Skip to main content

Topic: Can you fake a hi-res track? (Read 3058 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • wakibaki
  • [*]
Can you fake a hi-res track?
Is the HF content of >RB sampled tracks sufficiently correlated to the musical content that it would be possible to tell if HF content from a different track were superimposed on an up sampled RB track?

Obviously you can't hear it. So I've seen people complaining about early high-res releases that had no HF content, but I believe this was discovered by other means than listening.

I could do some experiments myself, but I guess there are a lot of people more experienced at looking at tracks in Audacity than I. I guess that if you put more work into it you might get a more convincing result, but how much work? Could you write a program to do the work? Do we need to treat these things with as much suspicion as retouched digital photographs are now? If not now, then tomorrow?
wakibaki.com

Can you fake a hi-res track?
Reply #1
Is the HF content of >RB sampled tracks sufficiently correlated to the musical content that it would be possible to tell if HF content from a different track were superimposed on an up sampled RB track?

Obviously you can't hear it. So I've seen people complaining about early high-res releases that had no HF content, but I believe this was discovered by other means than listening.

I could do some experiments myself, but I guess there are a lot of people more experienced at looking at tracks in Audacity than I. I guess that if you put more work into it you might get a more convincing result, but how much work? Could you write a program to do the work? Do we need to treat these things with as much suspicion as retouched digital photographs are now? If not now, then tomorrow?



What is a >RB track?

In general the easiest way to add false HF content is to use a nonlinear process to create correlated HF content from the LF content.  An example of such a process would be a half-wave rectifier.

I'd take the track, half wave rectify it, high pass filter the rectified content above around 16-18 KHz, and mix it  back in with the LF content.

BTW AFAIK the relevant Aphex patents have run out.
  • Last Edit: 08 April, 2015, 08:03:22 PM by Arnold B. Krueger

  • mzil
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #2
There are various, common hardware devices such as "aural exciters" which add (mix in) extra, higher, artificial harmonics to the existing musical signal  you feed them. They are said to brighten, enhance, and liven up the sound. I even know of some mini stereos sold decades ago which had BBE or other analogous circuits which worked similarly. There's no reason they (or a software version) couldn't be modified to work in the ultrasonic range and even to forensic analysis I bet if they were to be used judiciously their presence would be completely undetectable, except in instances where the smart scientist might realize that the specific musical instruments being reproduced (having been already  analyzed previously in regards to ultrasonic content) simply have no such harmonics in the first place.

I don't think tubas have much ultrasonic content, for example  but yes, "hi-re$"  > 22k musical content could easily be faked and made to look like it was harmonically related .
  • Last Edit: 08 April, 2015, 08:13:22 PM by mzil

  • Wombat
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #3
It is not that easy as some think. In realtime FFT even well made upsamples can be spotted.
I once tried and the best result was to start creating fake content with aliasing by bad upsampling. There still may be the lowpass visible. So first downsample to the cutoff. This means if at 20kHz the LowRes material has a visible cutoff resample it to a samplerate of 40.000 first. After that upsample with a hefty aliasing setting like 96kHz. Take care to have not to much because it will look like mirrored.
You need some tool to create some fake harmonics out of this garbage.
Optimal would be some pure tape noise you have from one other source and mix it.
A harmonic recreator alone won't do it btw.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

  • saratoga
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #4
Is the HF content of >RB sampled tracks sufficiently correlated to the musical content that it would be possible to tell if HF content from a different track were superimposed on an up sampled RB track?


RB == redbook?

And yes, adding high frequency content back into a recording is very simple. Almost any modestly nonlinear process will do that, including many that are used in mastering (dynamic range compression for instance). 

  • Wombat
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #5
There are various, common hardware devices such as "aural exciters" which add (mix in) extra, higher, artificial harmonics to the existing musical signal  you feed them. They are said to brighten, enhance, and liven up the sound. I even know of some mini stereos sold decades ago which had BBE or other analogous circuits which worked similarly. There's no reason they (or a software version) couldn't be modified to work in the ultrasonic range and even to forensic analysis I bet if they were to be used judiciously their presence would be completely undetectable, except in instances where the smart scientist might realize that the specific musical instruments being reproduced (having been already  analyzed previously in regards to ultrasonic content) simply have no such harmonics in the first place.

