Finally got it work on my Macbook Pro M1!!!

[jeantimex]

realtime-chat.mp4

Will share more details. Btw, this is a super cool project! Claude Code is great :)

[KoljaB]

Wow, that is really impressive. Thanks for sharing.

[jeantimex]

Here are my changes in the forked repo: https://github.com/jeantimex/RealtimeVoiceChat/tree/mac-support.

And I wrote down the details of changes for supporting Mac ARM64: https://github.com/jeantimex/RealtimeVoiceChat/blob/mac-support/MAC_SUPPORT.md.

The core issues are:

1. Coqui TTS engine depends on deepspeed library which is not supported on Mac ARM64.
2. The pvporcupine wake word detection engine is not supported on Mac ARM64.

[KoljaB]

Thanks a lot, great work!

[that-one-arab]

@KoljaB @jeantimex Why not open a PR and merge your changes?
This would be a very important contribution for mac users

[K-Schubert]

Amazing work thanks! PR would be nice.

[RishiBodha]

Yes, great work.. any updates? I am going to try and install this within the next day or so

[DrBrule]

Very helpful changes here!

[scalar27]

Thanks. Really nice. This fork works well on my M1 Mac. However, I can't get audio when I switch to orpheus. If anyone has figured this out please let me (us) know what changes to make. Additionally (for Kokoro or Orpheus) I'd like to serve the llm and tts directly with llama-server. Figuring out what changes are needed seems beyond my capabilities...