KyutaiMimi
Mimi
Mimi is a neural audio codec. It converts speech into a sequence of compressed tokens and reconstructs sound from them. It is a component used by speech models, not a conversational model by itself.
Research release
The tokens retain speech content and sound
Mimi combines semantic and acoustic information. Its streaming encoder and decoder process short segments as they arrive. A language model can predict these tokens without generating a waveform sample by sample.mimi: Kyutai model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | kyutai/mimi |
| Frame rate | 12.5 frames per second.mimi: Kyutai model card |
| Published configuration | 1.1 kilobits per second.mimi: Kyutai model card |
| Weight license | CC BY 4.0.mimi: Kyutai model card |
Related models
- Moshi Moshika
Kyutai · Moshi
A Moshi speech-dialogue model that listens while speaking, using Kyutai's Moshika voice.
Launch date not confirmedResearch releaseDownloadable - Moshi Moshiko
Kyutai · Moshi
A Moshi speech-dialogue model that listens while speaking, using Kyutai's Moshiko voice.
Launch date not confirmedResearch releaseDownloadable