Open IguteChung opened 7 months ago
In /tts_to_audio api, if USE_CACHE=False, it will use default file_name_or_path="out.wav" for XTTS.process_tts_to_file.
/tts_to_audio
file_name_or_path="out.wav"
XTTS.process_tts_to_file
However it get error if concurrent api calls happen.
torchaudio.save(output_file, torch.tensor(out["wav"]).unsqueeze(0), 24000)
I also get an error when processing 2 or more requests at the same time, I'm trying to fix it.
Is there a fix for this?
In
/tts_to_audio
api, if USE_CACHE=False, it will use defaultfile_name_or_path="out.wav"
forXTTS.process_tts_to_file
.However it get error if concurrent api calls happen.
torchaudio.save(output_file, torch.tensor(out["wav"]).unsqueeze(0), 24000)