Bits more important than kHz?

Psssssst ... don't let any of the 'audiophiles' around here know what is commonly accepted among professional sound engineers around the world. 

The 24 bits are indeed handy for later mastering. They define the dynamic range and 24 bits give them some headroom to manouever during mastering. A sample rate above 48 kHz on the other hand is a waste. The Nyquist theorem states that the sample rate of a recording must be twice that of the frequency which is to be recorded. Ths means that a 48 kHz recording can capture frequencies of up to 24 kHz. Human hearing usualy goes up to somewhere around 20 kHz at best. 

Pinkhamster for a variety of reasons including ADC filtering the maximum sampled signal is usually less than half the sampling frequency in the real world.

Simon

Digital recording is very critical to clipping and this is what every balance engineers has to avoid. 24-bit allows to work with a comfortable margin and still having 20 or 22 "real" bit captured. 

BTW 20 kHz is the theoretical upper range of human hearing. There might be a very very few golden eared people (one per million? don't know) older than 25 beeing able to hear such high frequencies, usualy this ability gets lost by growing older. For a person of lets say 40 years age 15 kHz would be a very good result. 

That's interesting. Begs a question for me as to why some people claim that super tweeters make a positive difference? Don't those operate operate at 20kHz and above? 

MDS

There are some more or less esoteric theories why frequencies of 18 kHz or higher may influence human listening experience. In my opinion everyone should decide for themselves. If someone feels better with a system upgraded in such way, why shouldn't he do? It's all about having fun with good music!

A lot of the talk about super tweeters when they came out was that by allowing the V high frequencies "out" (technical term) the harmonics were also liberated and "de-congested" (another tech term) the sound.

This maybe a load of cobblers or a very poor explanation form , or even both.

I read once that ultrasonic frequencies were capable of being detected by mastoid bone conduction rather than via the good old eardrum. Mastoid bone sound conduction is certainly part of normal hearing and, if I remember rightly, why our voices never sound the same when recorded and played back to us. If the hearing loss we all (will) suffer from as time gets the better of us is solely conductive hearing loss, which I'm fairly sure it is, then mastoid bone conduction should be unaffected. This might explain why even people of an advanced age claim to hear the benefit of supertweeters.

I'm only theoretically interested anyway: my 804Ss apparently are good up to at least 30kHz, which saves me wondering whether I need to spend more money on CDs or a set of supertweeters.

Mark

FWIW

I use a decent A to D to record audio off Vinyl for replay through my DAC. Experiments with various bit rates and frequency resolutions have consistently shown the biggest audio improvements to come form higher bit rates and in fact purely by coincidence i have settled on 24bit/48KHz as the best choice finding all I get from higher frequency recording is larger and larger file sizes which use up NAS space greedily.

Maybe it's not coincidence!

(Nice to see you back here Geoff ! Quite a change in your profile, but I may be guilty of not having followed the whole story. Hope all is well with you. Maurice)

Mark - I hope you're right. My Focal Electra's reach up to 40kHz apparently and I'd like to think I'm not missing over half of their capability. I also seem to remember reading somewhere that super-tweeters can improve the perception of bass performance which if true seems very curious.   

MDS 

I doubt that many people listen to pure sine waves and that the ear is a Fourier transform machine. It's rather a device to precisely detect timing in signals in real time, eg a breaking twig in the bushes behind you.

I find this article interesting:

http://www.yamahaproaudio.com/...temporal_resolution/

To also accurately reproduce changes in a signal’s frequency spectrum with a temporal resolution down to 6 microseconds, the sampling rate of a digital audio system must operate at a minimum of the reciprocal of 6 microseconds = 166 kHz. Figure 515 presents the sampling of an audio signal that starts at t = 0, and reaches a detectable level at t = 6 microseconds. To capture the onset of the waveform, the sample time must be at least 6 microseconds.

And this article:

http://www.physics.sc.edu/~kunchur/papers/HIFI-Critic-article-by-George-Foster.pdf

The article also points out that older people do not lose their abitilty to detect timing differences which would happen if the waning ability to hear high frequencies would be of concern for this.

Well, many who tread the 2ch plus sub route (as you did before the Electras, and I still do), notice the opening up of detail in the upper registers that a well set up sub brings to a system. 

Last time you were round at mine the sub was "always on" but next time I could A/B it and turn it off.

I've always looked at it in terms of what we consider to be a bass note..

In terms of my taste in music it is either a bass guitar or something from the big skins on the drumkit... However, these notes are summoned forth either by an impact or the release of a pluck. This starting point is the high frequency leading edge to that note... It defines the note as cleanly as the tweeter allows. Great tweeter with great bass drivers to follow up results in a wonderful bass note..

At least it does in my experience. 

I once downloade the very same song in

- 16/44,1 WAV

- 24/44,1 WAV

- 24/88,2 WAV

- 24/176,4 WAV

- 24/352,8 WAV

and I put all of them on the very same USB stick and did some listening test with my Naim DAC and to my ears there was a great improvement in overall sound quality going from 16 Bits to 24 Bits whereas increasing the sampling frequency from 44,1 to 88,2 only changed the sound very, very minimal ( I am not sure If I would be able to tell them apart in a real blind test to be honest). Going higher than 88,2 did not change anything to my ears.

