API Reference¶

Details on API request and response protocols for the ASR, MT, TTS and Video Interpolation services have been provided here.

ASR (Speech-to-Text)¶

The endpoint for ASR service is https://asr.iitm.ac.in/internal/asr/decode

Request keys¶

Key	Description
`file`	the media file to be transcribed
`language`	the language of the source audio/video in all lowercase (eg: `hindi`, `english`)
`vtt` (optional)	whether a webVTT caption file has to be generated. This is an optional value. It accepts two string values either `true` or `false`. By default, this is `false`. This can be used for captioning purposes.

Response keys¶

Upon successful service of the request, the API returns a JSON response with the following keys:

Key	Description
`status`	success
`time_taken`	time taken to transcribe the given audio/video in seconds
`transcript`	the transcription of the given speech input, as infered by the deployed model
`vtt`	WebVTT caption if it was requested. i.e, if the vtt key was set to `true` in the request, a WebVTT caption would be returned.

In case of a service failure, the API returns a JSON response with the following keys:

Key	Description
`status`	failure
`reason`	a reason for the failure in serving the request

Supported Languages¶

Usage¶

PythoncURL

import requests

files = {
'file': open('speech_sample1.mp3', 'rb'),
'language': (None, 'hindi'),
'vtt': (None, 'true'),
}

response = requests.post('https://asr.iitm.ac.in/internal/asr/decode', files=files)
print(response.json())

curl -v -X POST -F 'file=@speech_sample2.wav' -F 'language=english' \
-F 'vtt=true' https://asr.iitm.ac.in/internal/asr/decode

Sample audio files to test the API: english speech, tamil speech.

The ASR API accepts media files from most of the common formats such as .mp3, .mp4, .wav, .ogg etc.

Web Demo interface available at https://asr.iitm.ac.in/demo/asr

Last update: August 2024
Created: March 24, 2023