24-192 a scam?

I don't really have a valid opinion as I've never sat down and done a blind test between 24/16.

I've read a lot of comments online about people saying the 24 bit version sounds better but they fail to take into consideration the 24 bit version may be a different master.  It also fails to take into consideration the very real chance of bias.

I'd find a controlled test more informative than a hundred individual opinions.

Seen that, read it; when considering music rather than test tones, it's technically flawed.

Flawed in what way?

Unfortunately I don't have the references to hand (or the time to get them just now), but here is a summary:

1   Crest factor:   Music signals from some instruments can have a crest factor in excess of 20dB.

2   Temporal discrimination:  The human ear brain combination can break the Fourier limit and achieve temporal discrimination of uncorrelated signals below 10μs (for some people this can be as little as 4μs).

3   Sensitivity to correlated / uncorrelated signals:  The noise patterns from quantisation noise are a mixture of correlated and uncorrelated noise.  The human ear-brain combination can distinguish these signals, so does not integrate the quantisation noise in the way the author suggests as a possibility.

I have given the details of the references to the relevant academic works in a previous thread.

Interesting, Huge.  What are you arguing?  That the human ear does exceed the parameters of 16/44?  I'm not clear at all about your critique

I never find the "some people can..." argument very persuasive.  Some people can run 100 metres in 10 seconds or hit a golf ball 440 yards.  But the overwhelming majority of us cannot.  So the pursuit of 24/192 or the ultimate running shoes or driver may be a waste of money for us.

I have no pre-conceived notion here as to whether 24/192 is audibly better in a normal domestic environment or not.  If it is, I'll pay more for the new but if it's not it's a waste of money. I suppose what I am looking for is a persuasive argument for or against spending yet more money on this

One way forward would be to download a 24/192 file and listen to it. These are frequently offered for free by Linn, Qobuz and others. For example there will be a free download album from Qobuz available at the Bristol Show (how do they do it when they are supposed to have gone bust months ago?).

FWIW I can hear a slight difference but not enough to make it worth my while replacing stuff and I'm streaming rather than buying these days.

I think it is a nonsense. Reebok is certainly capable [in a competent and well produced recording] of resenting music totally to my satisfaction. 

Unless the basic master is superb, and also made at higher resolution than Redbook [and very very few are] then it is a nonsense to even contemplate it. 

At the production and basic recording stages then 24 bit can be useful as it allows for a margin of error in the setting of the levels for the basic recording, which can subsequently be assessed and preserved in all its dynamic potential within 16 bits for the finished CD issue. As for sampling rate, Nuquist theory covers it very well. 44.1 K is fine for the audible range of the human ear, at least for any sounds that could be generated by musical instruments or a sung line. The highest main fundamental frequencies any musical instruments make is in the range 3 to 4 kHz though most are typically in the hundreds not the thousands. If one considers the harmonic series  [the characteristic overtones] that define the timbres of instruments so we can recognise an oboe from a clarinet or such other differences, these are clearly dealt with by a top frequency of 9 KHz, which was for most of the period of 78 recordings all that was deemed necessary. When Full-frequency recordings emerged towards the end of the WW II, then the top frequency was rated at 14 KHz. This golden era of recording - particularly from Decca in th early LP era from 1952 onwards - is widely regarded as an era which has not been eclipsed by subsequent efforts, even though the top frequency presented is often 20 KHz these days.

ATB from George

I'm with you George. Reebok sounds good to me too.

John.

Dear John,

I am not going to fix that! Bloody auto-correction!

ATB from George

I think there are two sides to this. I have read that the reproduction of soundwaves outside of our range of hearing affects how we hear the frequencies that are within the range. Therefore, putting the 44.1KHz sampling rate of CD aside, the 20Khz frequency response limitation is a hinderence even if our hearing doesn't go that high. But that is academic. Don't know if it is true.

What I can say from my own A/B tests with the same material on 44.1Khz and then at different bit depths and sampling frequencies is that I cannot hear any improvement from the increase in sampling frequency from 44.1Khz through to 192Khz. BUT, I definately can hear a major difference with the increased bit depth from 16 to 24 bits. The differences were simply that it sounded clearer and had much better dynamic range. This makes sense because it is the bit depth that dictates the number of possible frequencies that can be encoded and the range.

A direct analogy with video is easy to follow here. The sampling frequency represents the video resolution and the bit depth the possible colours that are reproducible. You can keep increasing the resolution but after a certain point, your eyes cannot distinguish between then pixels on the screen. But increasing the bit depth and the colours and gradients continue to get noticeably smoother and lifelike. It is the same with audio. There are limited gains to be made from the sampling frequency but much more perceptable gains from the bit depth increase to 24 or newer 32 bit streams. Your hearing might have a limit on frequency response and sampling rate but the natural world has no limits on the scale of dynamic range or the gradation of frequencies withing your hearing (unless you go deaf from standing next to a jet engine in which case your peceptable dynamic range will be nil).

So although I hear a marked difference goign from 16bit to 24bit, I have yet to hear a differnce from going from 88.2Khz to 192Khz on the same 24bit bit depth. Of course, maybe the big stack of black naim boxes I have just cannot reveal the difference but I like to think it could :-)

Just turn it off George. I have. It's a PITA!

John.

George is certainly right in that the quality of the master is paramount, if that isn't absolutely top notch then 24/anything is a waste of space.

JSH, yes I am saying that for real musical signals the ear-brain combination can hear things that cannot be encoded in Redbook format.

