It seems that a disappointing number of the "high res" soundtracks available from retailers like HDTracks, ProStudioMasters, 7digital, Pono, Qobuz, etc are just upconverted from 16/44 masters. The Dark Knight Rises and Star Wars: The Force Awakens are examples of scores with high res releases of questionable quality.

Anyway, I was hoping that someone could provide me with step by step directions on how to analyze these sorts of high res audio files for myself. As I understand it, it cannot always be proven when files have been upconverted, but there are tools out there that can provide strong evidence one way or another.

At the moment, I'm using a Windows 10 PC.

Star Wars: TFA is not fake.
The mastering makes it look fake, but it was in fact recorded at 192/24, or higher.
This simple spectrogram proves it. Already we see the normal frequencies extending well
beyond 22.5kHz. So it wasn't upsampled. You can note two things from this. One, the presence
of white noise/ambiance between 60kHz and 96kHz. This is normal for HD, SACD, DSD, anything
really. I am not sure what causes it, while it is probably a result of the equalization of the Mic itself to boost
higher frequencies, I am not sure about that. I tested my own 192kHz studio mic and it exhibits the same
thing. So these recordings definitely aren't "fake" in terms of how they were recorded.
This track is EXTREMELY quiet. Which leaves a lot of headroom for white noise and gain
during the mastering process. You can even see the dynamic compression thanks to this.

https://dl.dropboxusercontent.com/u/68272387/Currently%20Listening/Rey's%20Theme.png

I think what a lot of people are mistaking for being "fake" is the louder cues. Take "The Falcon" for example.
This cue is very loud all the way through. What I feel a lot of people are doing is taking the extention of higher
frequencies and calling them "fake" because they look unnatural. When in fact it is just the mastering that
made them this way. I could take Rey's Theme and make it look the same. Just boost the EQ after 30kHz
and you'd achieve the same effect. Why they chose to do this is also a mystery. Just another dumb studio
practice



I only own 2 High-res audio albums. TFA and Vivaldi's 4 Seasons. I can assure you that neither of them are "fake".
You have to take a lot into account when making such a judgement. There are too many unknown variables
such as: the quality of the equipment, the resonance of the recording hall, the dynamics of the orchestra, the
recording engineer's taste for new or old equipment, DSPs used, you can't just look at something and call it
fake because it doesn't look right to you.

This doesn't answer your question but hopefully it sheds some light.
I've not got TDK:R so I can't look at that.

EDIT:
Here, I took 5 seconds to remaster "The Falcon"



Now the higher frequencies look more natural.
It's not fake. Just poorly mastered. As is LITERALLY EVERYTHING.

---------- Post added at 10:24 PM ---------- Previous post was at 10:02 PM ----------

Also another thing to note, some people will say it looks fake because the frequencies aren't there and insetad
the mastering team just added "noise". You can see this process in Rey's Theme, anything past 20kHz is practically "silent"
but this is another mistake. In this case, the frequencies are THERE, they're just super quiet.

The mics can only pick up what they hear and, if the gain was turned way down or some shit thing like that, you could yell
into the mic and it wouldn't hear you. (Figuratively speaking). This is evident in Rey's Theme. If we use an equalizer to
bring up everything past 20kHz, you'll clealy see the frequencies extending up to 40kHz before fading off into white noise.
The white noise is present because of how quiet the orchestra was recorded. Talk as quiet as you can into your home mic
and then boost it with EQ or gain, what do you get? White noise.

Take a louder track, The Falcon, Scherzo for X-Wings, etc, it is the complete opposite. There is NO white noise because
they are so loud. The dynamic volume of the orchestra was immense. And the frequencies extend well beyond
where they need to be for a 192kHz recording.

Star Wars: TFA is not fake.
The mastering makes it look fake, but it was in fact recorded at 192/24, or higher.
This simple spectrogram proves it. Already we see the normal frequencies extending well
beyond 22.5kHz. So it wasn't upsampled. You can note two things from this. One, the presence
of white noise/ambiance between 60kHz and 96kHz. This is normal for HD, SACD, DSD, anything
really. I am not sure what causes it, while it is probably a result of the equalization of the Mic itself to boost
higher frequencies, I am not sure about that. I tested my own 192kHz studio mic and it exhibits the same
thing. So these recordings definitely aren't "fake" in terms of how they were recorded.
This track is EXTREMELY quiet. Which leaves a lot of headroom for white noise and gain
during the mastering process. You can even see the dynamic compression thanks to this.

https://dl.dropboxusercontent.com/u/68272387/Currently%20Listening/Rey's%20Theme.png

I think what a lot of people are mistaking for being "fake" is the louder cues. Take "The Falcon" for example.
This cue is very loud all the way through. What I feel a lot of people are doing is taking the extention of higher
frequencies and calling them "fake" because they look unnatural. When in fact it is just the mastering that
made them this way. I could take Rey's Theme and make it look the same. Just boost the EQ after 30kHz
and you'd achieve the same effect. Why they chose to do this is also a mystery. Just another dumb studio
practice



I only own 2 High-res audio albums. TFA and Vivaldi's 4 Seasons. I can assure you that neither of them are "fake".
You have to take a lot into account when making such a judgement. There are too many unknown variables
such as: the quality of the equipment, the resonance of the recording hall, the dynamics of the orchestra, the
recording engineer's taste for new or old equipment, DSPs used, you can't just look at something and call it
fake because it doesn't look right to you.

This doesn't answer your question but hopefully it sheds some light.
I've not got TDK:R so I can't look at that.

EDIT:
Here, I took 5 seconds to remaster "The Falcon"



Now the higher frequencies look more natural.
It's not fake. Just poorly mastered. As is LITERALLY EVERYTHING.

---------- Post added at 10:24 PM ---------- Previous post was at 10:02 PM ----------

Also another thing to note, some people will say it looks fake because the frequencies aren't there and insetad
the mastering team just added "noise". You can see this process in Rey's Theme, anything past 20kHz is practically "silent"
but this is another mistake. In this case, the frequencies are THERE, they're just super quiet.

The mics can only pick up what they hear and, if the gain was turned way down or some shit thing like that, you could yell
into the mic and it wouldn't hear you. (Figuratively speaking). This is evident in Rey's Theme. If we use an equalizer to
bring up everything past 20kHz, you'll clealy see the frequencies extending up to 40kHz before fading off into white noise.
The white noise is present because of how quiet the orchestra was recorded. Talk as quiet as you can into your home mic
and then boost it with EQ or gain, what do you get? White noise.

Take a louder track, The Falcon, Scherzo for X-Wings, etc, it is the complete opposite. There is NO white noise because
they are so loud. The dynamic volume of the orchestra was immense. And the frequencies extend well beyond
where they need to be for a 192kHz recording.

What program are those screenshots from? And can you remaster audio with it?

What program are those screenshots from? And can you remaster audio with it?

The screenshots were from a small program called Spec. Just shows spectrograms.
So no. You can't do anything with it other than look at pretty lines.

The screenshots were from a small program called Spec. Just shows spectrograms.
So no. You can't do anything with it other than look at pretty lines.

Oooooooo. :p

I was wondering anyway, what program did you remaster with? Or what is the best program to do so?

Oooooooo. :p

I was wondering anyway, what program did you remaster with? Or what is the best program to do so?

It really comes down to personal preference of feature list. GUI, usability,
I like to use Audacity personally. I use Nyquist prompt to do a lot of command-line
editing so it's good for that, it's a light program, nice and simple.

It really comes down to personal preference of feature list. GUI, usability,
I like to use Audacity personally. I use Nyquist prompt to do a lot of command-line
editing so it's good for that, it's a light program, nice and simple.

Oh, nice. I use Audacity for nearly everything (editing, composing), Adobe Audition for some things (reverb, etc).

Oh, nice. I use Audacity for nearly everything (editing, composing), Adobe Audition for some things (reverb, etc).

Audacity is a handy tool for sure. A little reverb can go a long way, too.

A little reverb? Personally, I go all out on GREAT HALL OF VALHALLA - MAXIMUM SETTINGSSSSSSSSSSS

A little reverb? Personally, I go all out on GREAT HALL OF VALHALLA - MAXIMUM SETTINGSSSSSSSSSSS

YESSSSSSSSSSS

Star Wars: TFA is not fake.
The mastering makes it look fake, but it was in fact recorded at 192/24, or higher.

(...)

Take a louder track, The Falcon, Scherzo for X-Wings, etc, it is the complete opposite. There is NO white noise because
they are so loud. The dynamic volume of the orchestra was immense. And the frequencies extend well beyond
where they need to be for a 192kHz recording.

I�m sorry, but I have to disagree. Star Wars: TFA is as fake as The Dark Knight Rises. Only difference: they tried to hide it.

