Hydrogenaudio Forums

Bern

Apr 4 2004, 04:28

auCDtect - is a program for determining the authenticity of musical CD records, using the character of the audio information it contains, and predict its possible source is it an original studio record or a
reconstruction from a lossy encoded data (for example, MP3).

Current version: 0.6

Accuracy: >95% (for detecting mp3s)

atherean

Apr 4 2004, 07:23

The following has been moved by moderation from Bern's post to save space on HA portal. (See TOS # 14)
___________________________________________________________________

Features:
- Console-based
- Supports wildcards
- Detects wavs converted from MP3-s with different quilities (starting from lame -k, till lame -alt-preset standard)
- pretesting shows detetction of some files converted from MP3-s with --alt-preset insane --alt-preset extreme and --r3mix .
- Pretesting on some reconversions from OGGs too, shows approximately 80% accuracy for -q 8 and -q 9

License: Freeware

Location:
ftp://tta.iszf.irk.ru/aucdtect-lastest.zip (Windows version)
ftp://tta.iszf.irk.ru/aucdtect-lastest.tgz (Linux (i386) version)

Technical details:
http://tta.iszf.irk.ru/index.php?menu=aucd

We hope, it will be useful program...
___________________________________________________________________
Original post by atherean:

Thanks, been looking for something like this for a while, some of the russian cds i own sound suspiciously bad

/\/ephaestous

Apr 4 2004, 08:10

It actually works pretty well, it even detected some --xtreme mpcs

However, whenever dither is selected in the output options it never detects ANY wav as coming from a lossy source.

Rasi

Apr 4 2004, 08:49

just tried...

ripped from original cd to wav... aucdtect couldnt complain...

encoded to musepack -xtreme +xlevel

set playback options in foobar set to dither strong ATH Noise

converted to Wav PCM Fixed Point

aucdtect complains. file being from lossy source

/\/ephaestous

Apr 4 2004, 09:33

Small test:

CODE

aucdtect -m0 *.wav
auCDtect: CD records authenticity detector, version 0.6
Copyright (c) 2004 Oleg Berngardt. All rights reserved.
Copyright (c) 2004 Alexander Djourik. All rights reserved.
------------------------------------------------------------
Processing file: [MP3 - Dithered.wav]
Data analysis: [100%]
------------------------------------------------------------
This track looks like a high-quality CD record

Processing file: [MP3-No Dither.wav]
Data analysis: [100%]
------------------------------------------------------------
This track looks like an MP3 decoded record

Processing file: [MPC - Dithered.wav]
Data analysis: [100%]
------------------------------------------------------------
Could not determine the source of this record

Processing file: [MPC - No Dither.wav]
Data analysis: [100%]
------------------------------------------------------------
This track looks like an MP3 decoded record

Processing file: [Original.wav]
Data analysis: [100%]
------------------------------------------------------------
This track looks like a high-quality CD record

Processing file: [Vorbis - Dithered.wav]
Data analysis: [100%]
------------------------------------------------------------
This track looks like a high-quality CD record

Processing file: [Vorbis - No Dither.wav]
Data analysis: [100%]
------------------------------------------------------------
This track looks like an MP3 decoded record

The MP3 was --alt-preset standard. the mpc was --xtreme -xlevel and the vorbis version was GTune 3 beta 2 @ q8. All were encoded from Original.wav and decoded with foobar2000 with and without Strong ATH noise shaping.

The problem could be that dithering adds noise in the high side of the spectrum, so the analisis of the lowpass is borked (specially since the lowpass in APS is 19Khz anyway).

Johncan

Apr 4 2004, 16:31

I must be missing something obvious, but I can't get the program to work. I downloaded the zip file (for Windows) and put the files in their own directory. When I double-click on the auCDtect.exe nothing happens. Someone please tell me what I have missed.

I am running WinXP Pro.

John

krmathis

Apr 4 2004, 17:27

@Johncan.
It does not have a GUI, so you need to run it from a Command Prompt!

Solaris

Johncan

Apr 4 2004, 17:59

Solaris,

Thanks! I got it going now.

