StepAudio 2.5 Realtime is live!
Real-time voice that picks up what you actually mean — tone, pace, pauses, sighs, even the half-laugh mid-sentence.
- Top-tier paralinguistic perception — reads tone, pace, micro-emotions
- Bring-your-own persona via API — personality, backstory, quirks, language style
- 10,000+ native personas → millions of feature combinations
- 5 preset personas to try out of the box
- ZH/EN
RLHF-tuned to hold character even under roleplay stress tests.
Try it → https://t.co/3NZ1fJ4DOu
Model card:
attention — The mechanism that lets a model weigh which earlier words matter for the word it's currently processing — the core operation of a transformer.
RLHF — Reinforcement learning from human feedback — training a model to prefer answers humans rate as better, which turns a raw text predictor into a usable assistant.
Why it matters
A realtime voice model that reads paralinguistic cues like pace, sighs and half laughs, and lets you define a persona through the API instead of picking from fixed voices.
article👏🏻Congratulations！Step3-VL-10B was selected for HuggingFace Daily Papers…
postStep-Audio-R1.1 opens weights for a speech model that reasons in real time