You�re posting spectograms which prove exactly that: that those frequencies beyond 20 kHz are artificial. Real hypersonic frequencies are quite irregular; for example, a flute extends to something like 25-30 kHz (if it manages to stay above the noisefloor), percussion extends to 35-40 kHz, violins go up to 30 kHz. Besides those characteristic instrument frequency responses you�d see "dirt". Like residues causes by imprecise sampling, sines caused by CRT monitors on stage. The latter is visible - barely - on every TFA-release at precisely 15.734 Hz (to pick an example, Track 2 "The Scavenger" contains it). If it would be an actual HiRes recording, you�d see another sine at 31.468 Hz. Yet it�s completely absent. And not because they filtered it, only because it never was there in the first place.

Back to those characteristic frequency responses: the frequency response that IS there above 20 kHz on TFA is almost completely a straight line from 20 kHz to 96 kHz (192 kHz sampling rate) and its gain rises in perfect coordination with the gain of the music. To add insult to injury: every instrument, whether it�s a violin, percussion, brass or woodwinds, add the same kind of response. Not one instrument in the world behaves like that, not one microphone in the world behaves like that. I�ve never seen any HiRes recording exhibiting this odd behaviour. BUT: Exciters (https://en.wikipedia.org/wiki/Exciter_(effect)) behave like that.

An exciter is a DSP that creates artificial frequencies based on material that�s already there (fundamental frequencies). It takes this material (which is stuff we can hear, so depending on our age, from 20-20.000 Hz), synthesizes new stuff using 2nd and 3rd order distortions and adds these distortions to the original signal. This is called "harmonic synthesis" and this is what they did here.

Another explanation is that they used something called "Spectral band replication" which is a technique usually found in recent lossy codecs, to be specific, their low-bitrate variants (HE-AAC, for example). There, noise-like signals are added, based on transposing up harmonics from lower and midfrequencies.

And did I mention that you can hear what they have been faking? You�d need to apply a steep High Pass filter at 20 kHz (to erase anything below 20 kHz), then perform resampling from 192 to 44.1 kHz WITHOUT any aliasing filter (iZotope RX will allow you to do that - but please, never ever do that for other than scientific reasons!). With all the frequencies beyond 20 kHz folded back as aliasing into the audible passband you will then be able to hear those frequencies. And guess what, they sound like quantization distortions. Which would be... noise. Not what you would have on a real HiRes recording.

^ ^ ^

To add to this, the reason people are looking at spectrographs to spot a "fake" recording instead of listening to it is because human hearing cannot discern signals higher than between 18 and 22khz, making "high resolution" audio a completely academic exercise.

What we can and can't hear aside, the "high resolution" editions of TFA we've seen released here at FFshrine, whatever their source, are fake - unquestionably so. I would be prepared to testify as much in a court of law. I don't doubt for one moment that the original recording was made at a higher resolution, but the source material used for these uploads was clearly 44.1hz, or to put it another way, a CD rip. From 0 to about 22khz, you have real signal. Then you have 70khz-odd of noise. Somebody has taken a low-resolution recording, done *something* (almost certainly what SonicAdventure described above) to create new frequencies outside the range of our hearing spectrum.

Folks, it's noise - plain and simple. It's electronically-created distortion that somebody put there because they knew some smart-ass would put a spectrograph scan on the internet and they also knew that seeing "some" signal up in those parts would be enough to convince a large proportion of people that they're hearing genuine high resolution audio.

The scan that Zaralyth has posted as proof that the "high resolution" download is real, is in fact proof that it is not.

To add to this, the reason people are looking at spectrographs to spot a "fake" recording instead of listening to it is because human hearing cannot discern signals higher than between 18 and 22khz, making "high resolution" audio a completely academic exercise.

Well, I too looked at several spectograms. And analyzed the frequency with a plain old spectrum analysis... but then, I have a bit of experience and know when I see something odd. But I have to - partly - disagree about so called HiRes (proper terms: lowered noisefloor / extended frequency response) as I�ve found that it has its purpose when using DSPs with it.


What we can and can't hear aside, the "high resolution" editions of TFA we've seen released here at FFshrine, whatever their source, are fake - unquestionably so. I would be prepared to testify as much in a court of law. I don't doubt for one moment that the original recording was made at a higher resolution, but the source material used for these uploads was clearly 44.1hz, or to put it another way, a CD rip. From 0 to about 22khz, you have real signal. Then you have 70khz-odd of noise. Somebody has taken a low-resolution recording, done *something* (almost certainly what SonicAdventure described above) to create new frequencies outside the range of our hearing spectrum.

Just like you, I think that the original sessions were something like 24/176.4. Shawn Murphy did something similar all the way back to 'Munich', why should he have abandoned it within 10 years? I can only guess at the reasons why a fake HiRes release was given to retailers... but the one that�s most probable is that only one master was created: 24/44.1 or 24/48. From that master everything else was derived, with the HiRes version created to squeeze some more bucks out of batshit crazy audiophiles (and they are pretty daft, as they don�t understand technology at all).

Well, I too looked at several spectograms. And analyzed the frequency with a plain old spectrum analysis... but then, I have a bit of experience and know when I see something odd.

No disagreement here; my point was that if there WERE genuinely audible benefits to "high resolution" audio, pretty much anybody would be able to immediately tell what was fake and what wasn't by simply listening to it. The fact that we have to resort to staring at spectrograph scans before we can arrive at any reasonably concrete conclusions, I think, proves the nature of the con.

Absolutely; there is much to be gained by making your initial recording at a higher resolution and bit depth. With a lowered noise floor, you can set your levels a little lower whilst still capturing the full dynamics of whatever you're recording; so if a musician or a singer takes you by surprise with a particularly loud sound, it's not going to destroy your recording - you can sort it out later on in post production. Capturing higher frequencies is less useful although if you're in the business of restoration, the wider response can help you (and your noise reduction tools) discern noise from signal. As a rule, if you're doing extensive DAW work on a recording, you want to work on the highest resolution, highest bit-depth files you can get your hands on.

None of this has anything to do with the way the audio sounds; these advantages are all related to the practicalities of production.

At the end of the production line, after all the fancy work has been done, you're left with the distribution format; and there is absolutely no earthly reason why this should be above CD quality, ever. You can't hear the higher frequencies. For the end user, "high resolution" audio is, at best, merely the reason why seventy minutes of music requires 4gb of storage space instead of 300mb. There are theories that whilst the higher frequencies are inaudible, their mere presence in the playback chain could adversely affect sound quality because the reproduction hardware is trying to play them back; ie, a speaker which is trying to reproduce ultra-high frequency noise and distortion could very well end up distorting its reproduction of frequencies within the hearing spectrum as a result.

Any reproduction system that offers absolutely NO AUDIBLE BENEFITS WHATSOEVER, but can multiple the storage space requirements by a factor of up to ten and even has the potential to actually harm sound quality... in my book, it's a no-brainer...


Just like you, I think that the original sessions were something like 24/176.4. Shawn Murphy did something similar all the way back to 'Munich', why should he have abandoned it within 10 years? I can only guess at the reasons why a fake HiRes release was given to retailers... but the one that�s most probable is that only one master was created: 24/44.1 or 24/48. From that master everything else was derived, with the HiRes version created to squeeze some more bucks out of batshit crazy audiophiles (and they are pretty daft, as they don�t understand technology at all).

I suspect that the original sessions are made at a higher resolution; if not as high as 176.4 (sometimes chosen ahead of 192 for recordings that are explicitly intended for CD release because it's an easy multiple of 44.1 which makes it easy to fold down to CD audio standards) then certainly 88.2 or 96.

I think you're bang on the money as far as your "one master" theory; the CD and these "high resolution" downloads sound identical in every way. Sometimes a "nicer" mix will be made for the "high resolution" release or the vinyl mix - again, leading consumers to associate CD with poorer sound.

No disagreement here; my point was that if there WERE genuinely audible benefits to "high resolution" audio, pretty much anybody would be able to immediately tell what was fake and what wasn't by simply listening to it. The fact that we have to resort to staring at spectrograph scans before we can arrive at any reasonably concrete conclusions, I think, proves the nature of the con.

^^^^


Absolutely; there is much to be gained by making your initial recording at a higher resolution and bit depth. With a lowered noise floor, you can set your levels a little lower whilst still capturing the full dynamics of whatever you're recording; so if a musician or a singer takes you by surprise with a particularly loud sound, it's not going to destroy your recording - you can sort it out later on in post production. Capturing higher frequencies is less useful although if you're in the business of restoration, the wider response can help you (and your noise reduction tools) discern noise from signal. As a rule, if you're doing extensive DAW work on a recording, you want to work on the highest resolution, highest bit-depth files you can get your hands on.

That�s why everything I do is done within 32 bit floating point at 192 kHz. Upsampled, of course, using special filter designs ;)

About that musician / singer uttering a loud sound: you could repair (interpolate) him / her with iZotope's Declipper, of course. That one works well with 0 dBfs distortions. But apart from that, yes, using high bit-depths you increase the headroom (or footroom, as it were).


