Step-Audio / Step-Audio2
modelMultimodal audio models supporting 16.7Hz semantic tokenization and real-time "omni" interaction (speech-to-speech).
Model Details
Variants
| Name | Parameters | Notes |
|---|---|---|
| Step-Audio | — | — |
| Step-Audio2 | — | — |
Notes
Step-Audio released Feb 17, 2025. Step-Audio 2 released Jul 23, 2025.