KyutaiPocket TTS

Pocket TTS English 2026-09 revision 2

Pocket TTS English 2026-09 revision 2 is a downloadable text-to-speech checkpoint in the Pocket TTS family. Its official configuration identifies the trained weights and the components needed to generate english speech.

The configuration identifies this release

The configuration documents distillation on re-aligned Dora data instead of HiFi-TTS-2, using moving-average weights. This is a distinct revision of the September recipe.Pocket TTS english_2026-09_2: official release configuration

Specifications & access

Pocket TTS English 2026-09 revision 2 specifications
DetailValue and conditions
AccessDownloadable
Model identifierspocket-tts:english_2026-09_2kyutai/pocket-tts/languages/english_2026-09_2kyutai/pocket-tts-without-voice-cloning/languages/english_2026-09_2
Backbone depth6 transformer layers.Pocket TTS english_2026-09_2: official release configuration
Weight licenseCC BY 4.0.pocket-tts-without-voice-cloning: Kyutai model card
Voice-cloning weightsRepository acceptance required.pocket-tts: Kyutai model card

Related models

  • An October 2026 English speech checkpoint that retrains September revision 2 with a corrected training loader.

    Launch date not confirmedDownloadable