Can someboy explain me why is 192KHz Necessary

192KHz may not be necessary (unless you work for Sony)

Nyquist showed that 40Khz is the highest sampling rate need to recover a 20KHz time domain signal.

However, because you need to keep signals above that range from distorting your sampled signal (aliaseing) you low-pass your analog signal (anti-aliasing filters). This is why 44.1KHz was used as a sampling rate for CD's. The extra 4.1Khz allowed for the anti-aliasing filters. This filter need to be of a very high order and will introduce phase and amplitude distortion into the sampled signal (unless you spend a lot of money an the filter). If you increase the sampling rate the anti-aliasesing filters get much easier to design and build (and cheaper). Phase and amplitude distortion introduced by these filters will no longer be in the audible signal range.

Some sound sources have considerable content above 20 kHz. You may not be able to hear these frequencies directly, but they can produce psychoacoustic effects such as combination tones and beating that are audible. A 44.1 kHz sampling rate is not enough to encode this information which contributes to the realism of the sound; however, 96 kHz should get just about everything that might be there. I'm not sure if 192 kHz is strictly necessary, but there is the argument that having more than the bare minimum of samples per cycle at the highest frequencies allows for a better sounding high end. Also, there is an interesting paper related to this over at the Earthworks Audio site that you might want to read.

Thanks guys!
That's all very cool...

What about peak response accuracy?
44.1 KHz sounds like ie, if you sample a sequence of snare samples, the AD will not capture the transients equally... And maybe the same can be happening to Limiters and Gates... they may be not behave the same way twice, if the sampling miss some data on one beat and another data on the second beat of the same sample...
I have also made the test of samplig 20 KHz at 44.1, 96 and 192, and only 192 didn't show any kind of amplitud modulation. When I downsampled the 20Khz from 192 to 44.1 (everything inside protools) the tone showed the same amplitude modulation but a slightly differnt waveform. Could this mean that is better to sample at 44.1 instead of 192 and downsample afterwards if no DSP is involved in your project?

That's what I would tend to think.

The 192 controversy is heating up. There have been lots of discussion of this on the mastering web (see web address below) and no one seems to be able to give a definitive answer on why or why not 192. Lots of people are in the pro camp and just as many in the con camp.

In the FWIW department...Most microphones don't record well above the audible frequency range, Synths and keyboards don't put out a lot of information above the audible frequency range but guitars and other stringed instruments can put out some hi end frequencies as overtones without too much trouble as can some percussion, winds and brass instruments. The human ear can hear approximately 20 to 20000 Hz but some people can hear above this range and are bothered by ultrasonic burglar alarms which operate at the 22000 Hz level. So yes we can be influenced by out of band responses the question is how much difference this will make in what we perceive and WHO will make the most of this difference. It has been pointed out a lot of times that even the best recording of a live classical concert does not sound exactly like the concert and the reason is ........maybe being in the hall you hear things that you do not hear when you are listening to that same concert that was recorded on a CD.

Much of the popular music created today is lucky to have a frequency range of 20 to 8000 Hz, a dynamic range of 12 dB and a signal to noise ratio of 20 dB so higher quality equipment may not be needed for this type of music. In fact it may make some of it sound worse than it already does given that now you can hear a lot of the junk that in the old days was rolled off by the medium of analog tape or a 44.1 CD.

Various mediums have different frequency responses Telephone is approximately 300 to 3000 Hz, AM radio 80 to 8000 Hz and FM radio 50 to 15000 Hz. CDs can put out 1 to 20000 Hz but at the high end they are running into a brick wall filter which can do strange things to the upper frequency range.

So 96 kHz is probably a better all over bandwidth choice. 192 is nice but it takes approximately 5 times the storage space and there is not much equipment out there that will allow you do process 192 signals. The "Joe average" consumer is very happy with his 44.1/16 bit CD player and will have to be convinced big time if he/she is going to go for a higher bandwidth signal.

If you want to catch the debate on the 192 go to http://webbd.nls.net:8080/~mastering and read the ongoing discussion. You might find it interesting or not depending on your point of view.

the nyquist thing is just about being able to capture certain frequencies...but i would guess, correct me if i'm wrong, that , the more samples you have per second, the more accurate the digital waveform you capture, is..
right?

Perhaps the most extensive discussion on this topic in audio history can be found here:

http://recforums.prosoundweb.com/index.php/t/288/0

Heavy duty discussion. Metaphysical at times.

