Download
Step-Audio-EditX is an open-source, 3 billion-parameter audio model from StepFun AI designed to make expressive and precise editing of speech and audio as easy as text editing. Rather than treating audio editing as low-level waveform manipulation, this model converts speech into a sequence of discrete “audio tokens” (via a dual-codebook tokenizer) — combining a linguistic token stream and a semantic (prosody/emotion/style) token stream — thereby abstracting audio editing into high-level token operations. This allows users to modify not only what is said (the text) but also how it's said: emotion, tone, speaking style, prosody, accent, even paralinguistic cues. Because the model is trained with a “large-margin learning” objective over many synthesized and natural speech samples, it gains robust control over expressive attributes, and can perform iterative editing: e.g. you could record a line, then ask the model to “make it sadder,” “speak slower,” or “change accent to X.”
Features
- Token-based audio editing: converts speech to discrete token streams for high-level, language-like editing operations on audio
- Dual-codebook tokenizer design: separates linguistic content and prosody/style — enabling control over both what is said and how it's said
- Expressive editing: allows modifying emotion, tone, accent, speaking style, prosody, pacing, and other vocal attributes without re-recording
- Iterative editing workflow: supports multiple rounds of edits — e.g. change style, then adjust emotion, then pace, etc.
- Zero-shot TTS: generate speech directly from text + optional style/emotion instructions, in a controlled expressive voice
- Open-source model & code under permissive license — enabling integration, customization, and use in research, creative workflows, or production
Project Samples
