Hy-SimulMT-30B-A3B

Use it to translate live Chinese–English conversations, such as meetings, speeches, and interviews, while preserving the speaker’s vocal characteristics in translated speech.

Availability

Released; model resources linked

Pricing model

Not established

Deployment

GitHub repository provides the SimulInfer inference service

Privacy

Not established

Licence

Apache-2.0

• Supports bidirectional speech-to-speech translation between Chinese and English. ^[1]

  **Hy-SimulMT** is a real-time simultaneous translation model developed by Tencent Hunyuan, supporting **bidirectional speech-to-speech translation between Chinese and English**.

• Continuously receives incoming audio and produces translated text and speech. ^[1]

  It continuously receives incoming audio and translates as the speaker talks, producing both translated text and speech.

• Supports real-time voice cloning that preserves the speaker’s vocal characteristics. ^[1]

  With **real-time voice cloning**, it preserves the speaker's vocal characteristics while conveying the message accurately.

• The model was released with project and model resources linked. ^[1]

  Project and model resources are linked below.

• Speech output and voice cloning require a configured TTS service. ^[1]

  Speech output and voice cloning require a configured TTS service.

• The model card reports approximately 3 seconds StreamLAAL on the RealSI long-form audio evaluation. ^[1]

  In our **RealSI long-form audio evaluation**, Hy-SimulMT achieved a **StreamLAAL of approximately 3 seconds**, the lowest among the systems shown in the comparison.

• New models bring live speech translation and voice features Daily brief 2026-10-28