Skip to main content
Qwen Cloud home page
Docs
Realtime

Qwen-ASR server events

WebSocket server reference

Events the server sends during a WebSocket session.
User guide: For an overview of features and sample code, see Realtime speech recognition.

error

Sent when a client or server error occurs.
Example
{
  "event_id": "event_B2uoU7VOt1AAITsPRPH9n",
  "type": "error",
  "error": {
  "type": "invalid_request_error",
  "code": "invalid_value",
  "message": "Invalid value: 'whisper-1xx'. Supported values are: 'whisper-1'.",
  "param": "session.input_audio_transcription.model",
  "event_id": "event_123"
  }
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always error.
​
object
body
Error details.

session.created

First event after connection. Contains default session settings.
Example
{
  "event_id": "event_1234",
  "type": "session.created",
  "session": {
  "id": "sess_001",
  "object": "realtime.session",
  "model": "qwen3-asr-flash-realtime",
  "modalities": ["text"],
  "input_audio_format": "pcm16",
  "input_audio_transcription": null,
  "turn_detection": {
      "type": "server_vad",
      "threshold": 0.5,
      "silence_duration_ms": 200
  }
  }
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always session.created.
​
object
body
Session configuration.

session.updated

Sent after your session.update event is processed. If processing fails, an error event is sent instead. For other parameter descriptions, see session.created.
Example
{
  "event_id": "event_1234",
  "type": "session.updated",
  "session": {
  "id": "sess_001",
  "object": "realtime.session",
  "model": "qwen3-asr-flash-realtime",
  "modalities": ["text"],
  "input_audio_format": "pcm16",
  "input_audio_transcription": null,
  "turn_detection": {
      "type": "server_vad",
      "threshold": 0.5,
      "silence_duration_ms": 200
  }
  }
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always session.updated.

input_audio_buffer.speech_started

Sent in VAD mode when speech starts in the audio buffer.
Triggered each time audio is added to the buffer, unless speech start was already detected.
Example
{
  "event_id": "event_B1lV7FPbgTv9qGxPI1tH4",
  "type": "input_audio_buffer.speech_started",
  "audio_start_ms": 64,
  "item_id": "item_B1lV7jWLscp4mMV8hSs8c"
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always input_audio_buffer.speech_started.
​
integer
body
Milliseconds from the buffer start to speech detection.
​
string
body
ID of the user message item to be created.

input_audio_buffer.speech_stopped

Sent in VAD mode when speech ends in the audio buffer. Immediately followed by a conversation.item.created event with the user message item.
Example
{
  "event_id": "event_B3GGEYh2orwNIdhUagZPz",
  "type": "input_audio_buffer.speech_stopped",
  "audio_end_ms": 28128,
  "item_id": "item_B3GGE8ry4yqbqJGzrVhEM"
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always input_audio_buffer.speech_stopped.
​
integer
body
Milliseconds from the session start to when speech stopped.
​
string
body
ID of the user message item created when speech stops.

input_audio_buffer.committed

Sent when the input audio buffer is committed.
Example
{
  "event_id": "event_1121",
  "type": "input_audio_buffer.committed",
  "previous_item_id": "msg_001",
  "item_id": "msg_002"
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always input_audio_buffer.committed.
​
string
body
ID of the previous conversation item.
​
string
body
ID of the user conversation item to be created.

conversation.item.created

Sent when a conversation item is created.
Example
{
  "type": "conversation.item.created",
  "event_id": "event_B3GGKbCfBZTpqFHZ0P8vg",
  "previous_item_id": "item_B3GGE8ry4yqbqJGzrVhEM",
  "item": {
  "id": "item_B3GGEPlolCqdMiVbYIf5L",
  "object": "realtime.item",
  "type": "message",
  "status": "completed",
  "role": "user",
  "content": [
      {
    "type": "input_audio",
    "transcript": null
      }
  ]
  }
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always conversation.item.created.
​
string
body
ID of the previous conversation item.
​
object
body
The conversation item.

conversation.item.input_audio_transcription.text

Sent frequently with real-time recognition results.
Example
{
  "event_id": "event_R7Pfu8QVBfP5HmpcbEFSd",
  "type": "conversation.item.input_audio_transcription.text",
  "item_id": "item_MpJQPNQzqVRc9aC9zMwSj",
  "content_index": 0,
  "language": "en",
  "emotion": "neutral",
  "text": "",
  "stash": "Beijing's"
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always conversation.item.input_audio_transcription.text.
​
string
body
ID of the associated conversation item.
​
integer
body
Index of the content part that contains the audio.
​
string
body
Detected language. If you set the language request parameter, this value matches that setting. Possible values:
  • zh: Chinese (Mandarin, Sichuanese, Minnan, and Wu)
  • yue: Cantonese
  • en: English
  • ja: Japanese
  • de: German
  • ko: Korean
  • ru: Russian
  • fr: French
  • pt: Portuguese
  • ar: Arabic
  • it: Italian
  • es: Spanish
  • hi: Hindi
  • id: Indonesian
  • th: Thai
  • tr: Turkish
  • uk: Ukrainian
  • vi: Vietnamese
​
string
body
Detected emotion. Supported values: surprised, neutral, happy, sad, disgusted, angry, fearful.
​
string
body
Confirmed text prefix. The model has finalized this part and will not change it.
​
string
body
Pre-recognized text suffix. A temporary draft that follows the confirmed part. The model may still correct it.
To get the most complete preview, concatenate both fields: text + stash.

conversation.item.input_audio_transcription.completed

Sends the final recognition result and marks the end of a conversation item.
Example
{
  "event_id": "event_B3GGEjPT2sLzjBM74W6kB",
  "type": "conversation.item.input_audio_transcription.completed",
  "item_id": "item_B3GGC53jGOuIFcjZkmEQ9",
  "content_index": 0,
  "language": "en",
  "emotion": "neutral",
  "transcript": "What's the weather like today?"
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always conversation.item.input_audio_transcription.completed.
​
string
body
ID of the associated conversation item.
​
integer
body
Index of the content part that contains the audio.
​
string
body
Detected language. If you set the language request parameter, this value matches that setting. Possible values:
  • zh: Chinese (Mandarin, Sichuanese, Minnan, and Wu)
  • yue: Cantonese
  • en: English
  • ja: Japanese
  • de: German
  • ko: Korean
  • ru: Russian
  • fr: French
  • pt: Portuguese
  • ar: Arabic
  • it: Italian
  • es: Spanish
  • hi: Hindi
  • id: Indonesian
  • th: Thai
  • tr: Turkish
  • uk: Ukrainian
  • vi: Vietnamese
​
string
body
Detected emotion. Supported values: surprised, neutral, happy, sad, disgusted, angry, fearful.
​
string
body
Transcription result.

conversation.item.input_audio_transcription.failed

Sent if recognition fails for the input audio, separate from other error events so you can identify which item failed.
Example
{
  "type": "conversation.item.input_audio_transcription.failed",
  "item_id": "<item_id>",
  "content_index": 0,
  "error": {
  "code": "<code>",
  "message": "<message>",
  "param": "<param>"
  }
}
​
string
body
Event type. Always conversation.item.input_audio_transcription.failed.
​
string
body
ID of the associated conversation item.
​
integer
body
Index of the content part that contains the audio.
​
object
body
Error details.

session.finished

Confirms that all recognition is complete. Sent after you send session.finish. You can disconnect after receiving this event.
Example
{
  "event_id": "event_2239",
  "type": "session.finished"
}
​
string
body
Unique identifier for this event.
​
string
body
Event type. Always session.finished.