Chatterbox doesn't work on Apple silicon due to the lack of CUDA.

[johnkea]

https://github.com/resemble-ai/chatterbox/blob/bf169fe5f518760cb0b6c6a6eba3f885e10fa86f/src/chatterbox/mtl_tts.py#L179C13-L179C65

Simple add one more param at 179 line, and make the following code:

torch.load(ckpt_dir / "s3gen.pt", weights_only=True)

be like this one

torch.load(ckpt_dir / "s3gen.pt", map_location=device, weights_only=True)

that will save one's time in the future.

[ebertti]

+1