Actions a_transcribe_file

Transcribe File

One-shot transcription of a pre-recorded audio file (URL or base64).

What does it do?

Converts an audio file (like a WhatsApp voice note) into text.

How does it work?

Uses a speech-to-text model to listen to the full file and return its exact transcription.

Why use it?

Your AI assistants don't have ears — they read text. This node is mandatory if you want the bot to understand voice notes your customers send.

Inputs

name	type
`audio_url`	string
`audio_base64`	string

Outputs

name	type
`transcript`	string
`provider`	string
`model`	string
`chars`	number

Branches

Fixed branches

out — Output

Configuration

model string required default: "openai/whisper-large-v3-turbo"

language string optional

audio_url_field string required default: "audio_url"

audio_base64_field string required default: "audio_base64"

mime_type "audio/wav" | "audio/mpeg" | "audio/mp4" | "audio/webm" | "audio/ogg" required default: "audio/wav"

output_field string required default: "transcript"

memory_key string optional

diarize boolean required default: false

max_speakers integer optional

speaker_roles object optional

Default configuration

{
  "model": "openai/whisper-large-v3-turbo",
  "audio_url_field": "audio_url",
  "audio_base64_field": "audio_base64",
  "mime_type": "audio/mpeg",
  "output_field": "transcript"
}

Emitted events

transcript.done
Transcription completed.
transcript.error
Transcription failed.