At the end of the production line, after all the fancy work has been done, you're left with the distribution format; and there is absolutely no earthly reason why this should be above CD quality, ever. You can't hear the higher frequencies. For the end user, "high resolution" audio is, at best, merely the reason why seventy minutes of music requires 4gb of storage space instead of 300mb. There are theories that whilst the higher frequencies are inaudible, their mere presence in the playback chain could adversely affect sound quality because the reproduction hardware is trying to play them back; ie, a speaker which is trying to reproduce ultra-high frequency noise and distortion could very well end up distorting its reproduction of frequencies within the hearing spectrum as a result.

You know, not too long ago I used to actually believe in this crap. Until I started to read, test, measure and listen. But I still think that HiRes offers a tiny, tiny advantage. Even when all obstacles are removed. The advantage is about as big (small) as 2-4%: negligible and almost inaudible. To find something similar to compare the effect to: differences between several playback devices is bigger. I�ve also found that it yields a slight advantage to work with genuine HiRes files instead of upsampled ones. But that�s about it.


Any reproduction system that offers absolutely NO AUDIBLE BENEFITS WHATSOEVER, but can multiple the storage space requirements by a factor of up to ten and even has the potential to actually harm sound quality... in my book, it's a no-brainer...

Not only the potential, in lots of cases it does harm audio, I�ve measured it myself. Usually with audiophile boutique hardware... because companies producing those have A. a certain kind of customer B. not the best (or experienced) engineers C. often love tubes. I�m certain that many differences audiophiles usually perceive between 16/44.1 & 24/192 are caused by A. their belief (placebo, anyone?) and B. by badly engineered hardware.


I suspect that the original sessions are made at a higher resolution; if not as high as 176.4 (sometimes chosen ahead of 192 for recordings that are explicitly intended for CD release because it's an easy multiple of 44.1 which makes it easy to fold down to CD audio standards) then certainly 88.2 or 96.

Though today it wouldn�t make sense anymore to choose 176.4 because it can be resampled easily to 44.1. Resamplers have gotten so good during the last 10 years, it�s incredible. I recommend SoX, it�s one of the best and even free.

---------- Post added at 12:26 PM ---------- Previous post was at 12:20 PM ----------

Oh, I almost forgot, I made a HiRes file myself!

I turned one piece of Goldsmith's Legend into true and genuine HiRes music ;)



This is the track "Darkness Fails" and as you hopefully all know, there doesn�t exist any HiRes source. The only mastertape in existence is an early digital tape, made long before HiRes was developed. Looks not exactly like TFA but I might be getting there with a bit of tweaking and the use of an exciter. This isn�t an exciter, but I also won�t tell how I did it as people would copy the process to do fakes themselves.

About that musician / singer uttering a loud sound: you could repair (interpolate) him / her with iZotope's Declipper, of course. That one works well with 0 dBfs distortions. But apart from that, yes, using high bit-depths you increase the headroom (or footroom, as it were).

Yeah, there are ways of fixing it... but if you're recording at 24 or 32 bit, you can just set low levels and be sure. Once you've got your live recording set up, and you think your levels are right, the concert starts and suddenly your levels are crap. People play louder on the night than they do in rehearsal. Everything is done in such a tearing rush these days, there isn't the time (or the money) to test stuff properly. You just throw in as many microphones as you can get your hands on, record a hundred tracks, and sort out the mess later on when all the expensive musicians have gone home and you're no longer occupying a large studio.

I've seen it done. I attended a recording session a few years ago with the BBC National Orchestra of Wales for Doctor Who and got a taste for it. They have minutes to get set up. You don't get any real rehearsal time. The engineers turn up and blanket the stage with microphones, the orchestra turn up, sit down, open the music, the engineer hits "record", and off they go. If the take isn't great they do another one. If THAT take isn't good either they make do and mend, unless it went catastrophically wrong - but most of the time they move on to the next one and try to fix whatever went wrong in post production.

When I was there, I moaned at the recording engineer and waxed lyrical about the good old days. "Why do you need thirty microphones when Decca used to use three?"

The poor guy agreed with me, but pointed out what I didn't really realise at the time; to get it right with three microphones, you need to think carefully about where you're putting them, and you also need to give the orchestra and conductor time to do their balance... all of which takes time, and time is money, and nobody has either commodity these days. You get in, you do it as quickly as possible, and you get out. The live mix we heard in the control room wasn't BAD... but it wasn't going to fool anybody on an album or in the show. Back in the day, guys like Eric Tomlinson were making live mixes that could literally go straight to the vinyl cutter and straight onto the shelves of record shops. Try working that out today. Not gonna happen.


But I still think that HiRes offers a tiny, tiny advantage. Even when all obstacles are removed. The advantage is about as big (small) as 2-4%: negligible and almost inaudible.

May I ask what advantage you see, and what is your scientific corollary? I do not see how it could be given the inherent limitations of our species' auditory system. Our perception of sound ends 2000hz short of CD, and 16-bit audio provides for 96db of dynamic range which, whilst it doesn't quite exceed our perception (around 110-120db) it does exceed the dynamic range of ANY recording ever made and ANY playback system by a comfortable margin. (Perceiving a dynamic range of 120db is akin to the quietest sounds being completely inaudible except in a specially treated room and the loudest sound causing irreversible hearing loss in the listener). A live orchestral concert will not push 60-70db of range - more than covered by 16-bit audio - and 99.999% of commercial recordings will average about 10-20db of range, if you're lucky.

From whence does the benefit come? (I'm talking about benefits for the end listener - I don't dispute the usefulness of higher bit depths and higher sampling rates during production.)


Usually with audiophile boutique hardware... because companies producing those have A. a certain kind of customer B. not the best (or experienced) engineers C. often love tubes.

Yes, the certain kind of customer with a ludicrous amount of money, not much sense, and a fanatical belief that the thing with the higher number (decibels, bits, kilohertz, ���...) must by definition be better.


I�m certain that many differences audiophiles usually perceive between 16/44.1 & 24/192 are caused by A. their belief (placebo, anyone?) and B. by badly engineered hardware.

Not one reliable blind study has ever been publicised in which people can genuinely tell the difference between 16/44.1 and anything higher. The studies that claim to have done so are either sighted studies (ie, the listener knows which sample is which) or technically flawed in some way - the 16/44.1 sample is from a different mastering, volume matching has not taken place, or the sample has been intentionally crippled by the act of being derived from the 24/192 with faulty or misconfigured resamplers. Or the hardware is misconfigured leading to colouration of one of the sample types.

I have done blind tests on myself; I can sometimes discern 320kbps MP3 from FLAC depending on the source, but I can NEVER discern 16/44.1 from 24/192. Sometimes, I think I can, but the results of every single test I have ever taken reveal the truth; I'm guessing.

This reminds me of another similar blind test undertaken a few years ago with violins: The old accepted wisdom is that you can't beat a Stradivarius. A group of respected A-list violinists were asked to play (blind) a variety of violins from various sources and spot the Strad. None of them could. The success rate was around 50% - ie, they were guessing. Of course, after the results were in the violinists came up with a variety of nebulous reasons to "explain" their failure to distinguish between pricessless Stradivarius instruments and brand new machine-crafted instruments... and if you spent �20,000,000 on a 300 year-old violin and then realised that you could spend �1,000 on a brand new one and nobody could tell the difference... you would be thinking of reasons too.

http://blogs.discovermagazine.com/notrocketscience/2012/01/02/violinists-cant-tell-the-difference-between-stradivarius-violins-and-new-ones/#.Vo5Th1km_-s

The placebo is very simple; people need to feel like there is a benefit. This is especially pronounced if they've just spent a small fortune on hi-fi equipment (you can buy a turntable that costs �200,000; for the same money you could get a small house in London, or FOUR Jaguar F-Types) - if you've just bought something insanely expensive the placebo effect will kick in to convince you your money was well spent, so you don't feel like a prat. There is also the "keeping up with the Joneses" effect; if you can hear a difference and that stupid charlatan over there can't... then you, by definition, must be cleverer or otherwise superior in some way. How many times have you explained, using science and common sense, that "high resolution" audio is a con, and received a variation of the following snooty responses?
a1) Oh, well, I suppose my ears must be that much better than yours!
a2) My ears are musically trained and I have a Doctorate in music technology, so I'm not surprised you can't hear a difference.
b) My equipment is very expensive, and yours isn't, so that's why you can't hear a difference.
c) You're just saying that because you're an idiot who can't appreciate the finer things in life / you're poor and you can't afford a fantastic set of speakers like mine / you're jealous because you can't hear the difference, but I can.

If you're REALLY lucky, you'll get...

d) Well, I know the science, but I trust my ears and to me, the difference is like night and day, so there.

When asked to provide some scientific basis for their claims, those same people will a) get agitated and annoyed, b) get defensive, c) insult you, d) change the subject, or e) all of the above.


Though today it wouldn�t make sense anymore to choose 176.4 because it can be resampled easily to 44.1. Resamplers have gotten so good during the last 10 years, it�s incredible. I recommend SoX, it�s one of the best and even free.

SoX is absolutely fantastic.

