Bitflipper,
I'd still enjoy your thoughts on my previous comments.

But, I'd like to ask another question: Doesn't your explanation only hold true within the context of 44.1 digital distribution? Why does the signal HAVE to be bandpass limited at 20K?

So in theory something recorded at 192kHz digital and distributed on (for the sake of discussion) 2" high bandwidth analog tape may actually contain content that *someone* can hear?

best regards,
mike

ORIGINAL: altima_boy_2001


I still believe if frequencies above 20kHz matter at all then it should be possible to construct a sound file in 96/192kHz format consisting entirely of frequencies above 30kHZ (well above human hearing threshold) that can generate sound a human could directly hear or can be proven that humans can perceive in some other way through blind testing.

If anyone can create this demonstration then I will believe that these higher rates are really that important.

Actually that logic doesnt entirely follow

Take 2 sound sources, one making a 400 hz sound, and another playing an 800 hz sound. The waveform of the 800hz waveform has that ability to change the 400 hz sound because you get constructive or destructive interference as the two waveforms "merge". Similarly sounds over 20 khz have an effect on the sounds below 20 khz. So while you cant hear them they are affecting what you do hear. exactly how much, I have no idea.

"Why doesn't anybody do this?"

Clearly people say they can percieve the difference... they say it over and over and over again it's just that *the other* camp then calls the people making the claim crazy or uneducated.

at least that's my observation,

mike

So in theory something recorded at 192kHz digital and distributed on (for the sake of discussion) 2" high bandwidth analog tape may actually contain content that *someone* can hear?

Yes, if you specifically wanted to record ultrasonic material, you could do that with higher sample rates, assuming your distribution medium supported it, e.g. DVD at 96KHz. Nyquist's whole elaborate scheme is based on the assumption that only frequencies below 20KHz matter.

I've struggled with the idea that in the real world there are high frequencies that, though we cannot hear them directly, do affect sounds that we CAN hear. Much of what we hear in a cymbal crash, for example, is the result of the interplay of many unheard frequencies. Something like 75% of the sound energy of a cymbal is ultrasonic. No doubt if you recorded cymbals in a room with lots of very high-frequency absorption they would sound unnatural and dull.

However necessary they may be to lifelike recordings, all that ultrasonic interaction happens acoustically out in the real world, before it reaches a microphone. I am not convinced that there is any benefit to actually recording ultrasonic frequencies. For starters, how are you going to ever play it back?

How about if a clinically deaf person told you they heard something beyond YOUR range of hearing? Would you quote your physics textbook... or just call the person a fool??? Or maybe just think what if?

Point taken. It is entirely possible that some people can hear frequencies that you or I cannot hear. The traditional 20KHz figure is somewhat arbitrary, and skewed because of the narrow demographic used when it was determined. We know, for example, that Asians (on average) hear high frequencies better than Caucasians, probably the result of differences in average size of ear canals. It might explain why traditional music from Asia leans toward percussive high-frequency melodies and instruments. But it doesn't explain the popularity of Kiss in Japan.

I, too, have heard the story of Geoff Emerick's ability to discern a 3db dip at 54KHz. I suspect that anecdote has been repeated for the same reason people repeat urban legends and ghost stories -- it's fun to think that such an ability might really exist. Doesn't mean it does. A 3db change is close to the minimum amount of change you can hear in the most sensitive part of the spectrum. Put that in perspective: your room probably has peaks and nulls of 20db or more, and you're normally completely unaware of them. The only way you can hear a 3db difference is when there is a rapid change such as moving a slider on an EQ - a static 3db dip in a console channel would be nearly impossible to discern unless it were to suddenly change.

Question: Was Nyquist's Theory exclusive (and in anticipation of) to digital (quantized) audio or did he also apply it to bandwidth issues in the analog domain... things like analog sound recordings and or radio transmission?

Nyquist was specifically working toward digital (quantized) audio. He was financed by Bell Labs, so obviously the practical application was intended to be telecommunications, not entertainment. The motivation was all about bandwidth, or rather how to overcome bandwidth limitations of analog transmissions. Nyquist set about to determine the most efficient way to digitize audio based on whatever bandwidth was desired, not necessarily high-fidelity, although the theorem applies to all digital representation of analog data, including things like temperature readings and earthquake shockwaves.

