Qwen-ASR — Transcribe audio

Submit async transcription

POST

/api/v1/services/audio/asr/transcription

cURL

curl --location --request POST 'https://dashscope-intl.aliyuncs.com/api/v1/services/audio/asr/transcription' \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header "Content-Type: application/json" \
--header "X-DashScope-Async: enable" \
--data '{
  "model": "qwen3-asr-flash-filetrans",
  "input": {
    "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3"
  },
  "parameters": {
    "channel_id":[
      0
    ], 
    "enable_itn": false
  }
}'

{
  "request_id": "92e3decd-0c69-47a8-xxxxxxxxxxxx",
  "output": {
    "task_id": "8fab76d0-0eed-4d20-xxxxxxxxxxxx",
    "task_status": "PENDING"
  }
}

Async process

Async transcription follows a submit-poll pattern:

Submit a task: Send the request. The server returns a task_id.
Get the result: Poll the result API with the task_id until the task completes.

SDK

Call async_call() (Python) or asyncCall() (Java) to submit. The method returns a task object with a task_id.
Call fetch() with the task object or task_id. The SDK polls automatically until the task completes or times out.

REST API

Use this endpoint to submit a task and get a task_id.
Query the result with the task_id.

Audio formats

Audio file URLs must be publicly accessible. Each audio track in channel_id is billed separately. For example, [0, 1] for a single file incurs two charges.

Authorizations

string

header

required

DashScope API key. Get your API key from Qwen Cloud console.

Header Parameters

enum<string>

required

Must be set to enable for asynchronous task submission.

Available options:enable

Body

application/json

string

required

The model name. Only applicable to Qwen3-ASR-Flash-Filetrans.

object

required

The input object.

Show child attributes

string

required

The URL of the audio file to be recognized. Must be accessible over the Internet.

object

Additional parameters.

Show child attributes

enum<string>

If you know the language of the audio, specify it to improve recognition accuracy.

Available options:zh,yue,en,ja,de,ko,ru,fr,pt,ar,it,es,hi,id,th,tr,uk,vi,cs,da,fil,fi,is,ms,no,pl,sv

boolean

defaultfalse

Specifies whether to enable Inverse Text Normalization (ITN). Applicable only to Chinese and English audio.

boolean

defaultfalse

Controls whether to return word-level timestamps. When false, returns sentence-level timestamps. When true, returns word-level timestamps (supported for Chinese, English, Japanese, Korean, German, French, Spanish, Italian, Portuguese, and Russian). Also affects sentence segmentation: false uses VAD-based segmentation, true uses VAD and punctuation-based segmentation.

string

Context for customized recognition. Provide background text, entity vocabularies, and other reference information. Length limit: 10,000 tokens.

integer[]

default[0]

Indexes of audio tracks to recognize in a multi-track audio file. Index starts from 0. Default: [0]. Each specified track is billed separately.

Response

200-application/json

string

The unique identifier for this call.

Example:92e3decd-0c69-47a8-xxxxxxxxxxxx

object

Show child attributes

string

The task ID. Use this to query the task result.

Example:8fab76d0-0eed-4d20-xxxxxxxxxxxx

enum<string>

The task status.

Available options:PENDING,RUNNING,SUCCEEDED,FAILED,UNKNOWN

Example:PENDING

​Async process

​SDK

​REST API

​Audio formats

Authorizations

Header Parameters

Body

Response

Async process

SDK

REST API

Audio formats