First, Amplitude Domain:

If you take some types of struck instrument, the initial transient causes a much stronger pressure wave than the resulting sustained note.  The digital encoding must encompass that pressure wave in its entirety; however the 'reference' volume level of the instrument is that of the sustained note.  This reduces the available dynamic range available when real music is being recorded.  For continuous test tones (i.e. sine waves) as discussed in the paper this effect doesn't occur.  The usable 'musical' dynamic range of Redbook is thus reduced by 10dB to 30dB depending on the nature of the recording.  96dB - 30dB gives, at best, 66dB usable range and human hearing is better than that.

Second, Time Domain:

However, Nyquist theory only deals with continuous signals and for continuous test tones 44.1kHz is sufficient.  However, most real music has multiple uncorrelated signals (from the many instruments and voices) and the timing relationships between these are important as well as the individual signals.  Nyquist theory does not deal with the timing between different signals, and this is one of the areas where the human ear-brain combination excels.  Some people can distinguish 4μs differences, most people can distinguish 10μs to 15μs differences and even these cannot be encoded in Redbook (22μs sample interval).

Whether or not harmonics above 20kHz contribute to music or not is a different issue, and one on which I have yet to see definitive evidence.

The issue of dynamic range [as dictated in digital recordings] is that unless the dynamic range is stretched to fully utilise 24 bit recording technique, then a normal musical performance will only utilise a certain number of the bits offered by the recording mechanism , and so the extra bits are redundant.

If a 24 bit recording utilises only 14 of the bits, then mastering it at 16 bit standard will not affect the resultant dynamic. But it reduces the redundant bits from 10 to 2. 

The redundancy represents the safety headroom in a recording. 

Typically the most dynamic natural instrument performances of music come nowhere near the dynamic potential of the 16 bit Redbook standard.

That is why it was chosen. 

Initially the standard was going to be 14 bit, but there were examples where the margin of headroom was considered insufficient for practicle purposes. In reality most recordings outside the classical and Jazz genres come nowhere near challenging the 16 bit standard as over-compression is the norm nowadays. 

ATB from George

This link has comparisons for 16/44.1, 24/96, 24/192 as .wav downloads to try yourself :

http://www.soundkeeperrecordings.com/format.htm

Barry Diament covers this also well worth checking out. 

http://www.barrydiamentaudio.com

Blog :  https://soundkeeperrecordings.wordpress.com

In summary he explains the differences as in 192/24 will ring out longer on cymbals, that type of think. Higher Frequencies holding out further. Worth a read and some digging/research into his sites as there is some wonderful insight in there.

This is what in the old days in my law office we called our "sneaker network." 

George,

In respect of 14 bit, are you referring to the Philips 100, 101 and 104 CD players?  These used oversampling, dithering and quantisation noise shaping to simulate the last two bits as 14 bits was the limit of the economically available DACs at the time.

In respect of the dynamic range of recoded material, the usual measure is the ratio of the short time average SPL of the loudest passage and the short time average SPL of the quietest passage.  However for accurate recording you need the ratio of the highest transient peak pressure to the pressure of the smallest perceivable harmonic.  I have never seen this measured,  do you have any references you could point me to?

If one considers any instrument that is struck - by definition percussion instruments - there is no reference sound pressure from these instruments. All that happens after the initial attack is a steady and usually rapid reduction in sound pressure - called the decay.

To arrive at any standard volume for such instruments would require two arbitrary decisions to be made.

How long after the initial attack is the measurement to be made, and also how hard was the initial percussion made?

The biggest dynamic peaking issue in naturalistic recordings is in fact from choirs, where cancellation and summing of two [or more] voices singing the same line at not exactly the same pitch can produce some rather startling peaks in the signal from the microphone.

Even today this can cause trouble at the microphone itself and at the microphone pre-amp, long before it reaches the recording mechanism ...

As for timing at the infinitesimally small time periods that are smaller than CDs can bring out, someone is going to have to explain to me the relevance. When two musicians play together in the most pleasing ensemble, their timing is measured in the milliseconds rather than anything at the level a CD may not reproduce

Dear Huge,

No references on this, and only conversations from recording engineers who explained to me the practical issue of choral recordings as being the issue that is still beyond any microphone yet made, and beyond the paired microphone pre-amps as well unless a sufficient distance is set between the microphone and the voices so as to prevent peaking. All further issues down the recording chain are insignificant in comparison, as the issue of the real source is all but impossible to solve in the real world. Microphones live a tough life in the real world, and require a degree of robustness in construction to survive for a long service life ...

Placing the microphones further away is a real world work around that has been employed since the 1920s. The recording systems themselves have always dealt well with what ever the microphones send so long as the level is governed so that the dynamics are contained within the capability of the recording mechanism. The consequences of this are that it is very difficult to find a naturalist recording os acoustic instruments that peaks. If it does this is technician rather than system failure.

ATB from George

George,

As ever with instruments and music, you are absolutely correct; however I was trying to use examples that more people would understand intuitively.

The timing issue can occur within the sound from one instrument, as parts of the instrument can act as delay lines to certain transients and/or harmonics (I know you'll easily appreciate how this occurs, but some others won't).

Dear Huge,

I added another paragraph above. 

I think that there is widespread misunderstanding of these issues, and the people trying to market High Resolution notions are quite happy to promise the earth and blind with science, adding nothing substantial to the layman's understanding in the process. 

ATB from George

Dear George,

Yes if the microphones / preamps can't cope with the full acoustic signal, then it doesn't help if the recording system is extended to greater dynamic capability - the microphone will still be the limiting factor.  That's interesting.

4us differences in what.

AND, how is that relevant to listening to music constructed at an interval of 22us

Uncorrelated signals, and because it can't be encoded in Redbook.

Please see previous posts for full explanation.

But the only format for soul music.