I have a friend who makes a hobby/side business replacing DACs in CD players. The hot trend is to eschew oversampling and embrace the latest much more accurate clocks. He replaces the DAC and clocks with new parts.

This (oversampling in the context of D/A converters) is an entirely different subject. We were talking about sample rates while recording. I don't doubt that many cheap CD players use cheap DACs, and could be improved by replacement with better components. But if you're listening to a cheap CD player, it's not just the DAC that determines the quality, because if the manufacturer skimped on the digital components, they probably skimped on the analog section and the transport, too.

I've read some interesting articles about improving clocking in CD players. Interesting because most are laden with a lot of pseudo-scientific audiophile B.S. The truth is that jitter is not that big a problem in CD players (most are < 100ps), although that doesn't stop vendors from scamming clueless audiophiles with $10,000 CD players.

(Please don't infer from this that I assume your friend is a scammer, Mike! I honestly do not know if one can hear the difference between a stock CD player and one that's been reclocked. It would make for an interesting double-blind A/B test.)

ORIGINAL: John

In the end just what are the ones that say they have the remarkable ability to sense content above the normal range of hearing using to hear this content? The signal chain is not just the A/D converters. On playback (I assume we are talking about playback when we say we hear/sense this stuff) there are pres amps and amps used. Many have filters that cut off non audible frequencies. But lets say that we are using infinite band width amps. What of the transducers known as speakers able to do? In most cases they are pushed very hard to get a clean sound at 20 kHz. Most tweeters can't get to 20 kHz. What I find here is that those that support the notion that content beyond 20 kHz is important often do not have the equipment to reproduce frequency in that range. So how can they make this claim? I think it is the placebo effect and although they are sincere in this belief the truth is, it ain't there.

Best
John

Totally agree. The only real benefit I see in recording at 96 KHz is to stay future proof in case things change (as in actually adopting DVD-A as the standard format).

EDIT: Also, another benefit is that some plugs behave better at this higher frequencies.

I find this whole thing funny. Not one poster that supports the idea has ever said that they can hear this themselves they only relate anecdotal hearsay to plead their case. That alone should give one pause. Carry on gentlemen.

Best
John

To reap full advantage of 192kHz recording, your front end needs to be absolute world-class from top to bottom.
If that's not the case, the move to 192kHz will largely be insignificant.
FWIW, I'd focus more attention on instruments/mics/pre-amps. Improving the source will have a much larger impact on your final sound than moving from 88.2/96k to 192k.

FWIW, No one will decide to buy (or not) a record because of the sample-rate used to record it.

a studer multitrack can record up to 50 khz. Wich is a studio medium people allways adored for it's true reproduction of recorded material. 96khz sampling holds material up to 48khz wich is allmost the same. (Every sample is halve a sinewave, so 96khz/2= 48khz highest frequency you can record).

96khz/24bit can finally be put on on a consumer medium. HD DVD & Blu Ray. So finnaly people at home can enjoy the same quality wich was recorded in the finest studios then.

Muziekschuur

ORIGINAL: bitflipper

However necessary they may be to lifelike recordings, all that ultrasonic interaction happens acoustically out in the real world, before it reaches a microphone.

From my understanding, the interaction happens as the sound waves hit the microphone or any non-linear device like our ears. (Air is also non-linear with enough SPL but we are talking about recording music here, not thermo-nuclear explosions :).

I, too, have heard the story of Geoff Emerick's ability to discern a 3db dip at 54KHz. I suspect that anecdote has been repeated for the same reason people repeat urban legends and ghost stories -- it's fun to think that such an ability might really exist. Doesn't mean it does. A 3db change is close to the minimum amount of change you can hear in the most sensitive part of the spectrum. Put that in perspective: your room probably has peaks and nulls of 20db or more, and you're normally completely unaware of them. The only way you can hear a 3db difference is when there is a rapid change such as moving a slider on an EQ - a static 3db dip in a console channel would be nearly impossible to discern unless it were to suddenly change.

