Switch to Piper TTS - lightweight, fast, multilingual

Dockerfile

@@ -5,13 +5,27 @@ WORKDIR /app
 # Install system dependencies
 RUN apt-get update && apt-get install -y \
     libsndfile1 \
-    ffmpeg \
+    espeak-ng \
+    wget \
     && rm -rf /var/lib/apt/lists/*
 
 # Install Python dependencies
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
+# Pre-download voice models at build time (directly to /app)
+RUN cd /app && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/zh/zh_CN/huayan/medium/zh_CN-huayan-medium.onnx && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/zh/zh_CN/huayan/medium/zh_CN-huayan-medium.onnx.json && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/medium/en_US-lessac-medium.onnx && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/medium/en_US-lessac-medium.onnx.json && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/ms/ms_MY/sirius/medium/ms_MY-sirius-medium.onnx && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/ms/ms_MY/sirius/medium/ms_MY-sirius-medium.onnx.json && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/ja/ja_JP/kenji/medium/ja_JP-kenji-medium.onnx && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/ja/ja_JP/kenji/medium/ja_JP-kenji-medium.onnx.json && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/ko/ko_KR/hui_hao/medium/ko_KR-hui-hao-medium.onnx && \
+    wget -q https://huggingface.co/rhasspy/piper-voices/resolve/main/ko/ko_KR/hui_hao/medium/ko_KR-hui-hao-medium.onnx.json
+
 # Copy application code
 COPY main.py .
 

main.py

@@ -1,84 +1,70 @@
 from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
-from matcha_tts import MatchaTTS
+from piper import PiperVoice
 import soundfile as sf
-import torch
-import numpy as np
 import os
 import uuid
 from fastapi.responses import Response
 from io import BytesIO
 
-app = FastAPI(title="Matcha TTS Space")
+app = FastAPI(title="Piper TTS Space")
 
-# Load model on startup
-MODEL_NAME = os.getenv("MATCHA_MODEL", "matcha-tts/matcha-v1.0")
-print(f"Loading Matcha-TTS model: {MODEL_NAME}")
-try:
-    model = MatchaTTS.from_pretrained(MODEL_NAME)
-    print("Model loaded successfully")
-except Exception as e:
-    print(f"Failed to load model: {e}")
-    model = None
+# Load voices on startup (download if needed)
+VOICE_MODELS = {
+    "zh": "zh_CN-huayan-medium",      # Chinese
+    "en": "en_US-lessac-medium",      # English
+    "ms": "ms_MY-sirius-medium",      # Malay
+    "ja": "ja_JP-kenji-medium",       # Japanese
+    "ko": "ko_KR-hui-hao-medium",     # Korean
+}
+
+VOICES = {}
+
+print("Loading Piper voices...")
+for lang, model_name in VOICE_MODELS.items():
+    try:
+        voice = PiperVoice.load(f"{model_name}.onnx")
+        VOICES[lang] = voice
+        print(f"✓ Loaded {lang}: {model_name}")
+    except Exception as e:
+        print(f"✗ Failed to load {lang}: {e}")
+
+print(f"Loaded {len(VOICES)} voices: {list(VOICES.keys())}")
 
 class TTSRequest(BaseModel):
     text: str
-    language: str = "en"  # ISO 639-1 code
+    lang: str = "zh"  # zh, en, ms, ja, ko
     speaker_id: int = 0
-    noise_scale: float = 0.667
-    noise_scale_w: float = 0.8
-    length_scale: float = 1.0
+    sample_rate: int = 22050
 
 @app.post("/tts")
 async def tts(request: TTSRequest):
-    if model is None:
-        raise HTTPException(status_code=503, detail="Model not loaded")
+    voice = VOICES.get(request.lang)
+    if not voice:
+        raise HTTPException(400, f"Language '{request.lang}' not available. Available: {list(VOICES.keys())}")
     
     try:
-        # Synthesize audio
-        audio = model.synthesise(
-            request.text,
-            speaker_id=request.speaker_id,
-            noise_scale=request.noise_scale,
-            noise_scale_w=request.noise_scale_w,
-            length_scale=request.length_scale,
-        )
-        
-        # Convert to numpy array if needed
-        if torch.is_tensor(audio):
-            audio = audio.cpu().numpy()
-        
-        # Ensure it's 1D array
-        if audio.ndim > 1:
-            audio = audio.squeeze()
-        
-        # Create WAV in memory
+        audio = voice.synthesize(request.text, speaker_id=request.speaker_id)
         buffer = BytesIO()
-        sf.write(buffer, audio, 24000, format='WAV')  # Matcha outputs 24kHz
+        sf.write(buffer, audio, request.sample_rate, format='WAV')
         buffer.seek(0)
-        
         return Response(
             content=buffer.read(),
             media_type="audio/wav",
-            headers={
-                "Content-Disposition": f"attachment; filename=tts_{uuid.uuid4().hex}.wav"
-            }
+            headers={"Content-Disposition": f"attachment; filename=tts_{uuid.uuid4().hex}.wav"}
         )
     except Exception as e:
-        raise HTTPException(status_code=500, detail=f"Synthesis failed: {str(e)}")
+        raise HTTPException(500, f"Synthesis failed: {str(e)}")
 
 @app.get("/health")
 async def health():
-    return {"status": "healthy" if model is not None else "model_not_loaded", "model": MODEL_NAME}
+    return {"status": "healthy", "voices": list(VOICES.keys())}
 
 @app.get("/")
 async def root():
     return {
-        "message": "Matcha TTS API",
+        "message": "Piper TTS API - Fast multilingual TTS",
+        "voices": list(VOICES.keys()),
         "endpoint": "/tts (POST)",
-        "model": MODEL_NAME,
-        "example": {
-            "text": "Hello, this is a test of the Matcha TTS system.",
-            "language": "en"
-        }
+        "example": {"text": "你好，这是 Piper TTS 测试", "lang": "zh"}
     }

requirements.txt

@@ -1,6 +1,5 @@
 fastapi==0.111.0
 uvicorn[standard]==0.30.1
-matcha-tts==0.0.7.2
-onnxruntime==1.19.2
+piper-tts==1.2.0
 soundfile==0.12.1
 pydantic==2.8.2