Groq Speech to Text icon

Groq Speech to Text

Interact with Groq API for AI operations

Speech to Text

AI-generated

Overview

This node converts audio input into text using the Groq API's Whisper speech-to-text models. It supports audio input either from binary data within the workflow or from an external URL. Users can select different transcription models based on speed or accuracy, specify the output format (plain text, JSON, or detailed JSON with timestamps), and customize parameters such as language, prompt, and temperature. This node is useful for automating transcription tasks, such as converting recorded meetings, podcasts, or voice notes into searchable text.

Use Case Examples

  1. Transcribing a podcast episode audio file stored as binary data in the workflow to plain text.
  2. Downloading an audio file from a URL and transcribing it with detailed timestamps for subtitle generation.
  3. Using the English-optimized model to transcribe English audio with JSON output for further metadata processing.

Properties

Name Meaning
Audio Input Selects the source of the audio to transcribe, either from binary data in the workflow or from a URL.
Binary Property The name of the binary property containing the audio file when using binary data input.
Audio URL The URL of the audio file to transcribe when using URL input.
Model The Whisper model to use for transcription, balancing speed and accuracy.
Response Format The format of the transcription response output.
Language The language of the audio specified as an ISO 639-1 code (e.g., 'en' for English).
Prompt (Optional) Optional text to guide the model's style or continue a previous audio segment.
Temperature (Optional) Sampling temperature between 0 and 1 to control randomness in the output; higher values produce more random results.
Timestamp Granularities Granularity of timestamps to include in verbose JSON output, such as word-level or segment-level timestamps.

Output

JSON

  • transcription - The transcription result returned by the Groq API, which can be plain text or JSON depending on the response format.
  • model - The Whisper model used for the transcription.
  • responseFormat - The format of the transcription response as requested.
  • filename - The name of the audio file that was transcribed.

Dependencies

  • Groq API key credential for authentication with the Groq service

Troubleshooting

  • If the node fails to find the specified binary property, ensure the binary data exists and the property name is correct.
  • Errors related to HTTP requests when using URL input may indicate network issues or invalid URLs.
  • If transcription fails, verify the Groq API key credential is valid and has sufficient permissions.
  • Temporary audio files are created during processing; if file deletion fails, it usually does not affect the transcription but may require manual cleanup.

Links

Discussion