Comparing short audio files

[galarlo]

Hi, I'm interested in finding near-duplicate audio files. My dataset is about 3000 thousands short audio files, between 0.5 seconds to 5 seconds. Unlike Shazam, both the "target" audio (i.e. the songs in Shazam's case) and the user input are short, and both might contain noise.

Can this library help? If so, are there any recommendations for tuning parameters?

N.B - if a file is matched to multiple other files, it's fine - I have a less efficient algorithm that can verify which match is correct. In other words, I can handle some amount of false positives, but I don't want false negatives.

[timotech]

Well, I'm not the creator of the library, but 0.5 seconds to 5 seconds is quite small, you will have to try it and check your results. But generally, I believe the library should be able to help resolve your issue.

[galarlo]

@timotech thanks for the response.

Do you have any recommendations on configuration tuning?

[timotech]

Normally you specify the number of seconds in the line below: config.ResultEntryFilter = new TrackMatchLengthEntryFilter(2d);

Where 2d is 2 seconds. Meaning, it should match any sound from when it sees a match around 2 seconds into the song or sound. You can read more here: https://github.com/AddictedCS/soundfingerprinting/wiki/Save-fingerprints-and-identify-tracks and here: https://github.com/AddictedCS/soundfingerprinting/wiki/Realtime-Query-Command I think you can try the 0.5 seconds in that configuration, but I've not tried it.

[jeffward01]

One idea is to find your smallest file, and then multiply it by the number of times to get to seconds. So if your file is (1) second, you'd need to duplicate the file so it is (2) seconds, the caveat is this must be applied to all tracks.

There are pitfalls to this solution of course, but it is a workaround and maybe it will work, or perhaps with your type of tracks it wornt work.

@AddictedCS - There has not been much activity on this, I think we can close it if @galarlo does not reply

[galarlo]

@jeffward01 I assumed it doesn't work with short files because there isn't enough data for the algorithm to do it's magic. If that's right, I don't see how repeating the data will help. What am I missing?

[jeffward01]

Your correct. Your not missing anything, I’m mistaken.

On Sat, May 20, 2023 at 12:36 PM galarlo @.***> wrote:

@jeffward01 https://github.com/jeffward01 I assumed it doesn't work with short files because there isn't enough data for the algorithm to do it's magic. If that's right, I don't see how repeating the data will help. What am I missing?

— Reply to this email directly, view it on GitHub https://github.com/AddictedCS/soundfingerprinting/issues/201#issuecomment-1555948053, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADABSLZHKOSJ3KETGLRXVUDXHDXIHANCNFSM6AAAAAATS22V7E . You are receiving this because you were mentioned.Message ID: @.***>