It doesn't talk back just texts the answers... on mac os.

[johnkea]

It doesn't talk back just texts the answers... on mac os.

[eustlb]

Hey there,
you're likely running Parler-TTS on mps which is why you feel like it's not answering (way to slow...). Have you tried running with the optimal mac settings (see here) ?

Otherwise can you please provide your mac setup and the command you're running ?

[johnkea]

Hey, thanks for reply, what I meant is sound response, is the generated text supposed to be converted to audio and played then? So, I don't hear any sound there...

[andimarafioti]

It would be great to have more information to reproduce this 😅

[SutirthaChakraborty]

To solve run:

pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cpu
pip install Accelerate

python s2s_pipeline.py --device mps --stt whisper --stt_model_name turbo --tts melo --llm mlx-lm --local_mac_optimal_settings