app.py

import io
import sys
import os
import numpy as np
import soundfile as sf
import sherpa_onnx
from fastapi import FastAPI
from fastapi.responses import StreamingResponse
import uvicorn

ljspeech_config = sherpa_onnx.OfflineTtsConfig(model=sherpa_onnx.OfflineTtsModelConfig(vits=sherpa_onnx.OfflineTtsVitsModelConfig(model='model.onnx', lexicon='', data_dir='espeak-ng-data', tokens='tokens.txt'), num_threads=4))
if not ljspeech_config.validate():
	raise ValueError("Please check your config")

ljspeech = sherpa_onnx.OfflineTts(ljspeech_config)
app = FastAPI()

@app.get("/tts", response_class=StreamingResponse)
async def do_tts(text: str):
	audio = ljspeech.generate(text)
	f = io.BytesIO()
	sf.write(f, audio.samples, audio.sample_rate, 'PCM_24', format='WAV')
	f.seek(0)
	return StreamingResponse(f, media_type='audio/wav')

if __name__ == '__main__':
	uvicorn.run(app, host='0.0.0.0', port=7860)