Dear all,
I got some troubles with flac file , coz I do not know it's really real or fake.(It took me a long time to download, so I do not want to be a victim of cheating ). So I use Audio Checker and Spectro. They infact showed two different results, Audio Checker was 99% CDDA, but Spectro was 60% original.
Here it is :
I dont want to spam, but I think here is a more clearly example : The Audio Checker gave the result 100% but the spectrum given by Spectro looks like mp3's one , the bit rate is 1,441 kbps
(http://img193.imageshack.us/img193/1235/59468111.jpg)
Such questions come up occasionally on HA.. The short answer is that you can't know for sure whether something is a "genuine" FLAC or a transcode, those programs are merely (good) guessers.
Probably the best thing you can do is compare the files to those from another (trusted) source.
Also, you're misinterpreting the info from that spectrum analyzer. 60% size means 60% of the WAV. And the bitrate of FLAC and WAV will of course be different, that's the whole point of compression.
With Spectro you're just looking at frequency content in the two graphs. The 2nd screenshot is showing a recording without much high-frequency content, except for one little spike. Maybe the recording never had much treble content, or it did but has had noise reduction applied to it, or maybe it was taken from a radio broadcast. Or maybe it was subjected to lossy encoding at one point. I find Spectro to be useful for comparing two different lossy versions of the same recording, rather than spotting transcodes, although I've used it for that, too.
AudioChecker is a front end for AuCDtect (Google it). It performs statistical analysis of the frequency content, looking for suspicious patterns. The percentage it gives you is based on its own criteria and isn't really a percentage guarantee of anything. Try it on files you've created yourself, both lossless and lossy, and see what results it gives. I found that it wasn't all that reliable; the presence or absence of high-frequency content can really affect the numbers it gives.
If you're worried about the authenticity of music you get from questionable sources, there's an adage about gift horses that could be said to apply.
The only way you can be absolutely sure that a FLAC is from a lossless source is if you do the encoding yourself.
Otherwise, that's what you get for going on warezzzzzzzz torrent sites and downloading that stuff illegally.
The only way you can be absolutely sure that a FLAC is from a lossless source is if you do the encoding yourself.
Otherwise, that's what you get for going on warezzzzzzzz torrent sites and downloading that stuff illegally.
there are some sites that offer Flac download legally, such as bandcamp and Live Metallica.
I do not want to be a victim of cheating
How do you think the artist of the track you didn't pay for feels?
there are some sites that offer Flac download legally, such as bandcamp and Live Metallica.
True. However, people that use "coz" and "plzzzzzz" in their posts rarely visit and BUY from legitimate FLAC download sites.
Look at the line going across the ~16kHz mark especially near the end. Looks like the source was topped at 16kHz and higher frequencies were guessed and overlaid. ...Unless there is just background noise as if a CRT TV was in the background while the sound was recorded.
I downloaded the Spectro and tested with FLAC files I have. All of them are straight from CDs I own, and none of them are over 70%, lol.
As others said, they are just good guessers.
I downloaded the Spectro and tested with FLAC files I have. All of them are straight from CDs I own, and none of them are over 70%, lol.
*facepalm*
you're misinterpreting the info from that spectrum analyzer. 60% size means [the size of the FLAC is] 60% [of the size] of the WAV.
In other words, the percentage shown by Spectro has nothing to do with the percentage reported by AudioChecker. Spectro is just saying how big the file is, compared to the decompressed version. For example, if the FLAC is 7MB and the audio data you get out of it is 10MB, Spectro will say
Size: 7.00MB (70% of original)
Original Size: 10.00MB
It's not particularly useful info.
Look at the line going across the ~16kHz mark especially near the end. Looks like the source was topped at 16kHz and higher frequencies were guessed and overlaid. ...Unless there is just background noise as if a CRT TV was in the background while the sound was recorded.
Presumably the latter; I fail to see why a lowpass followed by some kind of guessed reconstruction would create such a line.
You know what, IMHO, you should buy that album in a original physical CD format and rip it on your own. That way, you won't have to go to sleepless nights thinking whether you downloaded a fake FLAC-encoded music. And also by doing that, you'll be more than a 100% sure it's genuine FLAC
You know what, IMHO, you should buy that album in a original physical CD format and rip it on your own. That way, you won't have to go to sleepless nights thinking whether you downloaded a fake FLAC-encoded music. And also by doing that, you'll be more than a 100% sure it's genuine FLAC
Not really. I've seen tracks that appeared to be MP3s on a legitimate album. These were some old live recordings added as a bonus and I guess that they didn't survive losslessly, but it shows that buying a CD is not a warranty of anything....though I have to add that I'm not 100% certain they were indeed MP3s. You never know.
