| from TTS.api import TTS |
| import torch |
| from typing import Optional |
|
|
| class TTSService: |
| def __init__(self): |
| self.device = "cuda" if torch.cuda.is_available() else "cpu" |
| self.model = self._load_model() |
|
|
| def _load_model(self): |
| """Ładowanie modelu XTTS v2""" |
| return TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(self.device) |
|
|
| def synthesize( |
| self, |
| text: str, |
| reference_wav: str, |
| output_path: str, |
| language: str = "en" |
| ) -> None: |
| """Generacja pliku audio z tekstu""" |
| self.model.tts_to_file( |
| text=text, |
| speaker_wav=reference_wav, |
| language=language, |
| file_path=output_path |
| ) |
|
|
| |
| XTTS = TTSService() |
