Chi Kim on Nostr: 😂 Kyutai released moshi v0.1, a fully native speech to speech chat model that ...
😂 Kyutai released moshi v0.1, a fully native speech to speech chat model that doesn't require separate ASR and TTS model. It's so dumb but kinda fun for toy. 🤣 Hopefully Meta will expand Llama to be like this.
Here's a simple way to try it out on Apple Silicon. You need to relaunch when you get an error once in a while, and you could try -q 8 If you have more than 16gb memory.
pip install moshi_mlx; python -m moshi_mlx.local_web -q 4
#Multimodal #LLM #AI #ML
Here's a simple way to try it out on Apple Silicon. You need to relaunch when you get an error once in a while, and you could try -q 8 If you have more than 16gb memory.
pip install moshi_mlx; python -m moshi_mlx.local_web -q 4
#Multimodal #LLM #AI #ML