Not a guarantee, sure—and I’ve seen that too (and seen threads here about it)—but I would suppose that lossy-sourced tracks on CDs are relatively rare (perhaps very). Plus, music enthusiasts may well be interested in that; whereas none care if someone else’s pirated music is not authentic or about helping them to investigate that possibility! That’s a risk that comes with the territory.
Very rare indeed.
Maybe this can help you in detecting what is "fake" lossless audio:
http://www.flickr.com/photos/60080664@N04/5485470134/ (http://www.flickr.com/photos/60080664@N04/5485470134/)
It's WAV file from an MP3 320kbps CBR source, and an original FLAC ripped from an original CD. Note that the fake WAV has a cut thoughout the track in 22kHz meaning it has retained the psychoacoustic tricks applied to its MP3 source. On the other hand, the original FLAC retains its full quality. This is probably the best example to know what is FAKE lossless file.
Maybe this can help you in detecting what is "fake" lossless audio:
http://www.flickr.com/photos/60080664@N04/5485470134/ (http://www.flickr.com/photos/60080664@N04/5485470134/)
It's WAV file from an MP3 320kbps CBR source, and an original FLAC ripped from an original CD. Note that the fake WAV has a cut thoughout the track in 22kHz meaning it has retained the psychoacoustic tricks applied to its MP3 source. On the other hand, the original FLAC retains its full quality. This is probably the best example to know what is FAKE lossless file.
Plenty of legitimate releases roll off the treble below 22kHz. Some even 20kHz. Not usually completely - the difference should be visible with close inspection.
Looking at this the other way round, I wonder how low a bitrate mp3 I could pass off as lossless, by adding fake hi-frequencies etc?
Cheers,
David.
Looking at this the other way round, I wonder how low a bitrate mp3 I could pass off as lossless, by adding fake hi-frequencies etc?
The following spectrograph is from a processed 192kbps CBR MP3. I see no reason why I couldn't replicate this process with 128 or maybe even lower. If I were honestly trying hard to dupe someone, I'd fix the little bump at 16kHz, but meh, this works as proof-of-concept, and as concrete evidence why you should never trust a spectrogram to determine a transcode.
(http://i.imgur.com/zws69.jpg)
You
cannot use spectrographic analysis to
prove anything when it comes to lossy audio.
Not a guarantee, sure—and I’ve seen that too (and seen threads here about it)—but I would suppose that lossy-sourced tracks on CDs are relatively rare (perhaps very).
I bought a pressed, pirate CD sourced from MP3 off of eBay once. Orbital's Blue Album, FWIW. I got bit because there was an MP3 release that had certain tracks on it. I wanted the certain tracks, but the MP3 release was bogus. However, some enterprising soul over in Asia (I forget where) took that bogus release and pressed it. I mistook the eBay sale as a legit release and bought it. It was silver and silkscreened and everything! Not even a burn. The packaging had no catalog number info or anything.
Looking at this the other way round, I wonder how low a bitrate mp3 I could pass off as lossless, by adding fake hi-frequencies etc?
Well, as you know, a good way to please the spectrograph types is to use an aggressive LossyWAV setting on an mp3, lossy on top of lossy which adds plenty of stuff to "re-fill" the spectrum. Perhaps a strap-line for lossyWAV - "half the filesize of lossless compression AND it makes your spectrograph look even better than the original".
C.
Hi,
If you want to be sure about the quality of FLAC, why don't take a WAV file (rip a track from a CD to a WAV), encode that WAV to FLAC, and then decode the FLAC back to WAV again. Than do a bit-by-bit comparison of the 2 WAV files. They should be the same. FLAC is nothing but a compressor/decompressor like WinZip or something like that (except it's in the form of a codec).
If you download stuff (bittorrents etc...), you will never know what the source is, and you can not do a qualified judgement on it.
Regards,
Peter
After my (several and probably embarrassing ) failed attempts, what's the bottomline? Could we know if it is true or FAKE?
@thesurfingalien: That wasn't what he asked about. He wanted to know how you can tell if a lossless file had been encoded from a lossy file.
@SonicBoom!: The bottom line is that you can sometimes make a guess that has a fair chance of being right, but there is no sure-fire way to tell.
@SonicBoom!: The bottom line is that you can sometimes make a guess that has a fair chance of being right, but there is no sure-fire way to tell.
...and it's far easier to be able to say "this is sourced from lossy audio" (when there are obvious clues), than it is to say "this has never been lossy-ly encoded" (because someone could have hidden all the clues, or it was such high quality lossy that there were few clues to start with).
If you want to be sure about the quality of FLAC, why don't take a WAV file (rip a track from a CD to a WAV), encode that WAV to FLAC, and then decode the FLAC back to WAV again. Than do a bit-by-bit comparison of the 2 WAV files. They should be the same. FLAC is nothing but a compressor/decompressor like WinZip or something like that (except it's in the form of a codec).