John

M

Apr 4 2004, 20:54

Nice work. Running the program directly from the command line it functions as described; however, for some reason it does not want to function with Speek's Batchenc frontend. Using the line auCDtect <infile> on "test.wav", it gives the following message:

CODE

auCDtect: CD records authenticity detector, version 0.6
Copyright (c) 2004 Oleg Berngardt. All rights reserved.
Copyright (c) 2004 Alexander Djourik. All rights reserved.
------------------------------------------------------------
error: can not open file test.wav
-------------------------------------------------------------------------------

- M.

zver

Apr 5 2004, 03:18

Well,this looks like the same company-developers which make tta lossless encoder,and i remember their post here when they were saying that their compressor will compress better then all present lossless codecs-at least they were claiming that????Somebody knows more about that like where to find a plugin for foobar couse i saw the binaries on their site and i wouldnt mind to test it to see it for muself

And one for developers-Any chance of making a gui for this piece of software??
Thanks!!

ald

Apr 5 2004, 05:34

QUOTE

Well,this looks like the same company-developers which make tta lossless encoder, and i remember their post here when they were saying that their compressor will compress better then all present lossless codecs-at least they were claiming that????

Who said? If you speek about a comparisons results.. Yes sometimes it compress better, but never always. Please, be correctly.

TTA compressor comparisons results

QUOTE

Somebody knows more about that like where to find a plugin for foobar..

Foobar, Apollo and Winamp plugins you can found in a TTA distribution package, also these plugins are available separately

TTA compressor download page

QUOTE

And one for developers-Any chance of making a gui for this piece of software??

Yes, all of GUIs are planed, but only 2 developers work with TTA project and we are very busy at now.

P/S: New TTA format coming soon, which is made especially for hardware support. It's will be released as a TTA1 format. Its compression characteristics corresponds to TTA 2.0 encoder results in fast mode (-e1). Also we start a work with new TTA-based format (not hardware oriented), to reach a maximum compression results.

-- Alexander

WaldoMonster

Apr 5 2004, 10:24

Alexander can we upload files that gave the wrong information to improve auCDtect?

auCDtect.exe only works when the wave files are in the same directory.
Otherwise I get the following error:

QUOTE

D:\Console\Codec>auCDtect.exe d:\temp\test.wav
auCDtect: CD records authenticity detector, version 0.6
Copyright © 2004 Oleg Berngardt. All rights reserved.
Copyright © 2004 Alexander Djourik. All rights reserved.
------------------------------------------------------------
error: can not open file test.wav

D:\Console\Codec>

ald

Apr 5 2004, 11:51

QUOTE

Alexander can we upload files that gave the wrong information to improve auCDtect?

Yes. Its possible. I can open FTP, but it will be better if you would post a codec name, its settings and program log with -v parameter.

QUOTE

auCDtect.exe only works when the wave files are in the same directory.

Thank you. It will be corrected quickly and also I'll inform you about it in this forum.

-- Alexander

WaldoMonster

Apr 5 2004, 14:37

Works pretty well with the current LAME and Musepack.
But has some problems with older LAME versions (post it next time).

WaldoMonster

Apr 5 2004, 15:14

Not good detected or "Could not determine" from 26 mp3 files:
All files decoded with LAME 3.90.3

This track looks like a high-quality CD record
LAME 3.86 - CBR256 (Bomfunk MC's – Freestyler)
LAME 3.88 - ABR256 (Darude - Sandstorm)
LAME 3.87 - CBR256 (Paolo Conte - Via Con Me)
LAME 3.87 - CBR256 (Stardust - The Music Sounds Better With You)

Could not determine the source of this record
LAME 3.87 - CBR256 (Brandy & Monica - The Boy Is Mine)
LAME 3.87 - CBR256 (Chef - Chocolat salty balls)
LAME 3.87 - CBR256 (Faithless - Don't Leave)
LAME 3.87 - CBR256 (Jennifer Paige – Crush)
LAME 3.87 - CBR256 (Moloko - The Time Is Now)
LAME 3.87 - CBR256 (Nomansland - 7 Seconds)
LAME 3.87 - CBR256 (Postmen – Crisis)
LAME 3.86 - CBR256 (Propellerheads - History Repeating (Shirley Bassy))
LAME 3.87 - CBR256 (Run DMC - It's Like That)
LAME 3.87 - CBR256 (The Artful Dodger – Rewind)
LAME 3.88 - ABR256 (Zita swoon - Hot hotter hottest)

ald

Apr 5 2004, 17:22

Thank you for your tests! Going to test auCDtect with all of available versions of LAME encoder.

