PerfectTUNES discussion

[Porcus]

Re: PerfectTUNES discussion

Also, Spoon: with your betas, I am used to not clicking the uninstall and reinstall button ... I suppose that's still the default thing to do?

(And I suppose you would save yourself some questions if you put that information in the stickied announcement thread )

[Porcus]

Re: PerfectTUNES discussion

Is it possible to report to the user how close a match the potentially duplicate files are? And categorize duplicates in something similar to the following order?

0. Filename, audio and metadata exact match. Allows easy discovery of duplicate folder contents.
1. Audio and metadata exact match: audio and metadata are identical, filename may differ. Think about whether this should checksum the entire file contents or have separate parsing and checksums of audio and meta data.
2. Audio-only exact match: audio data is bit-identical, metadata is different.
3. Audio fingerprint match only: audio data was probably encoded from the same source but there are some differences due to lossy encoding or non-accurate rips.

I agree (the exact level of detail could very well be up to discussion though, see below). What about, say, a pulldown to ignore metadata differences.

But, what I would like to see, is a "audio is bit-identical bar offset" kind of thing -- this requires some tolerance tuning, as it strictly speaking means that the tracks are not identical, one is not even a subset of the other, rather one begins first and the other ends last. And with different conventions on gaps appended or prepended ... this requires some work yes. And I'd like to have it working on folders -- basically if the utility can tell me that folder X and folder Y contains the same except a few frames at the beginning or end. (As of now, I am using cuetools for that ... reporting CRC w/o zeroes. Always looking for a better solution.)

Also, I would like the other side of the story: check for folders that are 'original' and 'remaster' (or 'remaster1' and 'remaster2') versions of the same album. Like, 'these two are the same album but way too different to possibly be the same (pre-)master tape'.



(As usual, I am throwing out suggestions à la carte, hoping some might serve whatever the purpose of this software really is ... which I don't know yet. But don't we all.)

[balma01]

Re: PerfectTUNES discussion

Tried de-dup and it works perfectly for CERTAIN MATCHES. The POSSIBLE MATCHES list contains tracks very very different, I can't understand how they can be considered POSSIBLE MATCHES...
WHISHLIST: to have an ACKNOVELEDGE button to put some 'desired' duplicates in another list (after CERTAIN MATCHES and POSSIBLE MATCHES).
Thanks Spoon.

[Spoon]

Re: PerfectTUNES discussion

The technology would not easily transfer to a library.

[TJOz]

Re: PerfectTUNES discussion

Spoon - is there a possibility of making the "fingerprinter" available as a library that could be called by other applications? I would be more than happy to pay for this sort of functionality.

I have about 13k tracks and don't want to delete duplicate tracks as I want to keep CDs complete, what I want to do is delete (and sell) CDs whose tracks are all duplicates and as I have many compilations this could be a few. Plus this is a great way of matching tracks, not just to identify duplicates, but to identify mistagged tracks that would otherwise not be found. I have already written an app that does this, but it relied on tags, so step one was getting tags right. The idea of the fingerprinter kills two birds with one stone - helps fix tags and identifiess duplicates by content not tags - brilliant!!

[bhoar]

Re: PerfectTUNES discussion

Is it possible to report to the user how close a match the potentially duplicate files are? And categorize duplicates in something similar to the following order?

0. Filename, audio and metadata exact match. Allows easy discovery of duplicate folder contents.
1. Audio and metadata exact match: audio and metadata are identical, filename may differ. Think about whether this should checksum the entire file contents or have separate parsing and checksums of audio and meta data.
2. Audio-only exact match: audio data is bit-identical, metadata is different.
3. Audio fingerprint match only: audio data was probably encoded from the same source but there are some differences due to lossy encoding or non-accurate rips.

Item *0 might be something you wish to leave to other tools, I suppose.

Brendan

[mickc]

Re: PerfectTUNES discussion

Thanks for the new beta - works fine now.

Mick

[dbfan]

Re: PerfectTUNES discussion

A is the answer

[bhoar]

Re: PerfectTUNES discussion

We are working on a fix for the missing (drawn) items

Yup, beta2 fixed the scroll/display issue for me. Thanks, Spoon!

Granted, it found 12396 Duplicates (out of > 43000 files) in my test data set. So, while very clean and simple, the current tool interface isn't a realistic way to deal with that number of dupes.

Any chance of grouping by probable album or artist? While I'm dreaming, I'd be more keen on a spreadsheet-like interface. Easier to find low hanging fruit that way. Sort groups by the lowest bitrate found in the group, etc.

Also, what, exactly is the count of duplicates?
a) total files involved in duplicate matches?
b) * of groups of duplicates?
c) Total files involved minus the number of groups? E.g. the number of files that would be removed if all but one of the files was deleted for each group?

Brendan

[Spoon]

Re: PerfectTUNES discussion

We are working on a fix for the missing (drawn) items

[Spoon]

Re: PerfectTUNES discussion

This is normal as the track is decoded and actually listened to generate a finger print.

[mickc]

Re: PerfectTUNES discussion

It took several hours to check my collection, but I presume it's doing more than just looking at titles.

Mick

[DaFork]

Re: PerfectTUNES discussion

Running de-dup right now. Seems very slow though (a track takes like about half a second to scan). Is this speed supposed 'normal'? Scanning my 12K files is indicated to last over 2 hours, the harddisk is not used for anything else.

In task manager I see 4 threads running 'fingerprintgenerator' taking CPU to full utilisation (double core CPU with hyperthreading, so 4 threads make sense). Likely this is normal behavior, but would like to see what speeds others get.

[DaFork]

Re: PerfectTUNES discussion

Running de-dup right now. Seems very slow though (a track takes like about half a second to scan). Is this speed supposed 'normal'? Scanning my 12K files is indicated to last over 2 hours, the harddisk is not used for anything else.

[mickc]

Re: PerfectTUNES discussion

Have about 43k files, and seeing the same issue as Mick. In addition, once I scroll down quite a bit into the blank boxes area, I can't scroll back up again to the top.

Brendan

Yes, I just tried scrolling down and got the same result.

Mick