app.py

import tempfile
import soundfile as sf
from fastapi import FastAPI, HTTPException
from fastapi.responses import FileResponse
from pydantic import BaseModel
from neuttsair.neutts import NeuTTSAir

# Initialize FastAPI app
app = FastAPI(title="NeuTTS-Air API", description="A FastAPI service for the NeuTTS-Air model.")

# Load the NeuTTS-Air model
# The path is relative to the working directory in the Docker container
MODEL_PATH = "neutts-air-q4-gguf"
try:
    tts = NeuTTSAir(backbone_repo=MODEL_PATH, backbone_device="cpu")
except Exception as e:
    print(f"Error loading model: {e}")
    tts = None

# Pydantic model for the request body
class TTSRequest(BaseModel):
    text: str
    ref_audio_path: str
    ref_text: str

@app.get("/")
def read_root():
    """Simple health check endpoint."""
    return {"message": "NeuTTS-Air FastAPI is running."}

@app.post("/tts", summary="Generate speech from text")
async def tts_endpoint(request: TTSRequest):
    """
    Generates a WAV audio file from text using a reference audio and transcript.
    """
    if tts is None:
        raise HTTPException(status_code=503, detail="Model is not loaded.")

    try:
        # Load the reference audio
        # Note: You must provide a valid path to an audio file
        # The user will need to upload their own reference audios or use pre-uploaded ones
        ref_codes = tts.encode_reference(request.ref_audio_path)
        
        # Perform inference
        wav_audio = tts.infer(request.text, ref_codes, request.ref_text)
        
        # Save the audio to a temporary file
        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
            sf.write(tmp.name, wav_audio, tts.codec.sampling_rate)
            filepath = tmp.name

        # Return the audio file
        return FileResponse(filepath, media_type="audio/wav", filename="generated_speech.wav")

    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Internal Server Error: {e}")