I don't think tubas have much ultrasonic content, for example  but yes, "hi-re$"  > 22k musical content could easily be faked and made to look like it was harmonically related .

Ops, was typing while you answered.
Like you say i bet if you have some talented person someone can create a dedicated Exciter to do just that.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

  • wakibaki
  • [*]
Can you fake a hi-res track?
Reply #6
OK, what about the other way round? Reliably detect a fake?
wakibaki.com

  • Wombat
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #7
OK, what about the other way round? Reliably detect a fake?

The nature of a perfect fake is that you can't i guess. Add some hugglebass and it will sound better. I doubt anyone really hears what is going on above 20kHz no matter what is claimed.
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

  • saratoga
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #8
OK, what about the other way round? Reliably detect a fake?



Depends how well its done.  Simple harmonic extension will probably be detectable with good enough time/frequency or statistical analysis.  More complex methods will be much harder, especially if you don't suspect in advance what method was used.

Can you fake a hi-res track?
Reply #9
OK, what about the other way round? Reliably detect a fake?


Depends on guessing which technique was used and looking for its "footprint" by determining the relationship between the real and the fake parts of the track.

  • augustine
  • [*][*]
Can you fake a hi-res track?
Reply #10
I don't understand what you mean by hi res however I am going to have a crack at answering on the off chance I understand you correctly. Could you be more specific about sampling rate and frequency ranges ?  Basically the vast majority of adults can't hear much at all over 16 kHz so anything you do above that won't be noticed by anyone except :

1. Golden eared audiophiles cheating by staring at spectrogram.
2. Teenagers and children
3. My wife

So you could cut and paste sympathy for the devil above  say 17000  kHz and superimpose it on groove is in the heart and I bet a lot of people wouldn't notice. Mahler might be different....
If you want to make fake harmonic content then izotope rx 4 is amazing at that. Experienced person studying a spectrogram should be able to detect a fake.
  • Last Edit: 08 April, 2015, 10:07:26 PM by augustine

  • nastea
  • [*][*][*]
Can you fake a hi-res track?
Reply #11
What if someone would use a Pioneer cd player with Legato Link Conversion® to create fake High Resolution audio by recording the analog output to a computer's soundcard line-in @ 96KHz/24Bit, would that somehow be detectable?

  • saratoga
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #12
What if someone would use a Pioneer cd player with Legato Link Conversion® to create fake High Resolution audio by recording the analog output to a computer's soundcard line-in @ 96KHz/24Bit, would that somehow be detectable?


Thats some really old oversampling, harmonic extension technique right?  If so, see above.

  • augustine
  • [*][*]
Can you fake a hi-res track?
Reply #13
What if someone would use a Pioneer cd player with Legato Link Conversion® to create fake High Resolution audio by recording the analog output to a computer's soundcard line-in @ 96KHz/24Bit, would that somehow be detectable?


Depends on quality of ad converter. Most computer sound cards have shit ad converters on board so probably yes.
Also pioneer cd player ? It's da converter we be 44.1 soooo.....
  • Last Edit: 08 April, 2015, 10:42:11 PM by augustine

  • mzil
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #14
I'm pretty sure I could make a pretty much undetectable hi-re$ fake. Take some normal music where from spectrographic analysis you can see some instrument with a known, stronger than typical ultrasonic content [ex. crash cymbal] get brick walled at 20k or so due to mics, studio recorders, processors, or the 44.1 medium itself. Take this in analog form and re-record through an ultrasonic wideband mixer and  recorder, and when that loud, boisterous crescendo comes along, [you know, the one that  looks on your wideband spectrogram like it probably original had a small, low level tail off into the ultrasonic octaves but it got chopped off?], and then rattle your keys briefly into an ultrasonic capable mic and mix it in at a low but detectible level for a fraction of a second, to recreate that chopped off part! [Watch an RTA as you do this and mix accordingly.]
  • Last Edit: 08 April, 2015, 11:14:31 PM by mzil

Can you fake a hi-res track?
Reply #15
What if someone would use a Pioneer cd player with Legato Link Conversion® to create fake High Resolution audio by recording the analog output to a computer's soundcard line-in @ 96KHz/24Bit, would that somehow be detectable?



