SteppedAudio 2.5 ASR released today.

Today, Jieye officially released the new generation automatic speech recognition model StepAudio 2.5 ASR. The core breakthrough of this model lies in the combination of speed and accuracy. It is the first to introduce the inference acceleration technology of large language models into the field of speech recognition. Based on the ASR+MTP-5 deep fusion architecture, the measured inference speed is increased by 400%, latency is reduced by 60%, the peak inference reaches 500 tokens/s, and the inference cost is reduced by 80%.