search

worldveil / dejavu

Audio fingerprinting and recognition in Python
MIT License
6.36k stars 1.43k forks source link

Total fingerprints less than confidence #138

Open JArsov opened 6 years ago

JArsov commented 6 years ago

Hi,I have read this issue, and @worldveil answered:

It's the maximum number of fingerprints that could possibly be aligned in time from the sample audio against the reference database audio (fingerprints).

So, I have only one song fingerprinted in the database (Josh Woodward - I want to destroy something beautiful), and it generated a total of 118974 fingerprints in the database. Now, when I test the algorithm with uploading the same file, it gives me confidence value of 127098.

Shouldn't the confidence value always be lower of (or equal to) the total no. of fingerprints from the song? Or did I misunderstand something here?

Thank you, Jordan

morteza-git commented 6 years ago

hi This happens due to some hash repetitions

mauricio-repetto commented 4 years ago

@JArsov as @morteza-git said its because of that and a bug that I've detected and fixed on this pr https://github.com/worldveil/dejavu/pull/205 regarding hash collisions with the same offset. The pr is for dejavu migration to 3.6.6 with support to postgresql and the use of the new mysql connector. I've added a couple of new features as well. Feel free to try it and let me know your thoughts.

Thanks, Mauricio