LiveTranslate client events

This topic describes the client events for the qwen3.5-livetranslate-flash-realtime API.

Connect

Establish a WebSocket connection to start a session. The server sends a session.created event when the connection is ready.

Configuration	Value
Endpoint	`wss://dashscope-intl.aliyuncs.com/api-ws/v1/realtime`
Query parameter	`model=qwen3.5-livetranslate-flash-realtime`
Auth header	`Authorization: Bearer $DASHSCOPE_API_KEY`
Protocol	JSON text frames

Full URL:

wss://dashscope-intl.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime

session.update

Updates the session configuration after you connect. The server validates parameters and returns the full configuration, or an error if any value is invalid.

Example

{
  "event_id": "event_ToPZqeobitzUJnt3QqtWg",
  "type": "session.update",
  "session": {
    "modalities": [
      "text",
      "audio"
    ],
    "voice": "Tina",
    "sample_rate": 16000,
    "input_audio_format": "pcm",
    "output_audio_format": "pcm",
    "input_audio_transcription": {
      "model": "qwen3-asr-flash-realtime",
      "language": "zh"
    },
    "translation": {
      "language": "en"
    }
  }
}

Example with both the source and target languages set to English and text and audio output skipped:

{
  "event_id": "event_xxx",
  "type": "session.update",
  "session": {
    "input_audio_transcription": {
      "language": "en"
    },
    "translation": {
      "language": "en",
      "same_language_skip_options": {
        "skip_text": true,
        "skip_audio": true
      }
    }
  }
}

string

body

required

Always "session.update".

object

body

Session configuration.

Show properties

array

body

Output types. Valid values:

["text"] — Text only.
["text", "audio"] (default) — Text and audio.

string

body

Voice for audio output. See Supported voices. Default: Tina for Qwen3.5-LiveTranslate-Flash-Realtime, or Cherry for Qwen3-LiveTranslate-Flash-Realtime.

object

body

Input audio settings.

Show properties

string

body

Speech recognition model. When set, the server returns source-language text through conversation.item.input_audio_transcription.text and conversation.item.input_audio_transcription.completed events.Valid value: qwen3-asr-flash-realtime.

string

body

Source language. See Supported languages. If not specified, the model automatically detects the source language.

string

body

Input audio format. Valid values:

pcm (Default) — Uncompressed raw audio data.
opus — Lossy audio codec with low-latency support, suitable for VoIP scenarios.

integer

body

Input audio sample rate, in Hz. Valid values:

8000
16000 (Default)

string

body

Output audio format. Currently, this parameter can only be set to pcm.

object

body

Voice Activity Detection (VAD) configuration, used to control how speech start and end are detected:

Set to a configuration object (default): Enables VAD mode. The server automatically detects speech start and end, commits the audio buffer, and triggers translation responses. The client does not need to send input_audio_buffer.commit events.
Set to null: Enables Manual mode. The client manually commits the audio buffer by sending input_audio_buffer.commit events. The server automatically starts generating translation responses upon receiving the commit.

Show properties

string

body

VAD type. Must be set to server_vad.

float

body

VAD detection sensitivity. A lower value makes it easier to recognize faint sounds (including background noise) as speech; a higher value requires clearer, louder speech to trigger detection. Value range: [-1.0, 1.0]. Default value: 0.2.

integer

body

Minimum duration of silence (in milliseconds) required after speech ends. Once this duration is exceeded, the server determines that speech has ended, automatically commits the audio buffer, and triggers a translation response. Value range: [200, 6000]. Default value: 1000.

object

body

Translation settings.

Show properties

string

body

Target language. See Supported languages. Default: en.

object

body

Same-language output configuration. When the source and target languages are the same, the service can skip text output, audio output, or both. This parameter takes effect only when translation.language is set to zh or en.

Show properties

boolean

body

Specifies whether to skip text output when the source and target languages are the same.

boolean

body

Specifies whether to skip audio output when the source and target languages are the same.

boolean

body

Whether to enable voice cloning. Default: false. When enabled, the model clones the speaker's voice from the input audio for translated output. In this case, voice no longer accepts system preset voices and must be set to default or a voice ID previously created through the Voice Clone API.

object

body

Voice cloning options. Only applies when enable_voice_clone is true.

Show properties

string

body

Controls when voice cloning occurs. Valid values: never (use pre-cloned voice profile), once (clone at session start), always (clone before each response). Default: once.

input_audio_buffer.append

Appends audio bytes to the input buffer. The server uses this buffer for speech detection and submission timing.

Example

{
  "event_id": "event_xxx",
  "type": "input_audio_buffer.append",
  "audio": "xxx"
}

string

body

required

Always "input_audio_buffer.append".

string

body

required

Base64-encoded audio data.

input_audio_buffer.commit

Commits the input audio buffer. This event only needs to be sent in Manual mode (turn_detection set to null). In VAD mode, the server commits automatically and the client does not need to send this event. Upon receiving this event, the server returns an input_audio_buffer.committed event as confirmation and automatically starts generating a translation response (no additional event is needed to trigger the response). If the audio buffer is empty, the server returns an error event.

Example

{
  "event_id": "event_xxx",
  "type": "input_audio_buffer.commit"
}

string

body

required

Event type. Must be set to input_audio_buffer.commit.

input_audio_buffer.clear

Clears uncommitted audio data from the input audio buffer.

Example

{
  "event_id": "event_xxx",
  "type": "input_audio_buffer.clear"
}

string

body

required

Event type. Must be set to input_audio_buffer.clear.

input_image_buffer.append

Adds image data to the buffer from a local file or a real-time video stream. Image limits:

Format: JPG or JPEG. Recommended resolution: 480p or 720p. Maximum: 1080p.
Maximum size: 500 KB (before Base64 encoding).
Must be Base64-encoded.
Maximum rate: 2 images per second.
You must send at least one input_audio_buffer.append event first.

Example

{
  "event_id": "event_xxx",
  "type": "input_image_buffer.append",
  "image": "xxx"
}

string

body

required

Always "input_image_buffer.append".

string

body

required

Base64-encoded image data.

session.finish

Ends the session. The server responds based on whether it detected speech:

Speech detected: The server finishes recognition and sends conversation.item.input_audio_transcription.completed with the result, then sends session.finished.
No speech detected: The server sends session.finished directly.

Disconnect after you receive session.finished.

Example

{
  "event_id": "event_xxx",
  "type": "session.finish"
}

string

body

required

Always "session.finish".

​Connect

​session.update

​input_audio_buffer.append

​input_audio_buffer.commit

​input_audio_buffer.clear

​input_image_buffer.append

​session.finish

Connect

session.update

input_audio_buffer.append

input_audio_buffer.commit

input_audio_buffer.clear

input_image_buffer.append

session.finish