QUOTE

Not good detected or "Could not determine"

"Could not determine" - We going to reduce the count of such messages, at now its not a error.

QUOTE

This track looks like a high-quality CD record

This is a program error if its not true. For about of 200 of the tested albums we have approx. 5% of such files. At now the algorithm has only 2 of known unstable points for standard LAME 3.92 files and both of them are already fixed, but not released because the problem with dithered files is not resolved. Please wait for a version 0.7 for this tests. I think we can to release it in a next week.

P/S: This version (0.6) may to show a wrong result for a very small count of files, but we think it can be usefull. Thus, all of the album tracks must be tested to determine the CD source.

-- Alexander

WaldoMonster

Apr 5 2004, 19:20

I will test the program some more when version 0.7 arrives.
It would be nice if the program can return different errorlevels for the source prediction.

zver

Apr 5 2004, 21:02

QUOTE

Who said? If you speek about a comparisons results.. Yes sometimes it compress better, but never always. Please, be correctly.

I thought i saw a post before here,really apologize if it wasnt the case

Anyway just downloaded a package so will try to see how it goes.

mpcfiend

Apr 6 2004, 01:57

QUOTE

This track looks like a high-quality CD record
LAME 3.86 - CBR256 (Bomfunk MC's - Freestyler)
LAME 3.88 - ABR256 (Darude - Sandstorm)
LAME 3.87 - CBR256 (Paolo Conte - Via Con Me)
LAME 3.87 - CBR256 (Stardust - The Music Sounds Better With You)

Yes, well, this says more about the state of today's music than the state of LAME (God forbid

) or the state of the detection algorithms. How much you wanna bet some recordings come off the CD already having been compressed with a lossy codec?

Edit: Sorry, let alone have their dynamic range completely fuctwith as well.

ald

Apr 6 2004, 02:33

Sorry for offtopic, about TTA:

QUOTE

Anyway just downloaded a package so will try to see how it goes.

TTA project has revived not for purposes of reaching a best results in compression.
I think it can be a goal of the next of our projects. At now we done our work with a simple and effective lossless audio format for hardware implementation. This format will be released at soon as TTA1 (based on encoder v3.0). New encoder compression characteristics corresponds to TTA v2.0 encoder results in fast mode (-e1), but its can be compiled and executed on a several different platforms, has a minimal system requirements (em85xx compatible), 30% faster, provides immediately seeking and can to decode partially corrupted files. Thus, you can to test TTA v2.0, but I recommend you to wait for the format release.

-- Alexander

ald

Apr 7 2004, 06:27

QUOTE

D:\Console\Codec>auCDtect.exe d:\temp\test.wav
auCDtect: CD records authenticity detector, version 0.6
Copyright © 2004 Oleg Berngardt. All rights reserved.
Copyright © 2004 Alexander Djourik. All rights reserved.
------------------------------------------------------------
error: can not open file test.wav

D:\Console\Codec>

FIXED. Sorry for delay.

-- Alexander

xmixahlx

Apr 7 2004, 08:42

hey ald,

bumping this thread to flag you down for this:
http://www.hydrogenaudio.org/show.php/showtopic/20462

later

WaldoMonster

Apr 7 2004, 10:00

QUOTE (ald @ Apr 7 2004, 06:27 AM)

QUOTE

D:\Console\Codec>auCDtect.exe d:\temp\test.wav
auCDtect: CD records authenticity detector, version 0.6
Copyright © 2004 Oleg Berngardt. All rights reserved.
Copyright © 2004 Alexander Djourik. All rights reserved.
------------------------------------------------------------
error: can not open file test.wav

D:\Console\Codec>

FIXED. Sorry for delay.