If you download stuff (bittorrents etc...), you will never know what the source is, and you can not do a qualified judgement on it.
I think we're being charitable, and assuming that at least some people are interested in finding whether a legitimate source (CD, legally downloaded FLAC, etc) is truly lossless. In a few instances, it's clear they're not.
If someone downloads a lossless file illegally, and it sounds fine and looks fine, but they're still worried it might be a lossy version and this impacts their enjoyment, then they've largely created their own torture. I wonder who is putting out not-really-lossless lossless files, so good quality that you can't hear the problem, with all the clues removed? And if you can't hear the problem....?! Also, if a tree falls down in a forest with no one there, does it really make a noise?
etc
Cheers,
David.
If you want to be sure about the quality of FLAC, why don't take a WAV file (rip a track from a CD to a WAV), encode that WAV to FLAC, and then decode the FLAC back to WAV again. Than do a bit-by-bit comparison of the 2 WAV files. They should be the same.
But what if the CD was sourced from a lossy audio eg. MP3? Doing bit-by-bit comparison of the files would only result in "no differences" yet, it is from an MP3 source.
IMO, I think, well never know
I think we're being charitable, and assuming that at least some people are interested in finding whether a legitimate source (CD, legally downloaded FLAC, etc) is truly lossless. In a few instances, it's clear they're not.
If someone downloads a lossless file illegally, and it sounds fine and looks fine, but they're still worried it might be a lossy version and this impacts their enjoyment, then they've largely created their own torture. I wonder who is putting out not-really-lossless lossless files, so good quality that you can't hear the problem, with all the clues removed? And if you can't hear the problem....?! Also, if a tree falls down in a forest with no one there, does it really make a noise?
etc
Cheers,
David.
I accidentally found this thread while looking for a spectrum analyzer from which I could export the image - thanks to this very thread, I found Spectro!
More on topic, this is an excellent point (the latter paragraph, anyhow). If you can't hear the difference - and this may make audiophiles cringe - what's the problem? Clearly, our user-in-question can't pick out artifacts or anything by ear, so as far as he's concerned, it's transparent - for example, my family thinks I'm silly when I hear artifacts in songs that they don't. It doesn't dampen their happiness with a mix-CD in the slightest because of what
I hear, or see (when it comes to spectrograms) so to the owner/downloader, I fail to grasp how there is a problem here.
So what it really comes down to is: can you hear a difference between a legitimate master copy and your copy, or in this case, can you hear artifacts or any sound issues with which you are displeased?
but really don't exist a method to detec a FLAC that is make from transcoding an MP3?
@Antigen: Is there a method to know if someone is stupid just by looking at a (random) photograph of him/her? You are asking something alike.
Audio is audio, lossy encodings can have a signature which programs like aucdtect and audiochecker use to guess if it could be lossy. That's all it can be done.
When you see a low quality video like those made with older mobile phones, or some of those in youtube you know that it is lossy because you see lots of blocks and image distortion. With audio, the same can be done, but generally, audio is not encoded in such low quality so the signature is less obvious to these type of programs.
Maybe this can help you in detecting what is "fake" lossless audio:
http://www.flickr.com/photos/60080664@N04/5485470134/ (http://www.flickr.com/photos/60080664@N04/5485470134/)
It's WAV file from an MP3 320kbps CBR source, and an original FLAC ripped from an original CD. Note that the fake WAV has a cut thoughout the track in 22kHz meaning it has retained the psychoacoustic tricks applied to its MP3 source. On the other hand, the original FLAC retains its full quality. This is probably the best example to know what is FAKE lossless file.
Plenty of legitimate releases roll off the treble below 22kHz. Some even 20kHz. Not usually completely - the difference should be visible with close inspection.
Looking at this the other way round, I wonder how low a bitrate mp3 I could pass off as lossless, by adding fake hi-frequencies etc?
Cheers,
David.
Hi everyone.
I downloaded an album (that I owned on CD, that was destroyed mind you) that did just that. It had a dither applied to it and I could see using Q0Labs Frequency Analyzer that it was compressing on vocal passages and anything else that was not hard cymbals or drums. It was a decent encoding, but it fooled Audio Checker into thinking it was 100%CDDA. The album in question came out in 1988, so I am doubtful that the lossy encoding occurred in the studio.
I use Q0's software to verify FLAC. I use a Linear 1-22400hz at -120 to 0db to visually check to see if it is compressing anything. I agree with the rolloff as well. I have many albums and singles that roll off at 20Khz rather than 22Khz.
I read (maybe here) a few years ago some publisher's announced plan to issue CD's sourced from marginal lossy compressed as a form of copy protection. The theory was that the CD would sound good (or good enough), but if ripped and reencoded to 128 kb mp3 it would be noticeably sub-par.
...And then they realized that the music they want to protect is top-40 stuff and most of those people obviously don't care about sound quality whatsoever.