Version 0.7 released

[Admin]

- suport APE,FLAC,WV
 - decoders dll's
 - decoders API
 - new experemental algorithm
 - installer support
 - some minor visual changes

[georgetoy]

What most sets this program can compare? Maybe 200 000 mp3?

[Admin]

i test only on 5000 files, it's limit only by available computer memory.

[georgetoy]

Comments: Does not show well bitrate! CPU 100%! Cache does not work for another use!

[familiar]

This is an amazing software. Showing a very accurate result and given at relatively quick time. I'm wondering what programming language do you guys use to write this.

[Admin]

Thanks, it's written on C++.

[familiar]

I'm planning to make a similar program for my thesis, but perhaps far less complex than this. Could you give me any clue as in what should i learn about? Are you planning to make it commercial or is there a possibility releasing it in GPL/copyleft license.

[Anonymous]

почему программа у меня не работает? ведь автоматически начинает сравнить также как добавляю папку и жму ок?

[Admin]

familiar
you can write me email.

Anonymous
возможно файлы немогут быть декодированны

[CJ001]

Admin
My name is Colin I down loaded Similarity a few days ago, wow thanks, saved me weeks sorting, a really friendly and reliable programme, thanks for making it freely available.
 
I'm now looking to populate the Tags with correct data, and looking for a programme like your own that listens/analyses each track and then finds the right info from the the net. Do you know any, and any that are Freeware

[Admin]

CJ001
Winamp + autotag (GraceNote DB), Yahoo Music... (forget full name, also GraceNote DB).

[Raccoon]

I'm wondering if you could add support for comparing the individual channels within a WAV, so to determine if an audio sample is exactly mono (100%; digital copy), fairly mono (99%; probably from a mono->stereo analog source), to stereo (~85% or less).  Could you add such a feature?

Also, I'm trying to compare multiple CDs of similar, often same, material... but I'd like to find any anomalies such as errors in CD extraction (jitter, etc) and any where editing has occurred in a track.  Will your software be able to show a waveform graph highlighting the exact offsets in wave where audio diverges between 2 or more samples?

I'm also curious how your software handles time offsets, such as a track with a longer period of silence at the beginning, or a track that has several seconds of audio inserted in the middle (eg, an extended chorus/loop).

And finally, can your software determine that two tracks are similar/the same, if one of the tracks is truncated/incomplete?

[Admin]

1. about comparing mono: it can be done, but i don't have free time, now, maybe later.
2. It's difficult, not for such kind of program it differently, try to smooth such errors.
3. it's uses average track score, not time, in such case it's ignore silence or its be a small statistic fluctuation(error).
4. see 3

[Raccoon]

Hmm, I think I understand.

So your program doesn't compare a waveform side-by-side over time... Instead it looks at each individual waveform, creates a "fingerprint", and then compares multiple fingerprints?

How would Similarity handle two identical tracks, where one of the tracks is "reversed" (flipped backwards)?  Would both tracks be statistically alike and show as a 100% match?

[Admin]

1. Yes its use "fingerprint", because "fingerprint"ing give very big speed improvment on big collections of files, only after, can be done more precisly comparing, but in this time the program uses only "fingerprint". I have algorithms that can compare very precisly 2 sounds and map 1 sound to another and show similar portions of them, but it very, very slow 10000 times slower than simple comparing "fingerprint".
2. No, its changes some time related scores, but they scores must be similar (i dont know exactly how similar, i didn't test).

PS. experemental algorithm try to compare sounds without "fingerprint".