Noise Reduction for Better Transcriptions

[CowTheGreat]

Preprocess audio files to reduce noise before transcription to improve accuracy.
Utilize libraries like PyDub, Audacity, or pre-trained models for noise reduction. Handling noisy real-world data can be challenging.

[Swayam-code]

@CowTheGreat Can you please assign this issue to me

[CowTheGreat]

Okay I will assign it to you

[CowTheGreat]

@Swayam-code The problem is that currently if there is any music in the audio the transcription doesnt work, we have to remove the music and make it work

[Swayam-code]

@CowTheGreat
you can assign it to someone else. I have tried this. But I don't think I can handle this in better way.

[lukiod]

@CowTheGreat i would like to take this one on

[lukiod]

as for the approch i will be using spleeter which is state of the art music and voice sepration tool

[Swayam-code]

as for the approch i will be using spleeter which is state of the art music and voice sepration tool

Actually I have used spleeter , but didn't get desired results

[lukiod]

oh ok

[lukiod]

@Swayam-code since u tried it what problems were u facine with it ?

[Swayam-code]

@Swayam-code since u tried it what problems were u facine with it ?

There were multiple sources of noises that the model had to distinguish,but it somehow led to some crap output in Russian etc.,idk.

I tried refining it ,but it didn't seem to help.