I think some of the "older school" still find the mathematics of downsampling 176.4 to be comforting; any equation that works out to something-decimal-point-recurring is bad. That will produce noise... but 176.4 divide 4 = 44.1 is nice and fluffy and straightforward and pure...


Oh, I almost forgot, I made a HiRes file myself!

I turned one piece of Goldsmith's Legend into true and genuine HiRes music ;)


That is utterly, utterly priceless.

:rofldata:

Yeah, there are ways of fixing it... but if you're recording at 24 or 32 bit, you can just set low levels and be sure. Once you've got your live recording set up, and you think your levels are right, the concert starts and suddenly your levels are crap. People play louder on the night than they do in rehearsal. Everything is done in such a tearing rush these days, there isn't the time (or the money) to test stuff properly. You just throw in as many microphones as you can get your hands on, record a hundred tracks, and sort out the mess later on when all the expensive musicians have gone home and you're no longer occupying a large studio.

(...)

The poor guy agreed with me, but pointed out what I didn't really realise at the time; to get it right with three microphones, you need to think carefully about where you're putting them, and you also need to give the orchestra and conductor time to do their balance... all of which takes time, and time is money, and nobody has either commodity these days. You get in, you do it as quickly as possible, and you get out. The live mix we heard in the control room wasn't BAD... but it wasn't going to fool anybody on an album or in the show. Back in the day, guys like Eric Tomlinson were making live mixes that could literally go straight to the vinyl cutter and straight onto the shelves of record shops. Try working that out today. Not gonna happen.

Not that the Decca tree is always the best solution ;)

But it explains why many releases made during recent years lack quality. It mirrors what I see in mastering. Instead of preparing a release so that it will play well on many systems (which is, after all, what mastering is (or was?) supposed to do), releases are tailored to a specific playback device. And more often than not it only includes limiting peaks. Instead of altering dynamics or frequency response to accomodate different playback devices. When I "remaster" I always listen to the result on several devices, ranging from a fully blown surround system over a car stereo to an ancient radio in the kitchen, looking for dynamic issues, mono compatibility, etc. (I need to do that because I do all mastering with headphones). I guess, no one has the time to do that anymore... or no one is paid the money necessary.

And I�m envious that you attended a recording session for Doctor Who.

A year ago I was able to hear a live mix of an orchestra and compare it to the finalized mix (which I got to master). It sounded well enough but didn�t differentiate as much as I would have liked it to and some instrument groups were too soft. Back then, I assumed that many different microphone feeds were to fault, an assumption that might have been correct.

And are you a fan of the late Eric Tomlinson? Me too. But what about Alien or Star Wars: Episode IV? I�d call them hideous.


May I ask what advantage you see, and what is your scientific corollary? I do not see how it could be given the inherent limitations of our species' auditory system. Our perception of sound ends 2000hz short of CD, and 16-bit audio provides for 96db of dynamic range which, whilst it doesn't quite exceed our perception (around 110-120db) it does exceed the dynamic range of ANY recording ever made and ANY playback system by a comfortable margin. (Perceiving a dynamic range of 120db is akin to the quietest sounds being completely inaudible except in a specially treated room and the loudest sound causing irreversible hearing loss in the listener). A live orchestral concert will not push 60-70db of range - more than covered by 16-bit audio - and 99.999% of commercial recordings will average about 10-20db of range, if you're lucky.

From whence does the benefit come? (I'm talking about benefits for the end listener - I don't dispute the usefulness of higher bit depths and higher sampling rates during production.)

You�re right. You may have noticed, that I didn�t write that the tiny advantage directly translates into an advantage for your average listener (because it doesn�t) ;)


Yes, the certain kind of customer with a ludicrous amount of money, not much sense, and a fanatical belief that the thing with the higher number (decibels, bits, kilohertz, ���...) must by definition be better.

Tell me about it.


Not one reliable blind study has ever been publicised in which people can genuinely tell the difference between 16/44.1 and anything higher. The studies that claim to have done so are either sighted studies (ie, the listener knows which sample is which) or technically flawed in some way - the 16/44.1 sample is from a different mastering, volume matching has not taken place, or the sample has been intentionally crippled by the act of being derived from the 24/192 with faulty or misconfigured resamplers. Or the hardware is misconfigured leading to colouration of one of the sample types.

Are you referring to Meyer/Moran by any chance? Because objectivists abuse it for their purposes when they shouldn�t. Have you ever read the published paper? And have you ever wondered why the authors published a follow-up paper (that finally included a list of the media played back)? Because: some questions were left unanswered in their first paper. Which they should have been made aware of when they forwarded their paper to AES to be peer-reviewed. Peer-review means that several fellow scientists look for flaws in a scientific paper before it is published. To this day, I wonder why the AES allowed it to be published in the form it was published in. Now, I don�t want to bash their methodology, I would have done it exactly the same way; with one crucial difference: I would have selected musical examples myself. Their error was allowing participants (mostly your common audiophiles) to bring their favourite recordings on a HiRes-capable medium. But... roughly 60% of the music played during the trials was plain old 16/44.1, blown up by the lables to be re-released for HiRes. To make matters worse, a smaller part was derived from 24/48 or 16/48 while only a very small part was either genuine 24/96 or DSD (most of the time derived from analogue sources of course... and you know how much crap you have above 20 kHz, crap that isn�t remotely composed of instruments' overtones). That is what you get for trusting audiophiles: they will just believe what the labels tell them to. Cattle.

On the other hand, subjectivists usually cite the Hypersonic study by Oohashi et al. Which they shouldn�t. Just like with Meyer/Moran, the methodology is sound. Their conclusions however are not. To find evidence for hypersonic signals influencing humans, they studied brainpatterns of the participants. All very good... the problem was their interpretation of the patterns they were scrutinizing. The field of neuroscience is still in its infancy and despite making huge progress during the last 40 years, results need to be analyzed by experienced people who know what they�re looking at. What Oohashi and his team concluded, could have been easily caused by other things than the hypersonic effect. So, their conclusion is worthless. Another problem is that the study has not been repeated successfully. All pretty suspicious.

You see, I don�t fully trust either objectivists or subjectivists. Both are two sides of the same coin. And every side will try to crush the other and they will abuse everything to their respective advantage that comes their way, even if isn�t suitable. Both sides are examples of faith, yes, some sort of audio-based religion. Objectivists believe in numbers and reliable data telling the truth, subjectivists believe that only their ears tell the truth. In short:

Objectivists: fact-faith
Subjectivists: faith-faith

Naturally, I favour objectivists. Because anything audio is living science (physics and math). Music in its most common form (digital bits) is data, digits waiting to be processed. Playback / recording devices are engineered, thereby adhering to scientific principles. Conclusion: science is the only way to properly explain them (measurements, yay!).


I have done blind tests on myself; I can sometimes discern 320kbps MP3 from FLAC depending on the source, but I can NEVER discern 16/44.1 from 24/192. Sometimes, I think I can, but the results of every single test I have ever taken reveal the truth; I'm guessing.

I have done blind tests myself, too. When I could still hear the results, that is. 10 years ago I performed several blind tests to find out if 24/96 was offering an advantage. Back then, true 24/96 was hard to find, but, depending on the material, I did have statistically significant results (before you ask, I did it the proper way by first downsampling, then upsampling). A few months ago, I repeated some of those tests to find that I cannot hear a difference anymore. Or that I simply use better resamplers now. Or that my ears have gotten worse over the years. Well, I age as much as everyone, even though I won�t dare to admit it :D

And if you can discern mp3 from flac, try aac or opus. I can promise you that you will have a hard time hearing any difference at all.


This reminds me of another similar blind test undertaken a few years ago with violins: The old accepted wisdom is that you can't beat a Stradivarius. A group of respected A-list violinists were asked to play (blind) a variety of violins from various sources and spot the Strad. None of them could. The success rate was around 50% - ie, they were guessing. Of course, after the results were in the violinists came up with a variety of nebulous reasons to "explain" their failure to distinguish between pricessless Stradivarius instruments and brand new machine-crafted instruments... and if you spent �20,000,000 on a 300 year-old violin and then realised that you could spend �1,000 on a brand new one and nobody could tell the difference... you would be thinking of reasons too.

http://blogs.discovermagazine.com/notrocketscience/2012/01/02/violinists-cant-tell-the-difference-between-stradivarius-violins-and-new-ones/#.Vo5Th1km_-s

The placebo is very simple; people need to feel like there is a benefit. This is especially pronounced if they've just spent a small fortune on hi-fi equipment (you can buy a turntable that costs �200,000; for the same money you could get a small house in London, or FOUR Jaguar F-Types) - if you've just bought something insanely expensive the placebo effect will kick in to convince you your money was well spent, so you don't feel like a prat. There is also the "keeping up with the Joneses" effect; if you can hear a difference and that stupid charlatan over there can't... then you, by definition, must be cleverer or otherwise superior in some way. How many times have you explained, using science and common sense, that "high resolution" audio is a con, and received a variation of the following snooty responses?
a1) Oh, well, I suppose my ears must be that much better than yours!
a2) My ears are musically trained and I have a Doctorate in music technology, so I'm not surprised you can't hear a difference.
b) My equipment is very expensive, and yours isn't, so that's why you can't hear a difference.
c) You're just saying that because you're an idiot who can't appreciate the finer things in life / you're poor and you can't afford a fantastic set of speakers like mine / you're jealous because you can't hear the difference, but I can.

