Skip to main content

Task type: Automatic speech recognition

Services of the asr type accept a byte array as input and return JSON in a specific format. The stream recognition mode is supported.

System name: asr

List of services

ASR stands for Automatic Speech Recognition. The task of automatic speech recognition is the conversion of an audio stream into text.

Data types​

Predict request​

Request format:

{
  "audio_base64": "audio data"
}

Predict response​

The response format depends significantly on the specific service. The basic interface defines only the required fields, common and identical for all services of this type.

Required response fields:

{
  "text": "Hello"
}

In addition to the single text field, the response will contain a multitude of data specific to the particular service. For information about this data, see the service page.