AFAIK the effects of legato link are too weak. The Placebo effect explains most of its perceived benefits.

I did some experiments with replacing the ultrasonic band of my keys jangling file with harmonics of a redbook-standard version of the file, as well as spectrally shaped random noise.

Sonically, it seems to make no difference - the ultrasonic band can be filled with silence, natural sound, generated harmonics, or noise because its inaudible anyway!

  • krabapple
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #16
I'm pretty sure I could make a pretty much undetectable hi-re$ fake. Take some normal music where from spectrographic analysis you can see some instrument with a known, stronger than typical ultrasonic content [ex. crash cymbal] get brick walled at 20k or so due to mics, studio recorders, processors, or the 44.1 medium itself. Take this in analog form and re-record through an ultrasonic wideband mixer and  recorder, and when that loud, boisterous crescendo comes along, [you know, the one that  looks on your wideband spectrogram like it probably original had a small, low level tail off into the ultrasonic octaves but it got chopped off?], and then rattle your keys briefly into an ultrasonic capable mic and mix it in at a low but detectible level for a fraction of a second, to recreate that chopped off part! [Watch an RTA as you do this and mix accordingly.]



Please tell me you're kidding and you don't think anyone would actually go to that trouble. 


Can you fake a hi-res track?
Reply #17
I'm pretty sure I could make a pretty much undetectable hi-re$ fake. Take some normal music where from spectrographic analysis you can see some instrument with a known, stronger than typical ultrasonic content [ex. crash cymbal] get brick walled at 20k or so due to mics, studio recorders, processors, or the 44.1 medium itself. Take this in analog form


Which unfortunately will have no ultrasonic content to speak of.


Quote
and re-record through an ultrasonic wideband mixer and  recorder,


These are the general run of things these days. Analog audio gear has by convention had response up to 50 or 100 KHz for decades. 2496 and 24192 recorders are hard to avoid if you are in the recorder market.

Quote
and when that loud, boisterous crescendo comes along, [you know, the one that  looks on your wideband spectrogram like it probably original had a small, low level tail off into the ultrasonic octaves but it got chopped off?],


They don't call them brick wall filters for nothing. Stop band rejection is no less than 90 dB, and often more, much more.

Quote
and then rattle your keys briefly into an ultrasonic capable mic and mix it in at a low but detectible level for a fraction of a second, to recreate that chopped off part! [Watch an RTA as you do this and mix accordingly.]


The good news is that you already have a lifetime supply of this kyes jangling  sort of thing in the form of my keys jangling file. ;-)

If you have CEP or Audition there is a tool called Edit, Mix, Paste, Modulate which can be used to create any harmonic of a signal that you wish for. For other editors YMMV. I seem to recall that Reaper has some interesting tools There are harmonic generator VST plug-ins for Audacity and other programs, and there are some stand alone harmonic generator programs.



  • drewfx
  • [*][*][*]
Can you fake a hi-res track?
Reply #18
I'm assuming we're not talking about audible stuff but technical analysis.

If I were going to either detect fake or fake ultrasonic content myself, I would probably divide it into two categories:

1. Stuff with a decaying overtone series that reaches into ultrasonic territory.
2. Stuff like cymbals and other instruments made of metal where the frequency content might not be predictably decaying before it starts getting filtered out.

Then I think it would be a matter of how much time I was willing to spend.

  • mzil
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #19
I'm pretty sure I could make a pretty much undetectable hi-re$ fake. Take some normal music where from spectrographic analysis you can see some instrument with a known, stronger than typical ultrasonic content [ex. crash cymbal] get brick walled at 20k or so due to mics, studio recorders, processors, or the 44.1 medium itself. Take this in analog form and re-record through an ultrasonic wideband mixer and  recorder, and when that loud, boisterous crescendo comes along, [you know, the one that  looks on your wideband spectrogram like it probably original had a small, low level tail off into the ultrasonic octaves but it got chopped off?], and then rattle your keys briefly into an ultrasonic capable mic and mix it in at a low but detectible level for a fraction of a second, to recreate that chopped off part! [Watch an RTA as you do this and mix accordingly.]
  Please tell me you're kidding and you don't think anyone would actually go to that trouble. 


