HuggingFace and Cerebras Open-Source Real-Time Speech-to-Speech Demo

Thom_Wolf · x · 2026-07-03

HuggingFace co-founder Thom Wolf collaborated with Cerebras to launch a fully open-source real-time speech-to-speech demo, including both the model and code. Showcasing the current capabilities of open-source voice technology, they provided the demo and a blog post to encourage users to test, fork, and modify it. The accompanying video is the first recording, entirely unedited and played at actual speed.

Original post →

More from Multimodal

Multimodal channel →