Maia RouterInstall

Summary

Transcribe audio to text using either a binary file input or an audio URL, with configurable model, language, prompt, response format, and temperature.

Inputs

• transcribeModel (required) — ID of the transcription model to use (openai/gpt-4o-mini-transcribe or maia/whisper-1)
• inputDataMode — Whether to provide audio as binary data or from a URL (binaryData or url)
• binaryProperty (required) — Name of the binary data property containing the audio file (only applicable when inputDataMode is binaryData)
• audioUrl (required) — URL of the audio file to transcribe (only applicable when inputDataMode is url)
• transcribeAdditionalFields.language — Language of the audio (ISO 639-1 format)
• transcribeAdditionalFields.prompt — Optional text to guide the transcription
• transcribeAdditionalFields.response_format — Format of the transcription output (json, text, srt, vtt, verbose_json)
• transcribeAdditionalFields.temperature — Sampling temperature between 0 and 1

Output shape

The transcription result in the format specified by response_format, returned as the json property of the output item. The pairedItem property indicates the input item index.

When response_format is json, the output is a JSON object. For text, srt, or vtt, it is a string. For verbose_json, it is a detailed JSON object.

Examples

Example 1: Transcribe an audio file from a previous node's binary data

Set Input Data Mode to Binary File, Binary Property to 'data' (or the appropriate property name), leave other fields as default for English transcription in JSON format

Example 2: Transcribe audio from a public URL with Spanish language specification

Set Input Data Mode to Audio URL, provide the URL, set Language to 'es' under Additional Fields