Qwen ASR icon

Qwen ASR

使用阿里云 DashScope Qwen ASR 进行语音识别

Qwen ASR

AI-generated

Summary

Convert audio data from either upstream binary input or a specified URL into text using Alibaba Cloud DashScope Qwen ASR speech recognition models, with optional language and formatting settings.

Inputs

  • audioSource (required) — Select whether the audio input comes from binary data of an upstream node or a URL.
  • binaryPropertyName — The binary property name containing the audio data when the audio source is 'binaryData'. Defaults to 'data'.
  • audioUrl — The URL of the audio file to transcribe when the audio source is 'url'.
  • model (required) — Select the Qwen ASR model endpoint to use for speech-to-text transcription.
  • options.language — Specify the audio language to improve recognition accuracy; defaults to automatic detection.
  • options.enableItn — Enable inverse text normalization to convert spoken forms into written forms, e.g., transforming 'one hundred twenty three' into '123'.
  • options.systemPrompt — An optional system prompt string to customize recognition context.

Output shape

a JSON array where each item corresponds to an input item and contains the transcribed text, used model name, usage data, and the full ASR API response

Processes each input item independently; errors are thrown if required audio data is missing. The returned data includes multiple metadata fields for flexible workflow use.

Examples

Example 1: Transcribing audio binary data from an upstream node

Set 'audioSource' to 'binaryData' and specify the binary property name containing audio data.

Example 2: Transcribing audio from a web URL

Set 'audioSource' to 'url' and provide the 'audioUrl' parameter with the audio file address.

Links

Discussion