-- Alexander

Thanks very much for the update.
Works perfect also tested with long file names and spaces in the path.

Why do you not increase the version number with this update?

ald

Apr 7 2004, 11:55

QUOTE

Thanks very much for the update.
Why do you not increase the version number with this update?

I think its not required. Its not an update, just a small fixing ;-))

M

Apr 7 2004, 14:44

QUOTE (ald @ Apr 7 2004, 12:27 AM)

QUOTE

D:\Console\Codec>auCDtect.exe d:\temp\test.wav
auCDtect: CD records authenticity detector, version 0.6
Copyright © 2004 Oleg Berngardt. All rights reserved.
Copyright © 2004 Alexander Djourik. All rights reserved.
------------------------------------------------------------
error: can not open file test.wav

D:\Console\Codec>

FIXED. Sorry for delay.

-- Alexander

... and this fix also makes it usable with Speek's Batchenc frontend. Parameters are simply auCDtect <infile> for default performance, or auCDtect -m0 <infile> for "slow and most accurate."

- M.

Veej007

Apr 16 2004, 00:42

wait, can this detect transcodes?

Supernaut

Apr 16 2004, 08:03

QUOTE (Veej007 @ Apr 15 2004, 03:42 PM)

wait, can this detect transcodes?

If you transcoded A to B and the program detected A as "a decoded MP3" to begin with (which it currently does extremely well for me), it will most certainly detect B as decoded as well.

EDIT: but not transcodes as in "looks like the file you've given me was first encoded to Musepack and then converted to MP3"; it will just give you the usual "this track looks like a decoded MP3"

Jan S.

May 2 2004, 13:22

Could source code be made available? A foobar plugin would be very useful IMO to check bootlegs. Saves time not having to decode.

aSceT

May 4 2004, 02:11

QUOTE (Jan S. @ May 2 2004, 04:22 AM)

Could source code be made available? A foobar plugin would be very useful IMO to check bootlegs. Saves time not having to decode.

100% agreed! Foo plugin will be great! But as far as i understand you may not need source code for making this plugin - it can work with binary, like CLI encoder, and show log somewhere in separate window or in console

mat128

May 4 2004, 04:12

Any chance seeing binary for OS X [or source btw]? That way I could bring my laptop, extract a track from someone's cd and test it out before purchasing the said cd from that person!
If you dont have any mac to compile it well I don't know if I can really help...

Bern

May 6 2004, 03:41

QUOTE (mat128 @ May 3 2004, 07:12 PM)

Any chance seeing binary for OS X [or source btw]?

mat128,
We plan to support Macs, but might be to the end of this year, when we will have enought financials to buy a mac

. For now we support only INTELs.
About source code - unfortunately, we do not open the source code of this software

Best Regards,
Oleg Berngardt,
auCDtect developer.

WaldoMonster

Jun 29 2004, 09:01

auCDtect version 0.7 beta released
http://tta.iszf.irk.ru

JL6161

Jul 21 2004, 02:10

Bern:

Would it be helpful to you to get some minidisc and digital FM results? Some friends and I are getting very high probability MPEG results on some known minidisc recordings. If auC is analyzing spectral characteristics, is there any way to refine its recognition of MD/FM compression vs. source MP3 compression?

Bern

Jul 30 2004, 03:39

QUOTE (JL6161 @ Jul 20 2004, 05:10 PM)

Bern:

Would it be helpful to you to get some minidisc and digital FM results? Some friends and I are getting very high probability MPEG results on some known minidisc recordings. If auC is analyzing spectral characteristics, is there any way to refine its recognition of MD/FM compression vs. source MP3 compression?

We are planning to detect more accurately some other lossy audio formats (say, OGG) ,
but for now we do not have FM or minidiscs tests in our plans.

Thank you very much for idea and help,
Best Regards,
Oleg.

~*McoreD*~

May 3 2007, 06:33

QUOTE (Jan S. @ May 2 2004, 21:22)

Could source code be made available? A foobar plugin would be very useful IMO to check bootlegs. Saves time not having to decode.

Seems like a great program aucdtect. Is this project still active? I was looking for a program that could detect transcodes.