Have a read of this article regarding low latencies and the connection to higher frequency rates:

huub wrote: the nyquist thing is just about being able to capture certain frequencies...but i would guess, correct me if i'm wrong, that , the more samples you have per second, the more accurate the digital waveform you capture, is..
right?

Nyquist guarantees a completely accurate reconstruction of the signal as long as it contains no frequencies above half the sample rate. You can sample it more often, but you can't get any more accuracy, because its already 100% accurate.

But at the top of the bandwidth, the wave ends up looking like a sawtooth wave and is often not a very good representation of the shape of the original wave. Plus, as I said in my first post in this thread, there's evidence that frequencies above 22,050 Hz that contribute to the sonic experience.

David,

I am by no means a Digital Design Enginner, but if the ultrasonic frequencies are creating "audible" frequencies, these audible frequencies WILL be captured by a 22.05KHz bandwidth, right? There is still no "Hard Proof" of ultrasonic perception being able to "enhance" the audible band. Some of the "Desirable" traits of Hi-Res are actually Intermodulation Distortion created by the power amp!

I also read 40 odd pages of the ProSoundWeb link (A Very, very, extremely good thread BTW). Realize that what you see on your PC when you "zoom in" to the sample level is NOT what the audio looks like on the D/A output. The "stair stepping" you see will equate to frequencies above Nyquest (essentially high frequency square waves), and are filtered out by the D/A filter.

Nyquest works. ANY frequency below 22.05KHz (actually a bit lower to allow for filters, etc) can be represented ACCURATELY (phase and all) by a 44.1K sample rate. I was a bit unclear on phase and exact timing of the signals, but the way A/D's and D/A's work (WAY beyond my comprehension) DO allow the band-limited signal to be captured in its entirety. Even a 44.1/16CD will have a phase resolution equivalent to (I'm trying to recall from memory here) .25" at the standard listening position. Speakers can move more than that, and we certainly move our heads more than that!

Still blows me away! Read the link sdelsolray posted. It gets so deep, you will not believe it :) Even the Sales Manager of Apogee admits that 192KHz sampling is not "superior"! (PG 14 of the thread). The thread also goes into great depth on "ultrasonic perception".

Later :cool:

Hey Randyman. When I spoke of the ultrasonics creating 'audible' frequencies, I meant audible in a psychoacoustic sense. Ultrasonic waves can produce combination tones that can be perceived even though these tones do not actually exist, but if the ultrasonics are never recorded, our minds will never have the opportunity to invent these phantom tones and the realism will be compromised - at least that's the theory, and it sounds plausible to me.

When I spoke of waveshape, I just meant that at the top of the band, the A/D is only taking about two samples per period and this should not be enough to capture the subtle changes in the wave's shape - not that most of us, myself included, could probably tell the difference at those frequencies. Again, this is just what i've read.

I will definitely read the PSW thread. Thanks for the tip and for the discussion.

HOLY GOD! 49 pages in the PSW thread! I can't get involved in this... it's just not an important enough issue to my life. 44.1 sounds fine. Let's just make some music. Forget I said anything :)

going by figures, stats and what Mr Nyquist said, even i'd totally agree that anything as far as 192 is not required.

the only thing is that people have chosed 192 sound over even analog sound as sounding better! so what do you say to figures if the ears respond this way.

Given the state of delivery systems ( ones that use data compression) like MP3 ... and now the proliferation of I Pods ... I don't think that increased sample rates at the production end are going to benefit anyone, other than the manufacturers of converters.

I hoped / thought the sample rate thing had calmed down to a point where we could all concentrate on front end gear like pres and comps but I guess not. One thing for sure. For me, this reinforces my own observation that purchasing pres that are married to converters in an all in one package is a potential waste of resources. Riddle: How can you make a great mic pre obsolete? Package it with an converter that will become obsolete. That's the only way ...

I have heard the difference between lower and higher sample rates and I am a believer. I know people who hear over 22K and I have heard the difference higher sample rates make in a / b testing. I fall into line with guys like Rupert Neve who believe that the higher the sample rate, the deeper the bit rate, the better the sound.

On the other hand, I have to question if this really makes a difference in a world where recordings are delivered primarily on TeeVee or MP3 / I Pod thingies .. and being made by people with an obvious hearing deficit and a lack of musical and production values. Let's concentrate on getting 24 bit 44.1 to sound as good as it can first. This concept that technology can replace talent is completely what is wrong with the commercial music industry. Very few productions these days even take advantage of the capabilities of present day 24 / 48 systems offer. Just because MOTU, Digidesign and RME tells me I need the next generation of converters is not a good reason for me to upgrade. What I use works just fine IMO ...