import io import time import requests from fastapi import Request, Response, status from fastapi.responses import JSONResponse from injector import inject from injector import singleton from ..log.logging_time import logging_time from ..configuration import MeloConf from .blackbox import Blackbox import soundfile from melo.api import TTS import logging logger = logging.getLogger(__name__) @singleton class MeloTTS(Blackbox): mode: str url: str speed: int device: str language: str speaker: str @logging_time(logger=logger) def model_init(self, melo_config: MeloConf) -> None: self.speed = melo_config.speed self.device = melo_config.device self.language = melo_config.language self.speaker = melo_config.speaker self.device = melo_config.device self.url = '' self.mode = melo_config.mode self.melotts = None self.speaker_ids = None if self.mode == 'local': self.melotts = TTS(language=self.language, device=self.device) self.speaker_ids = self.melotts.hps.data.spk2id else: self.url = melo_config.url logging.info('#### Initializing MeloTTS Service in ' + self.device + ' mode...') @inject def __init__(self, melo_config: MeloConf) -> None: self.model_init(melo_config) def __call__(self, *args, **kwargs): return self.processing(*args, **kwargs) def valid(self, *args, **kwargs) -> bool: text = args[0] return isinstance(text, str) @logging_time(logger=logger) def processing(self, *args, **kwargs) -> io.BytesIO | bytes: text = args[0] current_time = time.time() if self.mode == 'local': audio = self.melotts.tts_to_file(text, self.speaker_ids[self.speaker], speed=self.speed) f = io.BytesIO() soundfile.write(f, audio, 44100, format='wav') f.seek(0) print("#### MeloTTS Service consume - local : ", (time.time() - current_time)) return f.read() else: message = { "text": text } response = requests.post(self.url, json=message) print("#### MeloTTS Service consume - docker : ", (time.time()-current_time)) return response.content async def fast_api_handler(self, request: Request) -> Response: try: data = await request.json() except: return JSONResponse(content={"error": "json parse error"}, status_code=status.HTTP_400_BAD_REQUEST) text = data.get("text") if text is None: return JSONResponse(content={"error": "text is required"}, status_code=status.HTTP_400_BAD_REQUEST) return Response(content=self.processing(text), media_type="audio/wav", headers={"Content-Disposition": "attachment; filename=audio.wav"})