MiniMax Speech 2.8
Speech 2.8 adds native filler/breath 'sound tags', 10-second voice cloning and cleaner audio, with Mandarin-Japanese accent fixes.
A text-to-speech update focused on human imperfections (um, uh, pauses) and cross-lingual accent bleed. Available in MiniMax Audio and the API.
- Date
- Friday, 23 January 2026
- Lab
- MiniMax
- Kind
- model
- Access
- closed API
Predecessor Speech-02 (April 2025, 30+ languages) per Wikipedia.
Sources
This record was checked against its sources on 6 October 2026. How we check