AI/ML API icon

AI/ML API

Choose from 300+ AI models, from Gemini and ChatGPT to DeepSeek and Llama.

Actions7

Speech Transcription

AI-generated

Overview

This node performs speech transcription, converting audio files into text. It supports transcription from either binary audio files uploaded within the workflow or audio files fetched from a remote URL. Users can select the transcription model, specify various transcription options to fine-tune the process, and choose which part of the transcription response to return, such as the full text, segments, or raw JSON data. This node is useful for automating the conversion of spoken content into text for further processing, analysis, or storage.

Use Case Examples

  1. Transcribing customer support calls recorded as audio files to text for sentiment analysis.
  2. Converting podcast episodes into text transcripts for content indexing and search.
  3. Automatically generating subtitles from video audio tracks.

Properties

Name Meaning
Model Name or ID Select the transcription model or specify its ID to use for converting audio to text.
Input Source Choose whether the audio input is provided as a binary file within the workflow or fetched from a remote URL.
Binary Property Name of the binary property containing the audio file when using binary input source.
Audio URL Public URL of the audio file to transcribe when using remote URL input source.
Extract From Response Select which part of the transcription response to return: segments, text, or full raw JSON.
Transcription Options Collection of fine-tuning options for the transcription request, such as intent detection, language detection, diarization, formatting, and more.

Output

JSON

  • text - The transcribed text from the audio input.
  • segments - Segmented parts of the transcription, if requested.
  • raw - The full raw JSON response from the transcription service, if requested.

Dependencies

  • An API key credential for the AI/ML API service providing speech transcription models.

Troubleshooting

  • Ensure the selected model supports speech transcription; unsupported models will cause errors.
  • Verify that the binary property name matches the actual binary data property in the input when using binary input source.
  • Check that the audio URL is publicly accessible and correctly formatted when using remote URL input source.
  • If the transcription response is empty or incomplete, try adjusting transcription options such as language detection or diarization.
  • Common error messages include unsupported operation errors if the operation parameter is incorrect, or HTTP request failures if the API endpoint or credentials are misconfigured.

Discussion