|
import torch |
|
from transformers import SpeechT5ForTextToSpeech, SpeechT5Processor |
|
import logging |
|
import numpy as np |
|
import soundfile as sf |
|
|
|
|
|
logging.basicConfig(level=logging.DEBUG) |
|
|
|
MODEL_ID = "microsoft/speecht5_tts" |
|
|
|
|
|
try: |
|
processor = SpeechT5Processor.from_pretrained(MODEL_ID) |
|
model = SpeechT5ForTextToSpeech.from_pretrained(MODEL_ID) |
|
logging.info("Model and processor loaded successfully.") |
|
except Exception as e: |
|
logging.error(f"Error loading model or processor: {e}") |
|
|
|
def synthesize_speech(text): |
|
try: |
|
inputs = processor(text, return_tensors="pt") |
|
device = torch.device("cuda" if torch.cuda.is_available() else "cpu") |
|
model.to(device) |
|
inputs = inputs.to(device) |
|
|
|
with torch.no_grad(): |
|
speech = model.generate(**inputs) |
|
|
|
logging.info("Speech generated successfully.") |
|
|
|
|
|
waveform = speech.cpu().numpy() |
|
sf.write("output.wav", waveform, 16000) |
|
return "output.wav" |
|
except Exception as e: |
|
logging.error(f"Error during speech synthesis: {e}") |
|
return None |
|
|