Kyutai STT (https://kyutai.org/next/stt) is Kyutai’s “speech to text” model, available in two versions open-source:

• STT-1B-en_fr (Documentation: https://huggingface.co/kyutai/stt-1b-en_fr): , which includes English and French with a voice activity detector,
• stt-2.6b-en (documentation: https://huggingface.co/kyutai/stt-2.6b-en), optimized for English and to be as accurate as possible.

These templates are used for Unmute.

Free and unlimited. Demonstration of the stt-1b-en_fr model directly on the site.

Test with the conclusion of the book “Digital civilization, let’s open the debate!” by the National Digital Council: