Sampling Rates

[kexodusc]

How big of a difference does sampling rate make on a digital recording?
I just got a few DVD-A's, and unfortunately I'm stuck listening to them in DD/DTS, which isn't altogether bad, but doesn't push the DVD-A format to its limit.
Is the sampling rate the biggest advantage DVD-A has over CD audio?

[mtrycraft]

How big of a difference does sampling rate make on a digital recording?
I just got a few DVD-A's, and unfortunately I'm stuck listening to them in DD/DTS, which isn't altogether bad, but doesn't push the DVD-A format to its limit.
Is the sampling rate the biggest advantage DVD-A has over CD audio?

Sampling rate is way over hyped for the consumer market. CD is already well above most everyones hearing range. Can you hear 20kHz sound? Is there sufficient amount of 20kHz in music to begin with at a high enough level to be audible, well over 100dB threshold minimums?

Don't worry about it. Later, when you need to upgrade the player, get one that is universal.

[Sir Terrence the Terrible]

Sampling rate is way over hyped for the consumer market. CD is already well above most everyones hearing range. Can you hear 20kHz sound? Is there sufficient amount of 20kHz in music to begin with at a high enough level to be audible, well over 100dB threshold minimums?

Don't worry about it. Later, when you need to upgrade the player, get one that is universal.

Mtry,

I fully understand you skeptisicm with regards to sampling rates, but it is unfounded. The sample rate defines the Nyquist frequency, or upper limit of the recording or playback of the digital audio. While the upper limit of the human hearing is 20khz(most cannot hear this high) you need a sample rate that goes beyond that frequency limit. With the current redbook sample rate of 44.1khz, the Nyquist frequency would be 22,050khz. In order to enforce that upper limit(and keep noise out of the digital system) steep brickwall filters are employed. These brickwall filters have audible effects such as time smearing that can be heard well within the range of human hearing(most time they they cause strings, cymbals and upper brass to sound harsh) If you were to measure the frequency response of a crash of cymbals, or the overtones of some string and brass instruments, you would find that significant energy can be registered as high as 40khz, so 44.1khz sampling rate is not enough to capture the FULL frequency response without some sort of aliasing gong on. Aliasing raises the noise level within the audible band of frequencies so it is desireable to have the cutoff frequency as high as it can be. This facilitates using filters with a long gradual roll off.

24/96khz (as 24/192khz) raises the Nyquist frequency to 43khz which is more than a octave above CD redbook cutoff frequency(22,050khz) so no brickwall filters are necessary(and no time domain smearing which results in a more linear response of a wide banc of frequencies) and the audible problems they bring.

Using a higher sampling rate insures that EVERY instrument within a orchestra will be recorded with harmonics and timbral textures intact. That is something the current redbook standard had a problem with from day one.

[mtrycraft]

Mtry,

I fully understand you skeptisicm with regards to sampling rates, but it is unfounded. The sample rate defines the Nyquist frequency, or upper limit of the recording or playback of the digital audio. While the upper limit of the human hearing is 20khz(most cannot hear this high) you need a sample rate that goes beyond that frequency limit. With the current redbook sample rate of 44.1khz, the Nyquist frequency would be 22,050khz. In order to enforce that upper limit(and keep noise out of the digital system) steep brickwall filters are employed. These brickwall filters have audible effects such as time smearing that can be heard well within the range of human hearing(most time they they cause strings, cymbals and upper brass to sound harsh) If you were to measure the frequency response of a crash of cymbals, or the overtones of some string and brass instruments, you would find that significant energy can be registered as high as 40khz, so 44.1khz sampling rate is not enough to capture the FULL frequency response without some sort of aliasing gong on. Aliasing raises the noise level within the audible band of frequencies so it is desireable to have the cutoff frequency as high as it can be. This facilitates using filters with a long gradual roll off.

24/96khz (as 24/192khz) raises the Nyquist frequency to 43khz which is more than a octave above CD redbook cutoff frequency(22,050khz) so no brickwall filters are necessary(and no time domain smearing which results in a more linear response of a wide banc of frequencies) and the audible problems they bring.

.

No, no no, steep brick wall filter is not used for Red Book 44.1 sampling. That is a misnomer:

http://www.mlssa.com/pdf/Upsampling-theory-rev-2.pdf

http://www.resolutionaudio.com/Up-Oversampling.pdf

http://www.simaudio.com/upsampling.htm

Using a higher sampling rate insures that EVERY instrument within a orchestra will be recorded with harmonics and timbral textures intact. That is something the current redbook standard had a problem with from day one

Don't need it, don't miss it anything that is above 22.05khz. You just cannot hear it. Absolute nonsense that you can or it matters. Anything that is audible gets recorded.

[Sir Terrence the Terrible]

No, no no, steep brick wall filter is not used for Red Book 44.1 sampling. That is a misnomer:

http://www.mlssa.com/pdf/Upsampling-theory-rev-2.pdf

http://www.resolutionaudio.com/Up-Oversampling.pdf

http://www.simaudio.com/upsampling.htm

Using a higher sampling rate insures that EVERY instrument within a orchestra will be recorded with harmonics and timbral textures intact. That is something the current redbook standard had a problem with from day one

Don't need it, don't miss it anything that is above 22.05khz. You just cannot hear it. Absolute nonsense that you can or it matters. Anything that is audible gets recorded.

Mtry,

Hit yourself on the head, it may clear the fog. We are not talking OVERSAMPLING, that is done at the DAC stage. We are talking about recording at a high sampling rate. BIG difference bud. Secondly, there is not an engineer on this planet that would not agree with me that recording at a 96khz sampling rate sounds noticeable better than at 44.1khz.(and 192khz sounds better than 96khz)

While you cannot hear above 20khz directly, transient information in some instruments is located above 20khz. If you limit the response of a signal or sharp attack at 20khz(or even 22,050khz), the transient information will sound blurred. With a 44.1khz sampling rate, a brickwall filter MUST be used because high frequency information has to be removed at 22,050khz(which is the Nyquist frequency for 44.1khz) or aliasing will occur out of band, and within the band of human hearing. If you are recording audio from 20-20khz, that filter has got to be brickwall because the signal level must drop below between -60 and -90db at 22,050khz so as not to be heard. This filter must do this at 20khz(and be at -60 to -90 at 22,050khz). I would say that is pretty brickwall Mtry. So it is not a misnomer at all as you stated. If no brickwall filter is used, then you would have to lower the cutoff point of the audio well into the range of hearing to eliminate any signals above 22,050khz

http://psbg.emusician.com/ar/emusic_...digital_audio/

http://grassomusic.de/english/frames...sh/digital.htm

Here is a challenge to you. If you don't think the sampling rate is important record the sound of a cymbal crash(or trumpet, glock, chimes) at 96khz sampling rate, listen, then downconvert it to 44.1khz and listen to it again. Since I have done this before I know what you'll hear ;>

[Andy2]

Mtry,

Hit yourself on the head, it may clear the fog. We are not talking OVERSAMPLING, that is done at the DAC stage. We are talking about recording at a high sampling rate. BIG difference bud. Secondly, there is not an engineer on this planet that would not agree with me that recording at a 96khz sampling rate sounds noticeable better than at 44.1khz.(and 192khz sounds better than 96khz)

While you cannot hear above 20khz directly, transient information in some instruments is located above 20khz. If you limit the response of a signal or sharp attack at 20khz(or even 22,050khz), the transient information will sound blurred. With a 44.1khz sampling rate, a brickwall filter MUST be used because high frequency information has to be removed at 22,050khz(which is the Nyquist frequency for 44.1khz) or aliasing will occur out of band, and within the band of human hearing. If you are recording audio from 20-20khz, that filter has got to be brickwall because the signal level must drop below between -60 and -90db at 22,050khz so as not to be heard. This filter must do this at 20khz(and be at -60 to -90 at 22,050khz). I would say that is pretty brickwall Mtry. So it is not a misnomer at all as you stated. If no brickwall filter is used, then you would have to lower the cutoff point of the audio well into the range of hearing to eliminate any signals above 22,050khz

http://psbg.emusician.com/ar/emusic_...digital_audio/

http://grassomusic.de/english/frames...sh/digital.htm

Here is a challenge to you. If you don't think the sampling rate is important record the sound of a cymbal crash(or trumpet, glock, chimes) at 96khz sampling rate, listen, then downconvert it to 44.1khz and listen to it again. Since I have done this before I know what you'll hear ;>

Thanks for articute the issue so well. The biggest misconception most people have
is that all CD players sound the same.

[mtrycraft]

Hit yourself on the head, it may clear the fog. We are not talking OVERSAMPLING, that is done at the DAC stage. We are talking about recording at a high sampling rate. BIG difference bud. Secondly, there is not an engineer on this planet that would not agree with me that recording at a 96khz sampling rate sounds noticeable better than at 44.1khz.(and 192khz sounds better than 96khz)



Hogwash. Please site som of the DBT to support his nonsense, thanks.

While you cannot hear above 20khz directly, transient information in some instruments is located above 20khz. If you limit the response of a signal or sharp attack at 20khz(or even 22,050khz), the transient information will sound blurred. With a 44.1khz sampling rate, a brickwall filter MUST be used because high frequency information has to be removed at 22,050khz(which is the Nyquist frequency for 44.1khz) or aliasing will occur out of band, and within the band of human hearing. If you are recording audio from 20-20khz, that filter has got to be brickwall because the signal level must drop below between -60 and -90db at 22,050khz so as not to be heard. This filter must do this at 20khz(and be at -60 to -90 at 22,050khz). I would say that is pretty brickwall Mtry. So it is not a misnomer at all as you stated. If no brickwall filter is used, then you would have to lower the cutoff point of the audio well into the range of hearing to eliminate any signals above 22,050khz


More unalderated hogwash, garbage. Brick wall filtering is not used exactely because of the oversampling.

