kyutai-labs/moshi
View on GitHub ↗ ★ 11,456 Python
Moshi is a speech-text foundation model and full-duplex spoken dialogue framework. It uses Mimi, a state-of-the-art streaming neural audio codec.
7 mentions on Hacker News by 7 people: 1 story, 6 comments · first 2024-09-18 · latest 2026-03-03
20082026
-
Thanks. Then maybe it’s similar to Moshi https://github.com/kyutai-labs/moshi?tab=readme-ov-file
comment by totetsu · 2026-03-03 · in Show HN: I built a sub-500ms latency voice agent from scratch -
You mean Moshi ( https://github.com/kyutai-labs/moshi )? Since Personaplex is just a finetuned Moshi model.
comment by woodson · 2026-03-02 · in Show HN: I built a sub-500ms latency voice agent from scratch -
…kyutai Moshi, maybe old now but has a MLX implementation of their speech to speech model: https://github.com/kyutai-labs/moshi
comment by doonielk · 2026-01-24 · in Ask HN: What's the current best local/open speech-to-speech setup? -
…model built on Llama-3.1-8B-Instruct for simultaneous generation of text and speech - moshi https://github.com/kyutai-labs/moshi speech-text foundation model using Mimi, a SOTA streaming neural audio codec - Mini-Omni…
comment by underlines · 2024-11-04 · in Hertz-dev, the first open-source base model for conversational audio -
…audio model from earlier this year that seems to perform similarly (and it runs on a MacBook): https://github.com/kyutai-labs/moshi
comment by blixt · 2024-11-04 · in Hertz-dev, the first open-source base model for conversational audio -
Yes. Kyutai released an opened model called moshi : https://github.com/kyutai-labs/moshi There's also llama-omni and a few others. None of them are even close to 4o from an LLM standpoint. But moshi is called a "foundational" model and U'm…
comment by Ey7NFZ3P0nzAe · 2024-10-05 · in Show HN: Open source framework OpenAI uses for Advanced Voice