Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Fakin' The Funk? Detect true quality of your Audio Files (Read 9278 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Fakin' The Funk? Detect true quality of your Audio Files



About
Fakin' The Funk? is a tool that helps you to detect the true quality of your audio files in one batch.
 
Nowadays you can download tunes from anywhere.
But what is the real quality of these files?
What if you could check those tunes to see if they’re really that good, and not upscaled 128kbps to 320kbps MP3s?
You could batch-process them all without having to manually check them 1-by-1 using a spectrum analyzer?
 
Import your music files.
Fakin' The Funk? will analyze the right bitrate and frequency.
If it finds a wrong bitrate or lower frequency-peak, it will show you the actual bitrate.
It’s up to you whether you want to rename those files, copy or move. Or just delete them.
Fakin' The Funk? Keep it real

Q: How fast does Fakin' The Funk? process my files?
A: That's hard to say, because it depends mainly on the speed of the HDD and CPU.
For instance, on a i7 with SSD, Fakin' The Funk? will process 1,000 files in 40 seconds. On the same machine, but importing from a SATA drive, it takes 72 seconds. Importing from an USB drive will be significantly slower.

Q: Does Fakin' The Funk? modify my audio-files?
A: No, Fakin' The Funk? will not modify your files. That means, the content will not change.
But - if you have selected one of those options when saving the results - the file might be renamed / copied / moved as you have selected!

Q: Can Fakin' The Funk? analyze every audio file?
A: No. Fakin' The Funk? supports the common audio-formats MP3, MP4/M4A, OGG, OPUS, FLAC, WMA, AAC, ALAC, MPC, SPX, SFX, TTA and WAV.
Fakin' The Funk? can *NOT* process any DRM protected files.
To get a reliable analysis result, the song length should be at least 10 seconds. Processing files shorter than 0.5 seconds is not recommended and might cause Fakin' The Funk? to fail.

Q: How precise is the value for "actual bitrate"?
A: The analysis process is very defensive. That means: If Fakin' The Funk? reports a file as "fake", then the chance is very, VERY high that there really is something wrong with it.
But being honest, it is just NOT possible to compute a 100% accurate bitrate from the pure audio-stream. So there MIGHT be false positives.



Video: https://www.youtube.com/watch?v=1Cf5acKmB8U
Link: https://fakinthefunk.net/en/
Facebook: https://www.facebook.com/FakeNoFunk/

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #1
TOS14?

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #2
TOS14?

Dude, i am not the author, and i don't even know who the author is. I've been using this tool for a few days and i wanted you all to know about it. Yes, it's my first post, but i am a long time lurker.

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #3
From the descriptions and screenshots (including the features section in the website), I am inclined to believe that this software is more misleading than helping. (And worse, for a paid software with a limited free version)

Quote
9. THIRD PARTY TOOLS
Fakin' The Funk? uses following libraries, which are licensed under MPL or LGPL
Project JEDI Code Library (JCL), see http://www.delphi-jedi.org/
JEDI Visual Component Library (JVCL), see http://www.delphi-jedi.org/
Virtual TreeView, see http://www.soft-gems.net/
taglib.dll (http://developer.kde.org/~wheeler/taglib.html)
tag_c.dll (http://developer.kde.org/~wheeler/taglib.html)
BASS.DLL, basswma.dll, bassflac.dll, bass_ape.dll, bass_aac.dll, bass_alac.dll bass_mpc.dll, basscd.dll, bassenc.dll, bassmix.dll, bass_sfx.dll, bassopus.dll: http://www.un4seen.com/
AudioGenie, http://www.audiogenie.de/  <----- I guess this is actually: https://sourceforge.net/projects/audiogenie/
VCLstyle utils, https://github.com/RRUZ/vcl-styles-utils/
The installer uses the freeware InnoSetup by Jordan Russel, http://www.innosetup.com

So, it uses known tools for decoding and reading files, but it is still unclear what it uses as analysis. If it is strongly based around frequency analysis (and specifically about high frequency cutoff) then it goes against this forum conceptions of how to judge quality. (And could also mean that the screenshot they use as an example, could actually show how much it fails).