Thanks for the links, J, very interesting.

Sorry, I stand corrected: this is a fascinating read. The HifiCritic paper in particular, where the author George Foster goes a long way into simplifying the conclusions of the research carried by Milind Kuchur, a physics professor at the University of South Carolina.

Many misconceptions and mysteries surround the perception and reproduction of musical sounds. Specifications such as frequency response and certain common distortions provide an inadequate indication of the sound quality, whereas accuracy in the time domain is known to significantly influence audio transparency. [...] Our recent behavioral studies on human subjects proved that humans can discern timing alterations on a 5 microsecond time scale, indicating that the digital sampling rates used in consumer audio are insufficient for fully preserving transparency.

This made me remember what Barry Diament once said about 24/192 recordings:

When I first heard properly done 24/192, it was a jaw dropper. For the first time in my experience, those reservations I have always had about digital, where I felt there were some things the best analog did better, simply evaporated. This is, to my ears, a bigger jump up in quality over 24/96 than that was over 16/44. It no longer feels like a great digital recorder or a great analog recorder. It feels like the recorder has been effectively removed from the equation and I am listening directly to the mic feed.

George Foster sums up Milind Kuchur's findings like this:

•    Kunchur makes a strong case for high resolution formats since CD sampling rates can only resolve timing differences down to 11μs, but clearly we need much higher resolution playback at least 24/192kHz- to reach levels where time smearing will be
inaudible. This is especially the case when listening to percussion where transients lasting less than 10μs are common. The Red Book CD standard was never good enough from this particular viewpoint.
•    I feel that speaker buyers, builders and designers need to think small. Large cones or panel diaphragms will mean that the different parts of the speaker will be at different distances from the ear. Kunchur calculates that for a 1.5m tall electrostatic panel 5m from the listener, the sound from the panel edge will take 650μs longer to reach the ears than from the centre – a figure 100 times the minimum audible delay criterion. This may explain why small drivers (eg my Jordan JX92s) give such a clear sound on transients.
•    The time alignment of drivers and the time resolution of the system are of great importance.
•    It throws up questions about the wisdom of bi-wiring; it is easy to lose the time coherence of the signal with small errors in connection practice.
•    Many of our established criteria and explanations may be based on results which were obtained using insufficiently sensitive test arrangements. What we take as established fact may well be the result of an inability to measure properly.

It looks like this PRaT thing might have some importance after all...

A must read ! Thanks again, J.

Maurice

Indeed. At least the statement "everything above 20kHz is irrelevant" should be taken with a grain of salt.

Oh, but beware: this is another story altogether.

Everything above an audio frequency of 20 kHz may or may not be relevant (and some posts above have pointed to reasons why it may, like harmonics or bone conduction), but what Kuchur finds is that everything above a sampling frequency of 44.1 kHz (and up to 167 kHz) is actually relevant. Not for correct audible frequency reproduction, but for correct timing resolution (of changes in frequency, amplitude, or most often, both).

That agrees with my findings when digitising my LPs.  I finished up ripping to 24/44.1.

John

That makes a lot of sense, Mike. Will be interested to hear your A/B comparison at some point. As to my own experience, I guess any loss in the treble department from removing my old sub was more than made up by that lovely beryllium tweeter in the Electras.    

+ 1

Fascinating stuff. It's very good news to read (once again) that timing is the most important aspect in music reproduction. It's also very good news to read that the ear-brain system's timing ability is not dependant on hearing sensitivity.

But.... I'm having difficulty trying to square that with Mike's remark on the importance of the high-frequency leading edge in defining the beginning of a note. If we lose high-frequency hearing, then don't we also lose some of our ability to detect timing, or have I missed something ?

Jan

Interesting... A few years ago I was chatting to Ingvar Öhman the designer of the Guru range of speakers and he was saying that the impulse response of his designs was more critical than the frequency responce when it came to the musical impact. It's all in the timing...

I do wonder about the merits of extended high frequency. Try listening to a signal at 12kHz , it is incredibly  high pitched. At  around 14kHz have a vague sence of something being there, but cant get any meaningful info from the sound. Try at 18kHz and its gone for me completely. True if you are a 20 years or less you can start to hear these frequencies, but it does fall away quite quickly with age.

So with regard to transients (and in consideration of how the ear works) I can't  see how a sharp transient with frequency elements above ones ear cut off are going to be relevant. But absolutely agree phase alignment of the radiated audible energy is essential for a coherent and accurate sound.

I suspect hf performance it more to do with the performance  of the extended tweeter, diffuser  and crossover within the audible range, rather than anything happening outside of the audible range.

Simon

Jan,

I think there is a bit of confusion between audio hi-frequency (pitch) and sampling frequency. The higher sampling rate means that the individual smaller more frequent samples will capture the start of the sound curve (or rather the ears ability to distinguish it) regardless of its pitch and so improve one's perception of the music's timing.