How to avoid big % Dropped

Question

How to avoid big % Dropped

Opened this issue 7 months ago · 5 comments

Hello i want to scan all collection
I read that to improve accuracy, you need to increase the density. I increased the Desity parameter from 20 to 100 and the result is a higher % Dropped. (for example YOUTUBE 01 D2.pklz (79.70% dropped)) Some of the records became less recognizable than with standard Density = 20
How can I increase the output file and decrease the % Dropped to increase the overall recognition accuracy. I take as a basis the material I want to recognize low quality audio from tapes from the early 90's, so accuracy is important. It doesn't matter how much space the database of fingerprints will take up.

Answer 1 · 2024-03-06T12:41:54.000Z

“% dropped” comes from buckets overflowing in the database. You can increase the database size with the bucketsize flag; the default is 100. The database in memory will grow in proportion, and at some point you’ll run out of RAM. DAn.

…

On Wed, Mar 6, 2024 at 07:27 lightcomc ***@***.***> wrote: Hello i want to scan all collection I read that to improve accuracy, you need to increase the density. I increased the Desity parameter from 20 to 100 and the result is a higher % Dropped. (for example YOUTUBE 01 D2.pklz (79.70% dropped)) Some of the records became less recognizable than with standard Density = 20 How can I increase the output file and decrease the % Dropped to increase the overall recognition accuracy. I take as a basis the material I want to recognize low quality audio from tapes from the early 90's, so accuracy is important. It doesn't matter how much space the database of fingerprints will take up. — Reply to this email directly, view it on GitHub <#94>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAEGZUKBROI5XN6CAC73KYLYW4DRZAVCNFSM6AAAAABEI6VCB6VHI2DSMVQWIX3LMV43ASLTON2WKOZSGE3TCMZYHE3TKMI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

Answer 2 · 2024-03-06T19:50:14.000Z

Thank you i'll check thatt parameter to find optimal size and memory comsumption in process, speed is not critical. Started from 4x of current size (400 mb to 1200 mb database) i have about 12 000 tracks (64 gb of music) in each scan

Answer 3 · 2024-03-07T01:28:35.000Z

You can try to use my fork with some minor changes. Recognition speed is increased and memory consumption is reduced. - https://github.com/ZhymabekRoman/audfprint-enhanced

Answer 4 · 2024-04-09T23:57:38.000Z

You can try to use my fork with some minor changes. Recognition speed is increased and memory consumption is reduced. - https://github.com/ZhymabekRoman/audfprint-enhanced

Are you doing any active commits anymore?

Answer 5 · 2024-04-10T13:12:32.000Z

You can try to use my fork with some minor changes. Recognition speed is increased and memory consumption is reduced. - ZhymabekRoman/audfprint-enhanced

Are you doing any active commits anymore?

Yeap