If you're REALLY lucky, you'll get...

d) Well, I know the science, but I trust my ears and to me, the difference is like night and day, so there.

When asked to provide some scientific basis for their claims, those same people will a) get agitated and annoyed, b) get defensive, c) insult you, d) change the subject, or e) all of the above.

Here is another one:

e) Blind tests are a start, not the end

But sadly, all of what you�ve written above about blind tests and people reacting to it, is true. The golden-ear-myth alone gets me every time and infuriates me. The arrogance! Fuck, I�m guilty of arrogance too, so scrap that. However, one cannot ignore the flaws of blind tests. Or, better said, the flaws objectivists tend to ignore because it doesn�t fit their agenda (and they have one, despite all their assurances to the contrary - again, just like the subjectivists):

a) every blind test is only valid for that particular setting (your ears and brain, setup, material reviewed, circumstances, hardware) and you cannot extrapolate universal truths using the results from one pass. But you can calculate a certain likeliness for an event to repeat / not repeat itself. If some setup is repeated often enough under different circumstances, then one can say the results are valid enough to have a general meaning. For objectivists, even one pass at something is enough to "prove" something now and forever (just look at hydrogenaudio). Which is pretty much not scientific.

b) blind tests elicit a stress reaction from participants. Which might hide or obscure any result you want to find out. Sadly, this often is one of the objections used by subjectivists to deride them. Stupid. I once tested my partner. I prepared two files, one was the original, the other was the same file where I applied a 1.5 dB boost of frequencies around 1.000 Hz. After listening for about half an hour using foobar, my partner was exhausted and never wanted to repeat something like that.

c) there actually is a difference between trained and untrained listeners. My partner couldn�t hear any difference with the setup described above. Even though blind testing determined that we can discriminate a change in frequency response as small as 1 dB. My partner isn�t interested in audio at all, so no trained ears there. I could hear the difference... easily. Even though I suck at pinpointing a particular frequency responsible (I tested myself using the Philips Golden Ear Challenge and was shocked how much I need the visual feedback of spectrum analysis).

Please, remember that the results above are in no way a general statement about things regarding my partner. I would need to repeat them with other people to have reliable results.

I�ve found that blind tests are fantastic to gain a quick impression, to make a usable assessment of audio quality in a short amount of time. I�ve also found that more differences will show if I give it time, don�t do a listening test comparison and don�t listen consciously to, say, a device I want to judge. To conclude this paragraph: sighted tests are useless, blind tests are a valid method, long term but unconscious exposure reveals differences independent of opinion and marketing hype.

You see, I don�t fully trust either objectivists or subjectivists. Both are two sides of the same coin. And every side will try to crush the other and they will abuse everything to their respective advantage that comes their way, even if isn�t suitable. Both sides are examples of faith, yes, some sort of audio-based religion. Objectivists believe in numbers and reliable data telling the truth, subjectivists believe that only their ears tell the truth.

I think I object to the word "faith" because it implies something that cannot be backed up with empirical evidence. It is a belief you hold despite evidence suggesting that belief is invalid. You can't have faith in facts, because facts are facts - you don't need to believe in a fact for it to be a fact - lack of belief in a fact is ignorance but does not change the shape or nature of the fact. Truth is fact, and fact is truth. Folk who push "fact" as just another word for "belief" are invariably trying to discredit facts in order to credit beliefs, or to put it another way, they've got more bias than an open reel tape.

From my perspective, I won't try to crush the opposition because I want to be right, or because I want to prove them wrong; I don't care about one-upmanship. I care about what's true. A fact trumps a belief every single time. There are some things I believe but I can't verify; I believe that tea tastes better when you add the milk to the tea instead of vice versa. I believe my uncle's wife is a hideous and disgusting affront to the human species. A true scientist will ALWAYS be ready to change his beliefs in the face of evidence. The evidence directs his beliefs. Audiophiles have a belief and then look for evidence to corroborate it - they will misread, misquote, selectively quote, and quote unreliable sources until they have "evidence" to support their beliefs; or they will withdraw immediately back to "faith" - a state of mind in which one voluntarily choses to disregard evidence and go with their beliefs.

In some cases, I find blind tests to be pointless because they are usually an attempt to prove that which doesn't need to be proven because it's already factual. IE, the statement "sample rates above 44.1khz are pointless for human listeners" is factually correct and 100% verifiable. It doesn't require testing or proving. It's a statement about the nature of our physiology; unless you're not a human being, that statement applies to you and to me and to everyone. But people come along who want to ignore the science, ignore the common sense, ignore the facts, and suddenly we have to introduce a testing methodology to verify it. (And "verify it" is code for "give credence to baseless assumptions made by ignorant people".)

In the case of high resolution audio, a blind test does nothing but provide you with another "I told you so" which the true audiophile will nonetheless ignore. All it has to "prove" is that in a controlled environment, a human subject cannot identify audible differences between 44.1/16 audio and higher resolution audio assuming a) both versions come from the same master, b) the high resolution master is genuine, c) the 44.1/16 version was made properly and with transparent resampling algorithms, and d) the playback equipment is working properly and performs identically over a diverse range of sample rates.


And are you a fan of the late Eric Tomlinson? Me too. But what about Alien or Star Wars: Episode IV? I�d call them hideous.

Tomlinson is my audio God, in the film scoring world anyway - in the classical world it's the old Decca guys all the way. As far as Alien and Episide IV, I consider both to be fine recordings which have unfortunately suffered at the hands of time or idiotic people at the controls during the remastering - Star Wars infinitely moreso, which is ironic because I find Star Wars to be the superior recording. The room sound is electric. You can listen to that score, close your eyes, and your brain can almost create imagery of the recording session. How else can you explain the Star Wars scores starting off sounding pretty incredible on vinyl and getting progressively worse in every CD release? I think Alien sounds quite nice in the Intrada; not perfect, and not anywhere NEAR what was actually recorded at the sessions, but better than we need to appreciate Jerry's score.

My dream is to build a time machine and somehow get into the sessions of every Tomlinson-engineered score and quietly connect up the desk output to a digital recorder. Failing that, I'd go back in time and somehow convince the owners of the master tapes to treat them right. Master tapes from classical music recordings as early as the late 1940s survive in good to excellent condition - all these sad stories about lost masters, destroyed masters, unplayable masters could have been avoided just by treating those tapes with respect... but like the producers of Doctor Who, by the time they realised the cultural value on those reels, it was too late and the damage had been done.


And if you can discern mp3 from flac, try aac or opus. I can promise you that you will have a hard time hearing any difference at all.

In my ABX tests, with AAC and Opus all bets are off ninety-nine times out of one hundred, although there are still occasional "killer samples" that give the game away - even at high bitrates. By far the majority of times my results are clearly guesswork. Lossy codecs are at an unprecedented state of maturity and although they are less useful in the read world than they once were, the fact that a lot of people find hard to stomach is that lossy codecs are now transparent - obviously dependent on the codec under discussion, the type of sound under discussion, and the encoder itself. (Nobody is going to tell me that MP3s from 1996, encoded with BladeEnc at 128kbps, are transparent. Even my 88 year-old Grandad can ABX them reasonably well, and he wears a hearing aid in each ear. But good bitrate encodes of AAC, or Opus, or Ogg, or even MP3 using a modern encoder? Quite probably transparent in the majority of cases.)


...releases are tailored to a specific playback device. (...snip...) I guess, no one has the time to do that anymore... or no one is paid the money necessary.

Sadly, that's a factor which never changes - in the seventies recordings were mastered so they sounded pretty good on crappy plastic record players. Today, they're mastered so they sound pretty good on Apple earbuds.


Are you referring to Meyer/Moran by any chance?

I'm afraid I don't remember the specifics of the study, but I do remember that even the most cursory examination of it unmasked it as a fraud. All sorts of shenanigans went on that compromised its validity and it is clear that the testers were biased in favour of high resolution audio.


I have done blind tests myself, too. When I could still hear the results, that is. 10 years ago I performed several blind tests to find out if 24/96 was offering an advantage. Back then, true 24/96 was hard to find, but, depending on the material, I did have statistically significant results (before you ask, I did it the proper way by first downsampling, then upsampling). A few months ago, I repeated some of those tests to find that I cannot hear a difference anymore. Or that I simply use better resamplers now. Or that my ears have gotten worse over the years. Well, I age as much as everyone, even though I won�t dare to admit it

I used to have a sound card which would introduce all manner of distortions and strange sounds (in the audible spectrum) whenever you sent it audio higher than 48khz. You could've experienced something of the same phenomenon, or something in the same ballpark - you're right, in 2005 genuine high resolution audio was hard to come by, but equally (if not more) rare at the time was was hardware that could play it back properly. A lot of old sound cards resampled all output on the fly; so in your early blind test, the 24/96 sample may well have been through some garbage processor and you wouldn't have even known. You could have also experienced what I discussed before; if a speaker or a pair of headphones is actively trying to reproduce those high frequencies (ie, it is responding physically to the sound) that could colour its reproduction of the audible spectrum.

