Stairway launches stepaudio 3 series, achieving multiple global firsts in voice models

On September 15, 2026, Stairway officially launched the StepAudio 3 series, comprising StepAudio3Realtime, StepAudio3ASR, StepAudio3TTS, StepAudio3Gen, and StepAudio3Music. These models achieved top rankings in the Artificial Analysis rankings, with several securing the global first position. All models are now available on the Stairway Star Open Platform. StepAudio 3 Realtime excels in real-time voice interaction, achieving a 98.9% score in the Conversational Dynamics ranking and a 99.7% accuracy in Speech Reasoning. It can understand context, manage interruptions, and perform complex reasoning during conversations. The model also supports parallel processing of tasks and asynchronous execution, allowing for seamless user interaction during ongoing dialogues. StepAudio 3 ASR combines high-precision speech recognition with contextual understanding, achieving a WER of 1.7% in non-streaming speech recognition. It supports multiple languages, dialects, and specialized domains, enhancing accuracy in fields like healthcare and finance. StepAudio 3 TTS generates natural-sounding speech with emotional and tonal variations, mimicking human speech patterns. StepAudio 3 Gen and StepAudio 3 Music offer advanced capabilities in audio and music generation, enabling creators to produce complex audio content with minimal input.