All Products
Search
Document Center

Alibaba Cloud Model Studio:Real-time audio and video translation

Last Updated:Nov 12, 2026

Qwen-LiveTranslate-Realtime translates speech in real time and can use images or video frames as context. It returns translated text. Some target languages also support spoken translations.

The series supports WebSocket, WebRTC, and AOQ. The series guide covers qwen3.8-livetranslate-flash-realtime, qwen3.5-livetranslate-flash-realtime, and the qwen3.5-livetranslate-flash-realtime-2026-05-19 snapshot. The current qwen3.5-livetranslate-flash-realtime model has the same capabilities as that snapshot.

Protocol Selection

For general guidance, see Choose an access method. See the access and event references for the inputs, outputs, and event fields available for your model.

  • With WebSocket, send speech in audio events and optionally include images as context for translation.

    Browsers natively support WebSocket. Follow the model's authentication requirements and keep long-term API keys on the server.

  • With WebRTC, use audio tracks to send the original speech and receive the spoken translation. Use the DataChannel for model events.

  • With AOQ (AI over QUIC), use the client SDK to transmit media and model events on separate tracks.

    The client SDK is available for Android, iOS, HarmonyOS, Windows, macOS, Electron, and Linux. The AOQ access guide links to SDK downloads and platform-specific API references.

Video input over WebRTC or AOQ depends on the model and use case.

All three protocols can be used for Token Plan. Available models and regions are subject to the specifications outlined in Token Plan Personal Edition and Team Edition.

Events and Other References

Voice Styles