackgroundTasks,
+    text: str = Form(..., description="Text to convert to speech"),
+    voice: str = Form("bf_isabella", description="Voice to use"),
+    speed: float = Form(1.0, description="Speech speed (0.5-2.0)")
+):
+    """
+    Generate TTS with Kokoro (Fast & Emotional)
+    
+    **Performance:**
+    - Max audio: 5 minutes (4500 chars)
+    - Generation: ~20-30 seconds on CPU
+    - Speech rate: ~900 chars/minute
+    
+    **Available Voices:**
+    - `af_heart`: American Female (warm)
+    - `af_bella`: American Female (professional)
+    - `am_adam`: American Male (confident)
+    - `am_michael`: American Male (friendly)
+    - `bf_emma`: British Female (elegant)
+    - `bf_isabella`: British Female (storytelling) ⭐ Best for long content
+    
+    **Example:**
+    ```bash
+    curl -X POST https://your-space.hf.space/api/generate \\
+      -F "text=Hello world, this is Kokoro TTS!" \\
+      -F "voice=bf_isabella" \\
+      -F "speed=1.0" \\
+      --output audio.wav
+    ```
+    """
+    try:
+        # Validate speed
+        if speed < 0.5 or speed > 2.0:
+            raise HTTPException(status_code=400, detail="Speed must be between 0.5 and 2.0")
+        
+        # Generate speech
+        output_path = generate_speech(text.strip(), voice, speed)
+        
+        # Schedule cleanup after response is sent
+        background_tasks.add_task(cleanup_file, output_path)
+        
+        # Return audio file
+        response = FileResponse(
+            output_path, 
+            media_type="audio/wav", 
+            filename=f"kokoro_{datetime.now().strftime('%Y%m%d_%H%M%S')}.wav"
+        )
+        response.headers["X-Character-Count"] = str(len(text))
+        response.headers["X-Voice-Used"] = voice
+        
+        return response
+        
+    except ValueError as e:
+        raise HTTPException(status_code=400, detail=str(e))
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"TTS generation failed: {str(e)}")
+
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=7860)

packages.txt

@@ -0,0 +1 @@
+libsndfile1

requirements.txt

@@ -0,0 +1,6 @@
+fastapi==0.115.0
+uvicorn[standard]==0.32.0
+kokoro-onnx
+soundfile
+python-multipart
+numpy