Tencent releases Hy-SimulMT for real-time Chinese–English interpretation
Hy-SimulMT is a two-way streaming speech translation model under Apache 2.0. Tencent provides the 30B-A3B model, SimulInfer inference service, and SimulTrain tooling. Spoken output and voice cloning require a separate TTS service. Tencent reports roughly three seconds on the StreamLAAL latency metric for long-form RealSI audio.
Visit original source