There are too many potential variables; comparing the results of a test ten years ago wth the results of a test today, when so much about the original test is unverified or forgotten to time. I know that when I ABX myself I rarely document anything. I couldn't tell you what I used to resample with, what I used to playback, what samples I used, etc.

That said, it's absolutely true - as we age our hearing deteriorates. We don't really notice it on a day to day basis, but some of those frequencies we used to be able to hear as children are now lost to us forever. Luckily as humans we can get away with losing more than half of our high frequency hearing and still function perfectly well.

(Edit 22:50 GMT to expand a few points and correct some silly spelling mistakes.)

Best thread ever! Really interesting discussion and no one is yelling... yet.

I have only ever bought 3 hi-res albums in my life, 2 were SACD (Star Trek Nemesis OST and "Yo-Yo Ma plays John Williams") and one was DVD-A (a Mozart album) about 15 years ago. Besides the 5.1 surround on the DVD-A, my ears could not hear any other improvements so I decided not to spend my money on hi-res audio anymore. The surround on the DVD-A sounded odd as well, I don't think I like sitting in the middle of the orchestra, I just want to be in front of it. So, if I can't hear any of the benefits why waste my money. I'm not going to stare at spectograms on my computer, CD quality is just fine for me and my ears. In fact, as my ears get older it is becoming increasingly harder to hear the difference between iTunes Plus/256 mp3/CDs so, I've actually started buying more stuff through iTunes/Amazon. This blasted age thing! At some point, having good speakers will not even matter because I won't be able to hear them anyway!

I think I object to the word "faith" because it implies something that cannot be backed up with empirical evidence. It is a belief you hold despite evidence suggesting that belief is invalid. You can't have faith in facts, because facts are facts - you don't need to believe in a fact for it to be a fact - lack of belief in a fact is ignorance but does not change the shape or nature of the fact. Truth is fact, and fact is truth. Folk who push "fact" as just another word for "belief" are invariably trying to discredit facts in order to credit beliefs, or to put it another way, they've got more bias than an open reel tape.

That is too easy, I think. The aggressiveness with which the two groups discuss these matters, points to something else than just empirical evidence on the side of objectivists. Subjectivists are able to elicit pure rage in objective people and the fact that they�re able to, evokes strong shades of religious fear in me. I�ve witnessed myself the viciousness with which scientists attack each other, they even perform studies only to prove their point. It all boils down to "I�m right and you�re not. And I back up my righteousness with empirical data and fact. You do too? Then your data / conclusion / interpretation is wrong" - and so it goes back and forth. Yet they all learned the ways of science - which should in theory extinguish this behaviour. But the belief is there: I have the better numbers, so my belief is correct - and I can prove it. It�s a mixture of who�s got the bigger ego and the better numbers. I�ve seen it many times in academia, and I hated it. To me it was just another tepid belief system bolstered by the assumed appearance of objectivity. And I continue to see it on, for example hydrogenaudio where everyone who even smells remotely like a subjectivist, is attacked immediately (truth to be told, why audiophiles continue to be there when the intentions of the forum are clearly known, is beyond me). My point is, they have decided it�s time for war.

Subjectivists have decided the same, believing in the mighty power of their ears, the brittleness of jitter and the fiery sound vomited by tubes, all the while ignoring that their brain plays the crucial part in their hearing. The consequence? An objectivist will be attacked the minute he enters stevehoffman.forum or computeraudiophile (and why he would enter there, is beyond me, too). I don�t like the atmosphere in both circles. Audiophiles are loony and delusional, objectivists should get the stick out of their arse. And both sides should get their nose down.

Please, tangotreats, excuse my harsh wording, it really isn�t directed at you. I hope I explained it why I�m so fed up with these discussions led by these people. To me, this is all religious faith (even fundamentalistic to an extent) and if there�s one thing I abhor, it�s religion. People could be working together marvellously if they wouldn�t hold their respective belief systems in front of them like a monstrance.

And I don�t mistake fact for belief (fact can lead to belief), nor am I biased like open reel tape (you�re 31 and know about bias? Who are you??) :D


From my perspective, I won't try to crush the opposition because I want to be right, or because I want to prove them wrong; I don't care about one-upmanship. I care about what's true.

Then you are a good person and a welcome change to what I�ve experienced before.


A fact trumps a belief every single time.

Yes!


There are some things I believe but I can't verify; I believe that tea tastes better when you add the milk to the tea instead of vice versa. I believe my uncle's wife is a hideous and disgusting affront to the human species. A true scientist will ALWAYS be ready to change his beliefs in the face of evidence. The evidence directs his beliefs. Audiophiles have a belief and then look for evidence to corroborate it - they will misread, misquote, selectively quote, and quote unreliable sources until they have "evidence" to support their beliefs; or they will withdraw immediately back to "faith" - a state of mind in which one voluntarily choses to disregard evidence and go with their beliefs.

I�d sure want to meet your aunt! :D

A true scientist will always be ready to change his opinion in face of new evidence, true. But how many true scientists are out there, people who are able to erase their ego from the equation? And even if there are some, when their results will be published, how many people will use them to misrepresent them?

And I have problems to see this as black & white. My experience is that audiophiles start by looking for explanations, not evidence, to support their belief. And when they find that nothing can be explained (which happens often) they resort to the loony-bin stuff. Only then it gets ridiculous.

I want to see the good in people. Scientists and audiophiles alike are looking for explanations. Scientists have developed valid methods to do so, aimed at having significant impact by being based on empirical data, accuracy and repeatability. Audiophiles lack those instruments (better: don�t accept them) and that�s why they so often fail. In my eyes, audiophiles are people with too much money and not enough brains (knowledge).


In some cases, I find blind tests to be pointless because they are usually an attempt to prove that which doesn't need to be proven because it's already factual. IE, the statement "sample rates above 44.1khz are pointless for human listeners" is factually correct and 100% verifiable. It doesn't require testing or proving. It's a statement about the nature of our physiology; unless you're not a human being, that statement applies to you and to me and to everyone. But people come along who want to ignore the science, ignore the common sense, ignore the facts, and suddenly we have to introduce a testing methodology to verify it. (And "verify it" is code for "give credence to baseless assumptions made by ignorant people".)

If you would write "sample rates above 44.1 kHz are probably pointless for the majority of human listeners" I would agree immediately. You see, science is not that simple, it�s a lot of shades of grey and wording is important with statements like these, I think. Otherwise people will misunderstand. Thinking about it, there�s a lot going on right now in that field, several studies have been published, pointing towards a possible influence of aliasing filters (and I don�t mean Meridian Audio's attempt) which might explain why extended frequency responses might sound different. Lots of "mights"... and it might all come down to botched data or biased researchers. Time and repeated research will tell.


In the case of high resolution audio, a blind test does nothing but provide you with another "I told you so" which the true audiophile will nonetheless ignore. All it has to "prove" is that in a controlled environment, a human subject cannot identify audible differences between 44.1/16 audio and higher resolution audio assuming a) both versions come from the same master, b) the high resolution master is genuine, c) the 44.1/16 version was made properly and with transparent resampling algorithms, and d) the playback equipment is working properly and performs identically over a diverse range of sample rates.

Yes, just look at the crap they sell as High End audio hardware! 44.1 kHz material perfectly played back, but 96 kHz material fucked up beyond belief. Transparent performance with several sample rates is something I�ve experienced to be pretty rare. And transparent resampling algorithms? Not with the most commonly used ProTools. And I don�t think I�ve ever encountered a master that was the same for 96 kHz or 44.1 kHz releases.

But does that make the claim that 96 kHz sounds different any less true? After all, truth lies in the eyes of the beholder, however deluded that person may be. And the sad fact right now is that the majority of 96 kHz releases sound indeed different - for all of the above reasons. I fear we have to accept this situation for the time being, the HiRes train is running at full steam. "We must join with him, Gandalf. We must join with Sauron."


Tomlinson is my audio God, in the film scoring world anyway - in the classical world it's the old Decca guys all the way. As far as Alien and Episide IV, I consider both to be fine recordings which have unfortunately suffered at the hands of time or idiotic people at the controls during the remastering - Star Wars infinitely moreso, which is ironic because I find Star Wars to be the superior recording. The room sound is electric. You can listen to that score, close your eyes, and your brain can almost create imagery of the recording session. How else can you explain the Star Wars scores starting off sounding pretty incredible on vinyl and getting progressively worse in every CD release? I think Alien sounds quite nice in the Intrada; not perfect, and not anywhere NEAR what was actually recorded at the sessions, but better than we need to appreciate Jerry's score.