I wouldn't be inclined to since I have no interest in promoting Hi-re$, nor am I a con artist trying to dupe the general public for my own gains, however for the growing segment of the audio industry with a strong financial interest in promoting Hi-re$ gear and downloads...

By doing this entirely in the analog domain, as I'm suggesting, it helps cover the tracks from forensic analysis looking for fakery. The ultrasonic component to the sound really is there in the analog sound, entering the microphone, before the music gets digitized by the Hi-re$ recorder, without a doubt; it isn't artificially generated by digital software/hardware manipulation and hence subject to leaving finger prints indicating foul play.
  • Last Edit: 09 April, 2015, 02:53:43 PM by mzil

  • mzil
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #20
The good news is that you already have a lifetime supply of this kyes jangling  sort of thing in the form of my keys jangling file. ;-)


The problem with using your well known recording of the keys, rather than starting fresh per recording with real, new, ultrasonic sound from keys or whatever, generated on the fly, is a careful analysis could show that the fakery's ultrasonic content matched, exactly, a small segment of another known ultrasonic recording out there, namely your keys jangling file which has been posted now in many forums.
  • Last Edit: 09 April, 2015, 03:09:49 PM by mzil

Can you fake a hi-res track?
Reply #21
The good news is that you already have a lifetime supply of this kyes jangling  sort of thing in the form of my keys jangling file. ;-)


The problem with using your well known recording of the keys, rather than starting fresh per recording with real, new, ultrasonic sound from keys or whatever, generated on the fly, is a careful analysis could show that the fakery's ultrasonic content matched, exactly, a small segment of another known ultrasonic recording out there, namely your keys jangling file which has been posted now in many forums.


With an audio editor and a little time - a lot less time than it would take to make a new recording, the existing recording could be edited and re-eqed to be something else.

However, the nonlinear distortion technique strikes me as having far greater potential for non-detectability.

  • mzil
  • [*][*][*][*][*]
Can you fake a hi-res track?
Reply #22
Non-detectability would be the whole point when making an audio forgery to promote some Hi-re$ scam, not how quickly the forgery can be accomplished. If some company like Meridian was exposed as having promoted a song to evaluate their Hi-re$ gear with which was actually just a fake they would have egg on their face.

Can you fake a hi-res track?
Reply #23
Non-detectability would be the whole point when making an audio forgery to promote some Hi-re$ scam, not how quickly the forgery can be accomplished. If some company like Meridian was exposed as having promoted a song to evaluate their Hi-re$ gear with which was actually just a fake they would have egg on their face.


That is true, but it is likely that a goodly proportion of classical recordings (particularly of small ensembles) that claim to be broadband, actually are. 24/96 recording is a standard feature of just about every pro digital recorder or converter sold in the last 10-15 years. These days you can wipe you butt with 24/192 converters.

While mics with flat response > 20 KHz  are not common, one can do a lot of classical recording with as few as two or three of them.  There are a fair number of wideband omnidirectional mics and a few cardoids. 

When you get into multi-micing such as is commonly used on popular and rock recordings, the costs can get kinda intimidating if you try to go 100% wideband.  However, most mics don't have response that is brick-walled, so even though they are rolling off  typically @ 12 dB/octave > 14-20 KHz, there is still something going on.  Also, if you put your few wideband mics on the percussion and the few brass instruments with lots of highs, that gives you genuine wideband content that gets mixed in with the rest. Also, you can put a direct box on the electronic instruments and bypass the mic response limits.  Even the cheapest direct boxes cruise right past 20 KHz and are still pretty flat up to say 30 or 40.
  • Last Edit: 10 April, 2015, 05:00:46 PM by Arnold B. Krueger

  • wakibaki
  • [*]
Can you fake a hi-res track?
Reply #24
I seem to recall that there were a couple of sites that were pulled up for selling supposedly hi-res material that was then discovered to have no HF content. This was a minor embarrassment in some quarters, particularly where the tracks had already received favourable reviews.

It just occurred to me that an unscrupulous owner might find their website business so jeopardised by such a discovery, that rather than replace their inventory, they might seek to 'rectify' the situation by some form of batch processing.

Not that I'd seek to spread FUD amongst hi-res enthusiasts that they're listening to undetectable frauds.
  • Last Edit: 10 April, 2015, 07:51:34 PM by wakibaki
wakibaki.com