Anyway to extract Transcript from Hindi audio files ? (Hindi audio files to Hindi transcripts)

Futureized

Futureized

Skilled
I found couple of tools but all are paid, can we extract transcript from audio files in HINDI ?
(didnt liked the accuracy, as basic grammar was completely messed, making a complete different meaning)

Need to extract in source language only = Hindi
 
tearphones said:
A friend works at this startup: https://symbl.ai/platform/understanding-apis/transcription/

They are advertising 1000 mins of free transcription, and hindi is supported. I don't know if they have a tool, though: you may have to write some python.
Click to expand...
Thanks a lot, they provide solutions to websites/app developers with there data model.
Signed up for same and its too extensive, cannot locate any pre-made tool to test my audio files.
Thanks for sharing (its worth a bookmark, will come handy in future)
 
You can trying using the Live transcribe tool from Research at Google. This is available in the Play store. It supports live transcription of audio into the language selected. This is plain transcription which just transcribes what you say literally in the selected language. So if language selected is Hindi and you play an English audio file it will transcribe that into Hindi , no translation involved. You'll have to play the audio file near the microphone of the phone/device where you have installed Live Transcribe But keep in mind a privacy caveat. Hindi transcription is online based and needs an Internet connection on the device. Offline translation is not available yet for Hindi. And what is being played may be stored and used by Google.

There is a roundabout :) way to do this, again using Google's speech 2 text resources i.e. youtube and youtube-dl(thirdparty tool). This utilizes the Auto translate feature available on Youtube.
Rough steps involved:
1) Convert you audio file to video by adding some still images or something like that. This is needed because youtube needs a video file for upload.​
2) Now wait after uploading for Google to process the video and make captions/subtitles available (not sure if there's a way to force this to happen immedidately)​
3) Download the subtitles in the language needed using youtube-dl.​
​
Another method which is simple to try out is the Voice typing feature feature available in Google docs on a Windows 10 PC. I came across this when going down the rabbit hole of transcription in local languages. There may be a privacy caveat here too so use your discretion.
(Before following the steps below ensure that your microphone is functional and you are able to record.)
  1. Open a blank or existing Google doc
  2. Go to Tools > Voice Typing in the Menu.
  3. In the small always on top Pop-up with the microphone symbol that opens up select Hindi in the drop down list of languages and click on the microphone.
  4. Play the sound file in a silent room near the mic and it should type what it hears in Hindi in the doc. Click the microphone symbol again to stop capture.

I've tried this on a Windows 10 machine using Chrome and it worked properly. There may be some mistakes in the transcription but I found the overall accuracy rate to be very good. It should be as it has the resources of Google behind it. You might to adjust the speed of playback of audio if the speaker is speaking rapidly.
For some reason it failed to work in Vivaldi browser on Windows 10 and I was not able to get it to work on Vivaldi in Linux (PopOS) too.
 
Last edited:
Back
Top