Uh oh, you will not like what I will write now ;)

Tomlinson could be fantastic, yes. But he could also be pretty bad. And I�ve not yet heard a Star Wars score by him that sounds good to my ears, whatever the version. It doesn�t help that I don�t like the music for the first three movies. Besides, I would not let any vinyl version near me. Vinyl playback is the most unreliable method ever invented for mass distribution and I was so glad when I was able to buy CDs. An aging pick-up, needle, wrong cartridge alignment... the list goes on and on. All add up to a more or less severe frequency error mixed with distortions and incredibly bad channel separation. Just like HiRes, people love vinyl for all the wrong reasons.

Alien sounds quite nice, indeed. But only on my version ;)

Have I ever shared Supergirl here? It was easy to restore Tomlinson's sound from the botched Silva CD.

My favourite recordings of his would be Final Conflict and Night Crossing (the '94 release).


My dream is to build a time machine and somehow get into the sessions of every Tomlinson-engineered score and quietly connect up the desk output to a digital recorder. Failing that, I'd go back in time and somehow convince the owners of the master tapes to treat them right. Master tapes from classical music recordings as early as the late 1940s survive in good to excellent condition - all these sad stories about lost masters, destroyed masters, unplayable masters could have been avoided just by treating those tapes with respect... but like the producers of Doctor Who, by the time they realised the cultural value on those reels, it was too late and the damage had been done.

I�ll join you.

I didn�t know about the tapes. Is their condition that bad?


In my ABX tests, with AAC and Opus all bets are off ninety-nine times out of one hundred, although there are still occasional "killer samples" that give the game away - even at high bitrates. By far the majority of times my results are clearly guesswork. Lossy codecs are at an unprecedented state of maturity and although they are less useful in the read world than they once were, the fact that a lot of people find hard to stomach is that lossy codecs are now transparent - obviously dependent on the codec under discussion, the type of sound under discussion, and the encoder itself. (Nobody is going to tell me that MP3s from 1996, encoded with BladeEnc at 128kbps, are transparent. Even my 88 year-old Grandad can ABX them reasonably well, and he wears a hearing aid in each ear. But good bitrate encodes of AAC, or Opus, or Ogg, or even MP3 using a modern encoder? Quite probably transparent in the majority of cases.)

Have you seen what people sometimes use here? LAME 3.92. It�s 13 years old! And mp3 is so good these days... all of my music is stored as mp3 on my smartphone and my portable players (for simple compatibility reasons). Except the stuff that came from lossy sources, those are obviously stored in flac.


Sadly, that's a factor which never changes - in the seventies recordings were mastered so they sounded pretty good on crappy plastic record players. Today, they're mastered so they sound pretty good on Apple earbuds.

Really?? And I thought I was exaggerating things.


I'm afraid I don't remember the specifics of the study, but I do remember that even the most cursory examination of it unmasked it as a fraud. All sorts of shenanigans went on that compromised its validity and it is clear that the testers were biased in favour of high resolution audio.

It was the other way around: Meyer/Moran tried to find any evidence for or against such claims. Audiophiles hate this study with a vengeance and they try to attack the methodology. They shouldn�t, it�s perfectly sound (with the exception I mentioned before).


I used to have a sound card which would introduce all manner of distortions and strange sounds (in the audible spectrum) whenever you sent it audio higher than 48khz.

Soundblaster? I had one, too. Reliable, not very transparent, full of fancy stuff no one needs.


You could've experienced something of the same phenomenon, or something in the same ballpark - you're right, in 2005 genuine high resolution audio was hard to come by, but equally (if not more) rare at the time was was hardware that could play it back properly. A lot of old sound cards resampled all output on the fly; so in your early blind test, the 24/96 sample may well have been through some garbage processor and you wouldn't have even known.

Hehe, that�s why I wrote that I did it properly ;). I confirmed beforehand that my external sound card was genuinely capable of 96 kHz playback. Resampling wasn�t an issue since I used Kernel Streaming with foobar, the soundcard followed that protocol obediently. And, well, I measured it. Naturally, all DSPs were switched off, I would be bad at what I do if I would constantly use things that do things without me knowing what they do when.


You could have also experienced what I discussed before; if a speaker or a pair of headphones is actively trying to reproduce those high frequencies (ie, it is responding physically to the sound) that could colour its reproduction of the audible spectrum.

I used a HD-580 and then a HD-600. Both are free of distortions with any material, except at low frequencies with a sound pressure of more than 100 dB. Pretty easy to drive, with gain being the main problem, they need so much power. The only thing I�m unsure about to this day is the headphone amp inside the sound card. Back then, you weren�t able to buy decent amps and the weak dwarf I had in my own sound card might have produced distortions just by working at its limits.


There are too many potential variables; comparing the results of a test ten years ago wth the results of a test today, when so much about the original test is unverified or forgotten to time. I know that when I ABX myself I rarely document anything. I couldn't tell you what I used to resample with, what I used to playback, what samples I used, etc.

My methodology hasn�t changed ;) Well, the music and the hardware has changed but the rest stayed the same. A few months ago I still compared the genuine 96 kHz file against an upsampled 96 kHz file that was derived from a downsampled version (44.1) of the original 96 kHz file.

---------- Post added at 03:40 AM ---------- Previous post was at 03:32 AM ----------


Best thread ever! Really interesting discussion and no one is yelling... yet.

I love discussing with Tangotreats. In fact, I�ve missed people like him here on this forum.


I have only ever bought 3 hi-res albums in my life, 2 were SACD (Star Trek Nemesis OST and "Yo-Yo Ma plays John Williams") and one was DVD-A (a Mozart album) about 15 years ago. Besides the 5.1 surround on the DVD-A, my ears could not hear any other improvements so I decided not to spend my money on hi-res audio anymore. The surround on the DVD-A sounded odd as well, I don't think I like sitting in the middle of the orchestra, I just want to be in front of it. So, if I can't hear any of the benefits why waste my money. I'm not going to stare at spectograms on my computer, CD quality is just fine for me and my ears. In fact, as my ears get older it is becoming increasingly harder to hear the difference between iTunes Plus/256 mp3/CDs so, I've actually started buying more stuff through iTunes/Amazon. This blasted age thing! At some point, having good speakers will not even matter because I won't be able to hear them anyway!

For decades, I used to have the belief (there we have it: I was a believer - going to kick my ass for this stupidity) that HiRes really offered a difference. Subsequently, I�ve bought countless (50, 60, 70? Don�t know) SACDs over the years. Most of them contain genuine DSD / PCM recordings, how to copy them to the PC so that I can work on them? And DSD is such a pointless thing... And I won�t buy a 1st Gen PS3 for them, not for something that might, if at all, be a 2% increase in quality. Besides, I don�t hear that anymore so I�m in the same boat as you. Some things seem to become unimportant once you get older. I for one am so glad about it... I was so anal about these things ( people would think that I still am :D), if I could go back in time to visit my younger self, I would slap him in the face. So you have been the decent one, I have been the loony one.

Man, this is going to be an absolute hell to read. Also, is there a program that converts a 88.2 kHz to a 44.1 kHz file? Now hear me out. The album I'm trying to convert (Random Access Memories) pretty much doesn't qualify in the sense that there is a lot of empty space in the file. So much so that nothing would be lost if I down converted it. I would love to show an image, but the website isn't letting me.

In My opinion, the format of choice would be 20bit~48kHz max for listening. 16bit~48khz would be fine with me too though even at my age I can here the difference between 24bit~96khz and 16bit~44.1khz. But the advantage with 48khz is the Noise Floor, and really anything above that for storage, portable or not is insane to the end listener. The Industry used to know that 20bit~48khz is the max needed for archive as it offers the maximum perceived format. 24 or 32bit and 176.4 or higher is necessary to edit as Sonic said for headroom to not damage the file during mastering. It's easy to add noise to higher frequency but the TFA HiRes samples in my opinion are artificial, added, not Mastered, at least not professionally. Either from lack of know how or being stingy with the release of true bit depth. But that said, anyone with the know how can take a 16bit 44.1khz Lossless file and do a true remaster just as is the case with the industry itself. And there's nothing wrong with that as long as they know what their doing. The proof in in the pudding, the sound itself.

Man, this is going to be an absolute hell to read. Also, is there a program that converts a 88.2 kHz to a 44.1 kHz file? Now hear me out. The album I'm trying to convert (Random Access Memories) pretty much doesn't qualify in the sense that there is a lot of empty space in the file. So much so that nothing would be lost if I down converted it. I would love to show an image, but the website isn't letting me.

Wow, this sounds good. I�m afraid, Random Access Memories did not enter my consciousness when it was released. And it is a genuine HiRes album, it seems. Dirt produced by samplers (aliasing residue) or old synths can be seen easily. This album probably is a true 24 bit album with a genuine expanded frequency response. That empty space you talk about is probably caused by a low pass applied at 32 to 35 kHz, perhaps done to avoid intermodualtion distortion caused by high levels of ultrasonic sound (one of the dangers of HiRes Tango and I talked about).