The explanation I have read many times is that a 3dB bump at 54Khz in the analogue domain will give a phase shift down in the audible range. If Geoff Emerick was testing his newly delivered console and was switching identical signals between channels, the change would be instantaneous. That would make it much more obvious.

UnderTow

ORIGINAL: wicked
If you have a sin wave at 1 hz and you sample at 2 hz anything in between the two sample points will be averaged to the two points, like conect the dots so what if there are small spikes in the wave between the points, they will not be there.

Any small spikes will be frequencies higher than 1Hz which means they are above the 1/2 Fs. In other words, they are not part of the 1 Hz sine wave.

What is the highest absolute sample freq to use , Don't know but the lowest is 2fc

The highest needed is 2 X our audible range. :)

UnderTow

All I will say on this is that music is about much more than physics,

Its also about Voodoo too I guess.

You may have a point about music but not about sound.

It could be very informative if you would describe how you did the blind listening tests. It would be useful for you to list the equipment used and specs. What materials where used?



Best
John

ORIGINAL: bitflipper

So in theory something recorded at 192kHz digital and distributed on (for the sake of discussion) 2" high bandwidth analog tape may actually contain content that *someone* can hear?

Yes, if you specifically wanted to record ultrasonic material, you could do that with higher sample rates, assuming your distribution medium supported it, e.g. DVD at 96KHz. Nyquist's whole elaborate scheme is based on the assumption that only frequencies below 20KHz matter.

I've struggled with the idea that in the real world there are high frequencies that, though we cannot hear them directly, do affect sounds that we CAN hear. Much of what we hear in a cymbal crash, for example, is the result of the interplay of many unheard frequencies. Something like 75% of the sound energy of a cymbal is ultrasonic. No doubt if you recorded cymbals in a room with lots of very high-frequency absorption they would sound unnatural and dull.

However necessary they may be to lifelike recordings, all that ultrasonic interaction happens acoustically out in the real world, before it reaches a microphone. I am not convinced that there is any benefit to actually recording ultrasonic frequencies. For starters, how are you going to ever play it back?

While true that audio in the ultrasonics can interfere such that harmonics can project back in to your range of
hearing, that's not really the point. The microphones in the room are going to shave off things around the 20kHz to 22kHz range.
An SM57 tops out around 16kHz, an U87 at 20Khz, as does an AKG C414, etc.
It is a rare microphone that has response beyond this. If one were to track at 48kHz sample rate, you may reproduce what the
microphone recorded with 100% fidelity up to 24Khz .. well above the range of most microphones.

lots of mics go way beyond 20K, most aren't flat response past some point but it's a fallacy to think mics CAN'T respond to hi frequencies.

Call up Sweetwater and see if they'll send you the Earthworks demo CD. I would not have thought that anyone could convince me to buy a microphone based on a sample recording on a CD. But they've done an excellent job of doing just that. Whether it's because they go where only dogs can hear or just because they're very responsive to transients and high frequencies, they do sound really good regardless.

When this topic came up in a previous thread, someone pointed out that Earthworks microphones have a serious drawback, which is that load on the phantom power supply is so high that most converters could not handle more than one or two of these mics at a time.

ORIGINAL: SteveJL

As someone with an audiophile sound reproduction system, and having done blind-listening tests with other audiophiles, I can confirm that the difference in sound quality between 44.1k, 96k, and 192k is indeed audible, given sufficient-quality source material, ears and brain that can listen, a good listening environment, and an audiophile-quality sound reproduction system. All I will say on this is that music is about much more than physics, and no-one can truly explain the brain-component of the listening experience. If you want a shot at the best possible sound, you have to use the best equipment you possibly can (in the entire chain), record at the highest possible specs by the best-qualified engineer possible. You are on the right track. It's your music, have fun

You are only hearing differences in that particular converter performing at a particular sampling rate - it is not caused by sampling the higher frequencies. Typically, there's measurable distortion occurring in systems running at 192kHz and even 96kHz to a much smaller extent (this is due to the physical components of the converters not being able to handle the speed). Many people hear that distortion as being "more analog" or whatever. Just like tubes. These errors are measurably making a system less accurate at higher sampling rates. There's some great info that explains these situations (to an extent) in the white paper section at www.lavryengineering.com