Configure Translation

2026-07-13

Feature Overview

During real-time speech recognition (ASR), you can enable the translation feature to translate recognized audio content into the target language in real-time.

This feature supports the following two main configuration modes:
- Room-level translation: All streams in the room share one translation configuration.
- Stream-level translation: Set personalized translation configurations for different streams in the room (e.g., different target languages).
Currently supported translation models include: doubao-seed-translation, Qwen-MT, etc. For translation languages supported by different models, please refer to the official documentation of the corresponding model.
Supported methods for obtaining translation results:
- Server callback to get translation results.
- Real-time audio/video room signaling (ZEGO Express SDK) callback.

Core Parameter Configuration

Configure parameters when creating a real-time speech recognition task (StartRealtimeASRTask).

Parameter	Type	Required	Description
`RoomId`	String	Yes	RTC room ID
`RecognitionRange`	Int	No	Recognition range. 0: entire room, 1: specified StreamList
`StreamList`	array of object	No	List of streams to recognize, effective when RecognitionRange is 1.
`EnableTranslation`	Bool	No	Whether to enable translation
`Translation`	Object	No	Translation LLM configuration item
`SubtitleType`	Int	No	Subtitle delivery type via room signaling, default is 0: 0: No delivery 1: Recognition results only (e.g., when a user says "你好" in Chinese, delivers "你好") 2: Translation results only (e.g., when a user says "你好" in Chinese, delivers the translated "Hello") 3: Both recognition and translation results (e.g., when a user says "你好" in Chinese, delivers both "你好" and "Hello") If client UI needs to display subtitles, it is recommended to set to `1`, `2`, or `3`

Usage Examples

Prerequisites

Use a translation model supported by ZEGO cloud real-time speech recognition
Have already activated the translation service and obtained the model's API Key.

Start Recognition Task with Translation and Display Translation Results on Client

Get Translation Results

Get Translation Results via Server (Non-streaming)

Refer to the Receiving Callbacks document to get the Data result when Event is TranslationResult.

(Optional) Display Subtitles on Client

To display translation subtitles, please refer to the Display Subtitles document.