Skip to content

Speech to Text

Module · Stable

Transcribe audio to text via Groq Whisper (or OpenAI Whisper fallback).

Transcribes audio using Groq Whisper (GROQ_API_KEY) or OpenAI Whisper (OPENAI_API_KEY) as a fallback. Accepts audio_document_id (a stored audio artefact, e.g. a Scout Listen recording, read live from the document store), audio_path (file on disk), or audio_base64 (base64 string). language hint (ISO 639-1) improves accuracy. Used for WhatsApp voice note transcription, Scout Listen recordings, and as a standalone workflow step.

Use when an Incoming WhatsApp trigger (or Document Store) delivers a voice note as audio_base64 or audio_path, or when a Scout Listen step files a recording — read the stored clip with audio_document_id={{ wait_event_payload.memory_item_id }} — and downstream steps need plain text, for example to feed the transcript into Ask AI for intent classification or into Format to compose a reply.

If you already have text and need audio output, use Text to Speech instead. If you need to extract structured fields from a document image rather than an audio recording, use OCR.

Configured per use: audio_document_id, audio_path, audio_base64, mime_type, language, prompt.

  • text
  • engine
  • model

Auto-generated from the skill registry (load_skills()). Do not edit by hand.