3
$\begingroup$

I have audio files, most of them start with the same music, and then a conversation begins. I want to trim the part of the music (which can be varied in length). I have no labels, I can transcribe the whole file using off-the-shelf models, but the music itself contains words which are resulted in false positives. but I know to extract features from the audio, such as Mel spectrogram, pitch, etc. The music at the beginning of the file can easily be noticed by looking at the spectrogram or just at the sound wave (please see the following images).

I thought about using a knn with a high number of neighbors, and then filtering the audio based on its values. Is there a more obvious way?

Thanks!

Wave

Spectogram

$\endgroup$
2
  • 1
    $\begingroup$ knn on the spectrum (the Fourier transform) of each small bin sounds good for a first try, this would be supervised though. $\endgroup$
    – Valentas
    Dec 28, 2021 at 8:37
  • $\begingroup$ Thanks for your answer, eventually, since the data includes only phone calls, I have noticed that there is a "BIP" that separates the conversation from the music at the beginning. So I convolved it over files and achieved better results than k-means and GMMs. $\endgroup$ Dec 29, 2021 at 23:39

1 Answer 1

0
$\begingroup$

Eventually, since the data includes only phone calls, I have noticed that there is a "BIP" that separates the conversation from the music at the beginning. So I convolved it over files and achieved better results than k-means and GMMs.

$\endgroup$
1
  • $\begingroup$ Your answer could be improved with additional supporting information. Please edit to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers in the help center. $\endgroup$
    – Community Bot
    Dec 30, 2021 at 3:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.