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #4
So first, let me introduce myself: My name is Ulrich and I'm the author of Fakin' The Funk. I can confirm that mrkoelie is not related to me and my software in any way. I guess he just wanted to know your opinion on that tool he stumbled upon :-)

He sent me a link to this forum and asked me to reply here.

You are absolutely right, that a sole analysis of the frequency spectrum is not enough to determine if a track is "high quality" or not. But it's the other way round: If you have a file claiming to be 320 kbps but has absolutely nothing above 16 kHz, then you can be pretty sure that this file was upscaled from a 128 kbps version in any way. And I think (hope) you'd agree that upscaling a file does NOT improve its quality.

Fakin' The Funk does not claim to detect the 100% accurate bitrate of a file. But it does a sanity check if the claimed bitrate is in a reasonable range that a regular file of this quality would show.

Yes, Fakin' The Funk looks mainly for frequency cutoffs as described in this scientific paper:
http://www.briandalessandro.com/about/publications/2009_ACM_MMSEC.pdf
I know that not all files show a clear cutoff, especially Fraunhofer encoded files do not. For those ones, it tries to detect an average value: Lower quality files *do* show peaks way above 20 Khz, but only few.

So Fakin' The Funk does mainly a preselection. It's up to the user to review the results and decide if it's a fake (for his ears) or not.
But what I can say from my test is that it pretty good identifies upscaled songs. And that is exactly what it is designed for!

PS:
Fakin' The Funk does also a little more: It scans through the audioframes and tries to decode them. In case of corrupted files, it will notice that the file is broken - and not you in the middle of a party :-D

PPS:
One word about "paid software": 9.99 is really a VERY small amount of money. It's not to get rich, but to buy the coffee I need to keep on coding :-D

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #5
Hello @audioxp

About the paid/free issue: It's not as much about the quantity, but the justification for it. If the main feature might or might not work, it makes me question it. (PS: I'm a software coder, both, as a job, and as an open source advocate).


Now, about the audio quality detector:

Audio quality, and the research of objective evidence of audible difference or not-difference is in the essence of this forum. Note that this "objective evidence" is not based on calculated data, or in absolute value differences, but rather on the perception of them by a group of subjects that, to the best of their possibilities, have avoided to cheat. (this is not the "canonical" definition, but it's ok).

As such, we have studied the quality of lossy codecs a lot, and have collectivelly done several tests ( ABX and ABC-HR tests ) to determine the quality of different codecs at different bitrates.

Then, different codecs work in different ways respect to how they affect the frequency response of the audio, the lowpass frequency analysis would be quite useless for codecs like HE-AAC, or users that try to force the codecs to do what they are not designed to (like not using filter on a 128kbps mp3 file).

What I want to say is that you might have tested using only one side of the equation: I.e. to detect the transcodes that you have created, and assuming that most of the files work in the same way. As such, i think that the images shown in the main website as well as here, might actually be showing a false positive.

To know really if something is not how it should, an analysis of the file data would be needed, or if you want to go into the frequency analysis, you should emulate the encoder in order to detect the quantization that it needed to do in order to get that quality. Of course, a transcoding does not help on detecting that, but I still think it could be a better way to do so.

Basically, the problem is that while auCDText and tau analyzer are trying to see if a lossless file resembles a lossy file, you are looking at lossy files to see if they were already lossy files. (Oh.. and of course, I'm not even mentioning audacity or spek, since those are not designed to detect this).


Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #6
I just tried Fakin the Funk, and while I did find some files that were "fake", then I also noted something odd, which, after some research online, turned out to be just how the programme is made:
The programme can't say if a file is uncompressed wav or mp3 320 kbps. So, when I loaded wav files, it said "bitrate: 1411. Actual bitrate: 320".
It does give the files a thumbs up though, but it's still very confusing for the uninformed, and I find it very unfortunate that the programme doesn't tell you upfront that it can't distinguish between wav and mp3 320. I got a bit paranoid thinking there must be something wrong with my recording software or my ripping software, until I found out that it's simply how the programme is.
"What is asserted without evidence can be dismissed without evidence"
- Christopher Hitchens
"It is always more difficult to fight against faith than against knowledge"
- Sam Harris

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #7
You can skip the analysis in Opus (CELT MODE) files and HE-AAC files. They are all real full quality encodes!

