Fish Audio Drama 3 Heads to Alibaba Cloud
Alibaba Cloud is bringing Fish Audio’s Drama 3 to its platform, expanding access to the 35B-parameter open-weight TTS model for builders and creators. Drama 3 supports natural-language performance direction, mid-sentence emotion changes, multi-speaker scenes, voice cloning, and word-level editing.
This is a strong distribution move: Fish Audio gains Alibaba Cloud’s reach, while Alibaba adds a more director-friendly voice model to its AI stack. The model itself launched last month, so this announcement is about cloud availability and ecosystem expansion.
- –Natural-language direction makes expressive voice generation more accessible than parameter-heavy TTS controls
- –Word-level editing and multi-speaker generation target audiobooks, games, dubbing, and interactive media
- –The 35B-A3B mixture-of-experts design offers substantial capability while activating only 3B parameters per token
- –Open weights give developers a self-hosting option, while Alibaba Cloud can provide a simpler managed deployment path
- –Voice cloning and emotional control increase the need for consent, licensing, and provenance safeguards
DISCOVERED
1h ago
2026-11-04
PUBLISHED
1h ago
2026-11-04
RELEVANCE
AUTHOR
FishAudio
