Tasks

Transcribe Task

This task transcribes an audio file to text. Provided with a Stubber fileuuid, this task will transcribe the file and make the resulting text available in the data. This can be used for e.g. voice notes in WhatsApp, or the voice note field in Stubber.

# Basic usage

# Parameters

# files

required

array of objects

An array of file objects that should be transcribed. The supported formats are those supported by the whisper models, as listed here. The file size limit is 25MB.

In the basic usage example above, ~~stubpost.data.voicenote interpolates to an object containing a fileuuid. Written out explicitly without interpolation, the same input looks like:

# model

optional

string

The model used for the transcription. Supported models are whisper-1 and gpt-4o-transcribe-diarize. The latest information can be found here.

Default: whisper-1

# disable_model_response

optional

boolean

By default, the model will respond with the _update_from_transcribe_task feedback action. This can be disabled by setting this parameter to true.

Default: false

# Result

# Properties

# request

object

The files that were used in the task. This is returned as a sanity check to make sure that properties like the fileuuid, filename, and contentType are as expected.

# response

object or array

The response is either an object or an array, depending on the number of files. If only a single file is supplied in the files parameter of the task, the response will be an object with a text property. If multiple files are supplied, the response will be an array of objects, each containing a text property for the corresponding file.

# Examples

# Single voice note and disable model response

The task definition:

# Result

# Using multiple files

You can also pass multiple files to the transcribe task. With the Stubber voicenote field, you can record multiple voice notes in a single field, or you can add multiple voice note fields, and pass the first file of each field as an item in the files array. The latter approach is applied here.

The task definition:

# Result