StepFun and ACE Studio launch StepAudio 3 Music, a song-generation foundation model
StepFun_ai · x · 2026-09-17
StepFun and ACE Studio unveiled StepAudio 3 Music, StepFun's first music-generation foundation model that turns a prompt and lyrics into a complete song. Users can specify genre, mood, vocal character, instruments, key, BPM, and arrangement. One model covers four workflows: song generation, instrumental generation, music covers, and vocal-to-song arrangement. Powered by ABC-COT, it plans musical structure before synthesis, enabling iteration via prompt rewriting and ABC notation editing. An ABC-COT API is coming soon; interactive demo is live.
Related event: StepFun and ACE Studio Release StepAudio 3 Music Foundation Model(2 posts)→
More from Multimodal
- NVIDIA's Axolotl3D does occlusion-aware 3D shape completion from multimodal inputs — NVIDIAAI · 2026-09-17
- fal teases Volume 2 of its generative media report — adamho · 2026-09-17
- 'Er Loop': a dark comedy AI short film retelling Plato's Myth of Er — Electrical_Toe8598 · 2026-09-17
- Runway's Fall 2026 drop aggregates Fish Audio, MiniMax, Cartesia, Flux models in one platform — tlakomy · 2026-09-17
- Runway launches Flux Video Edit to reshape objects and style in any video — runwayml · 2026-09-17
- Gemini Canvas app turns prompts into 3D-printable .STL designs — GeminiApp · 2026-09-17