6 tracked versions of the stepaudio-3 family. Timeline ordered newest → oldest. Flagship version marked.
StepAudio 3 ASR Max transcribes audio into text, with contextual recognition of domain terms and mixed Chinese-English speech. The HTTP interface returns incremental and final text over SSE.
StepAudio 3 Chat Preview accepts speech or text through the Chat Completions API and returns text replies, with contextual dialogue and tool calling.
StepAudio 3 Gen Preview generates audio scenes from text and audio references, combining dialogue, sound effects, ambience and music in one clip.
StepAudio 3 Music Preview generates songs or instrumental music from descriptions, lyrics, vocals or reference audio. It supports accompaniment and arrangement workflows through an asynchronous API.
StepAudio 3 Realtime Preview supports full-duplex voice conversations, interruption handling, reasoning and tool use over a persistent WebSocket connection.
StepAudio 3 TTS converts text into speech with controllable delivery and streaming output. It offers voice synthesis and a separately billed voice-cloning operation.
See every provider hosting StepFun models at /provider/stepfun.
Fast Pivot — Access hundreds of AI models from a single API — competitive pricing, reliable uptime, enterprise-grade infrastructure.