 

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #8
@2012 : You can't skip analyzing a file just because the format is Opus or AAC: If you transcode a 128 kbps mp3 to Opus, then the quality is still bad... (but it claims to be "premium")

@board : The latest version 1.2 will no longer cut-off the bitrat at 320 kbps (e.g. wav, flac etc), since quite some pople got confused by this :-)



Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #11
It is quite hard to estimate the "true" source bitrate, and could be considered waste of effort. It is much easier to tell if a lossless file is likely sourced from lossy of any bitrate and replace it with a proper copy of full quality.

These test items have significantly reduced stereo. To assess the quality, I would generate spectrograms with SoX of all the files, so that the L-R (difference) channel is visible, and then quickly skip through the pictures and look for wide bands or gaps that fall into a grid. I've adapted the "Frontah" multi frontend for this purpose. Various quality testers, which also make SoX spectrograms as part of their functionality, usually output the images too small to be reliable, or not all SoX parameters can be tweaked.

T2 "looks" to be the best quality, with full spectrum up to 17k, but suspect because of the low cutoff. The others are obviously heavily lossy.

T1: http://i.imgur.com/HwpBayY.png


Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #13
Easy to ABX perfectly due to the approximate 100 mSec timing discrepancy.

I'm aware of the discrepancy. I didn't bother to sync the files because the idea is to see what quality this tool assigns to each file.

This would  work if only people could control their biases. Leaving such an obvious biasing influence in place obviously  defeats the purpose of DBTs.   If you don't fix it, it suggest  a malicious attempt to bias the tests to suit a personal agenda.


Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #14
@j7n
I know exactly how the spectrograms look. I'm just curious what quality this tool assigns to each of those files.

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #15
This would  work if only people could control their biases. Leaving such an obvious biasing influence in place obviously  defeats the purpose of DBTs.  If you don't fix it, it suggest  a malicious attempt to bias the tests to suit a personal agenda.

Who said anything about DBTs?
You had me thinking I was in the wrong thread for a moment there.

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #16
@2012 : t3 is reported as 128 kbps, t2 *would* have been reported as 160 kbps if it was not so short and had this "flaw" at the end.
The other ones show a spectrum that is reasonable according to the stated bitrate and therefore marked as OK...


Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #18
@2012 t6 is also flagged as OK. (for me, this spectrum looks "best" :-) as if it was the original source...)

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #19
@audioxp

Okay.

Here is how those files were generated:
Code: [Select]
t1.wav:  lossless -> Opus 64k
t2.wav:  lossless -> sinc lowpass 17k
t3.wav:  lossless -> FDK AAC VBR 4
t4.wav:  lossless -> LAME MP3 128k -> Opus 64k
t5.wav:  lossless -> FDK AAC VBR 0 (HE-AACv2)
t6.wav:  lossless -> lossywav -q P

t2 is actually the best one of the bunch (should be audibly lossless, at least for adults  ;) )

Now. I hope that you understand that I was being sarcastic when I wrote:
Quote
You can skip the analysis in Opus (CELT MODE) files and HE-AAC files. They are all real full quality encodes!

Source file:
https://archive.org/download/unsorted_files/t_original.wav

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #20
Well, maybe t2 is "best" (whatever that means), but nevertheless it is a fake 'cos it states to contain 1411 kbps but it's "only" ~160...

Re: Fakin' The Funk? Detect true quality of your Audio Files

Reply #21
Well, maybe t2 is "best" (whatever that means), but nevertheless it is a fake 'cos it states to contain 1411 kbps but it's "only" ~160...
Dunning Kruger is in full swing.