To answer your question: SoX & foobar. You�ll need foobar2000 (https://www.foobar2000.org/) and SoX (zip-archive) (https://www.hydrogenaud.io/forums/index.php?act=attach&type=post&id=7330).

SoX settings should look like this:



The passband setting determines the steepness of the aliasing filter. Set to 90% it removes a bit more treble content at the "benefit" of impulses audiophiles are so mad about. Since I mostly export lossless originals to lossy formats for portable use, it is a good thing to remove those frequencies anyway. Standard setting is 95%, you can leave it at that. You could tweak impulses further by altering the phase response. 0% would create files with no pre-ringing but lots of post-ringing AND a phase error (-> don�t do it ;)). Quality should always be set to "best", that way, SoX uses 32 bit floating point calculation.

---------- Post added at 01:52 PM ---------- Previous post was at 01:33 PM ----------


In My opinion, the format of choice would be 20bit~48kHz max for listening.

Yes, 24/48 is more than enough. In that case one could go with mp3 or any other lossy codec, provided they will be decoded with floating point precision.


16bit~48khz would be fine with me too though even at my age I can here the difference between 24bit~96khz and 16bit~44.1khz.

Can you? ;)


But the advantage with 48khz is the Noise Floor

Sorry, but 48kHz has nothing to do with the noisefloor.

Bitdepth = noisefloor
Sample rate = frequency response



The Industry used to know that 20bit~48khz is the max needed for archive as it offers the maximum perceived format

Did they use to know that? When? Where? I�ve never heard anything like this. Take the Grammy foundation (which attempts to preserve music for future generations) (https://www.grammy.org/grammy-foundation/preservation): for a digitization of an analogue master, they will not accept anything below 96 kHz. In case of a digital master, they will take the original - and if that�s 24/48, they don�t have another choice.


It's easy to add noise to higher frequency but the TFA HiRes samples in my opinion are artificial, added, not Mastered, at least not professionally. Either from lack of know how or being stingy with the release of true bit depth.

On the contrary, they knew exactly what they were doing. It was a decision by suits and Patricia Sullivan had to take the can for them - and while her masters can on occasion be pretty bad (if she�s paid enough, she�s wonderful though), she perfectly knows what she�s doing. Those distortions won�t be audible for anyone older than 30 (as they are starting within audible range).

You right on the Bitdepth, I had it reversed. As far as Sampling and Bitdepth 20bit~48khz was determined to be the maximum needed for archiving. Anything higher is unnecessary. And yes I've tested copies between 24~96 and 16~44.1 and I could hear a difference, as you said a few percent. They could have used 20bit 48khz for the CD medium but most people can't tell the difference.

Genuine blind tests, with identical test material produced from the same master, the high-resolution version verified as real, the lower-resolution version properly prepared, no DSPs of any kind, and playback equipment tested and verified as providing playback at all sample rates and bit depths without distortion?

Yes, but very minute. Mainly frequency. Not enough to care about. Like said above, not worth worrying about.

so some things up for tfa: stick to the cd :)

is the 24bit adding anything compared to the 16bit of the cd release of tfa?

so some things up for tfa: stick to the cd :)

is the 24bit adding anything compared to the 16bit of the cd release of tfa?

Yes. If it�s genuine, it has a lower noisefloor. Not that anyone would hear it. 16 bit has a noisefloor as low as -96 dB; the minimum level of most orchestral releases stays 50 dB above it - and that is already very, very soft. 24 bit is something around -144 dB. Well, noisefloor is a bit misleading. Below those values (-96 & -144 dB) any signal simply disappears.

The 16 bit release of TFA was properly dithered so you�d have the -144 dB performance encapsulated within a 16 bit package anyway.

Doing everything SonicAdventure said in his post resulted in the files looking like the frequencies were cut in half when using Spek. Was this supposed to be the correct result?

I�m sorry to retract my earlier statement(s): Star Wars: The Force Awakens is actually a genuine HiRes recording. Sort of.

I made an error in my judgement and I�m most likely confusing people further. But I�d hate to hide the fact I misrepresented facts, even though my recommendation will remain the same: use the CD / 16/44.1 download.



On the track "The Scavenger" you can see some dirt in the inaudible ultrasonic frequency band, specifically, some sines at roughly 45 kHz (left channel, circled). Then you can see transients typical for some instruments reaching up to 50 kHz (right channel, circled). You can even see points where material was edited / crossfaded, some further sines at roughly 63 kHz.





The characteristic frequency response of this particular instrument in the same track ("The Scavenger", right channel, circled) is a dead giveaway of a genuine HiRes recording.





"The Falcon", percussion. Up to 40 kHz you have genuine high frequency content (right channel, circled), above you have faked high frequency content.




So TFA is a mixture of genuine and faked ultrasonic content. I can only speculate about the reasons. The faked content was probably added afterwards to what was a well engineered recording (the fact that you have genuine ultrasonic content reaching up as far as 60 kHz, tells you everything about the quality of the technical equipment and the care (and time) with which it was used). The only reason I can think of is that they did it for effect, like, making it audible even for listeners who don�t hear ultrasonic content (about 99.99% of listeners) so that it screams HiRes HiRes HiRes!!! - but that�s about all I can come up with.

Again, I apologize for the confusion. Still, I recommend the CD / 16/44.1 download. The distortions visible in the third picture start within the audible frequency band, they might reach further down so in order to listen to undistorted audio, don�t go for the HiRes version.

So, Zaralyth, you were right! If I�m allowed to say this: well spotted :)

---------- Post added at 12:57 PM ---------- Previous post was at 12:52 PM ----------


Doing everything SonicAdventure said in his post resulted in the files looking like the frequencies were cut in half when using Spek. Was this supposed to be the correct result?

It should look like this:



Please note that the samplerate is at 44.100 Hz now (circled).

I've literally been beating myself up because I felt so stupid for being wrong but you think I might actually be right? Hmm, well, I'm not an expert (yet) and it was always clear that when you look at it the peaks were definitely extended on tracks like the falcon but to be acknowledged by an amazing editor in chief such as yourself, thank you it actually means a lot to me to hear that I am partially correct in my hypothesis..

in regards of the original trilogy in 192khz.... looking at the spectrum i dare to say it's FAKE 192khz. yes, it might contain noise up to 192khz, but do we really care about noise? i don't and hence me reducing the files down to 96khz seams to be logical step. dropping almost 3gb of pure noise (garbage) is a good thing.

in regards of the original trilogy in 192khz.... looking at the spectrum i dare to say it's FAKE 192khz. yes, it might contain noise up to 192khz, but do we really care about noise? i don't and hence me reducing the files down to 96khz seams to be logical step. dropping almost 3gb of pure noise (garbage) is a good thing.

Let's take a look, shall we.. Here is Imperial March



To quote SonicAdventure..


"The Falcon", percussion. Up to 40 kHz you have genuine high frequency content

Even for 30 year old transfers, I beleive these are genuine, but I'll leave that decision up to SonicAdventure. A lot of the frequencies are lost to the "noise floor" after 40-50kHz

and down converting to 96khz doesn't show any kind of cut-offs

I boosted the higher frequencies with equalization and you can see that the sonic frequencies extend --- genuinely --- up to 60kHz. But at this point they become to quiet and are just overpowered by white noise..

I've literally been beating myself up because I felt so stupid for being wrong but you think I might actually be right? Hmm, well, I'm not an expert (yet) and it was always clear that when you look at it the peaks were definitely extended on tracks like the falcon but to be acknowledged by an amazing editor in chief such as yourself, thank you it actually means a lot to me to hear that I am partially correct in my hypothesis..

No reason to beat you up. I mean, with this thread I�ve proven without a doubt that I make errors, so there you go. I�m not so high and mighty right now :)

---------- Post added at 11:17 PM ---------- Previous post was at 11:14 PM ----------


Even for 30 year old transfers, I beleive these are genuine, but I'll leave that decision up to SonicAdventure. A lot of the frequencies are lost to the "noise floor" after 40-50kHz

Darling, this is not my decision to make ;)

I can only do the same as you: look at spectograms (and listen to it). And I think - so far - that these are genuine 96 kHz recordings. Not 192 kHz but still - IMO - it wasn�t necessary to re-release them using HiRes. I have only listened to "Main Title" and "Asteroid Field" from Episode V so I cannot make any statement about the others.

---------- Post added at 11:32 PM ---------- Previous post was at 11:17 PM ----------

And right now I�m listening to the 1997 re-release of 'Empire Strikes Back'. Now with the re-release of the original OST in HiRes I might have the key I�ve needed for years to try and remaster it. But I don�t know. The RCA release sounds as if different people remixed it. Odd.

I boosted the higher frequencies with equalization and you can see that the sonic frequencies extend --- genuinely --- up to 60kHz. But at this point they become to quiet and are just overpowered by white noise..


i see... okay, i stand corrected and it's TRUE 192khz. but i still think it's okay to down-res to 96khz, since the extreme high frequencies are buried in the white noise.