tts_mockingbird / src /convert.py
khof312's picture
Load initial demo.
f5cf172
raw
history blame
682 Bytes
import torch
import IPython
from TTS.api import TTS
def convert_coqui(source_wav:str, target_wav:str):
'''
Use Coqui TTS for zero-shot voice conversion.
Inputs:
source_wav: Wav of the thing you want to say.
target_wav: Wav of the speaker you want to hear.
Returns:
Streaming wav and sampling rate.
'''
# Get device
device = "cuda" if torch.cuda.is_available() else "cpu"
tts = TTS(model_name="voice_conversion_models/multilingual/vctk/freevc24", progress_bar=False).to(device)
wav = tts.voice_conversion(source_wav=source_wav, target_wav=target_wav)
return wav, 24000 # Identified sampling rate of freevc24