And Ultrasonic information is not neede. THAT has been demonstrated and published.
Please consult AES on that. Or, if you require, I will site it for you.

You cannot hear it!!!.

[skeptic]

On recording, increasing sampling rate above 44.1K will not improve the recording. There is no reason to oversample on recording to capture sound which cannot be heard by human beings. However, on playback, there are a couple of advantages to oversampling. First, by sampling the same signal more times, the number of samples with errors as a percentage of the total number of samples may be reduced. Secondly, by oversampling and generating a signal which is the same as the 44.1K but repeated 4 times or 8 times as though it were recorded at a much higher sampling rate, the analog filter can be pushed out to a much higher frequency where it no longer affects the audible spectrum. By 1989, the best 20 bit players had approached the theoretical ideal and with newer 1 bit oversampled players, most others have come a long way towards matching or slightly bettering them. Fiddling with the analog frequency response to change the tonal balance of cd players slightly is hailed as a major improvement by some audiophiles but in reality, it offers no tangable benefits at all which cannot be achieved with proper use of an equalizer at far lower cost.

[Sir Terrence the Terrible]

On recording, increasing sampling rate above 44.1K will not improve the recording. There is no reason to oversample on recording to capture sound which cannot be heard by human beings. However, on playback, there are a couple of advantages to oversampling. First, by sampling the same signal more times, the number of samples with errors as a percentage of the total number of samples may be reduced. Secondly, by oversampling and generating a signal which is the same as the 44.1K but repeated 4 times or 8 times as though it were recorded at a much higher sampling rate, the analog filter can be pushed out to a much higher frequency where it no longer affects the audible spectrum. By 1989, the best 20 bit players had approached the theoretical ideal and with newer 1 bit oversampled players, most others have come a long way towards matching or slightly bettering them. Fiddling with the analog frequency response to change the tonal balance of cd players slightly is hailed as a major improvement by some audiophiles but in reality, it offers no tangable benefits at all which cannot be achieved with proper use of an equalizer at far lower cost.

Sir,

You are not just in left field on this one, you are in the wrong stadim, in the wrong city, and in the wrong country, and the wrong planet!!!! You don't oversample at the recording stage anyway, you sample. Oversampling is strictly a playback function. The same benefits you mention for oversampling apply to higher sampling, but higher sampling is more predictable, and reliable for best results. Higher sampling allows you to design filters that roll off gradually, as opposed to a brickwall fashion of filters currently in use for 44.1khz CD

First, using a sampling rate above 44.1khz DOES improve recording. Every engineer on this planet is in agreement with that. That is why both DVD-A, and SACD are EXTREMELY popular among audio engineers. Almost every recording done these days is both mixed, mastered, and archived at 24/96khz. Once again, it doesn't matter that you cannot hear it directly, most of the benefits of sampling at 96khz happen right in the audible range of hearing(sharper transients, more air in the mix, and more open sound)

Secondly, an EQ CANNOT do what sampling at a higher rate can do. If you try and do this, not only will you screw up the phase of the high frequencies, but you are like to burn out your tweeter in the process. Not to mention you just be introducing alot more noise to the mix.

I do not know where you got your information, but you need to give it back. It is wrong, and it will surely damage ones speakers if implemented.

DO NOT TRY THIS AT HOME!!!

[mtrycraft]

Sir,

You are not just in left field on this one, you are in the wrong stadim, in the wrong city, and in the wrong country, and the wrong planet!!!! You don't oversample at the recording stage anyway, you sample. Oversampling is strictly a playback function. The same benefits you mention for oversampling apply to higher sampling, but higher sampling is more predictable, and reliable for best results. Higher sampling allows you to design filters that roll off gradually, as opposed to a brickwall fashion of filters currently in use for 44.1khz CD

First, using a sampling rate above 44.1khz DOES improve recording. Every engineer on this planet is in agreement with that. That is why both DVD-A, and SACD are EXTREMELY popular among audio engineers. Almost every recording done these days is both mixed, mastered, and archived at 24/96khz. Once again, it doesn't matter that you cannot hear it directly, most of the benefits of sampling at 96khz happen right in the audible range of hearing(sharper transients, more air in the mix, and more open sound)

Secondly, an EQ CANNOT do what sampling at a higher rate can do. If you try and do this, not only will you screw up the phase of the high frequencies, but you are like to burn out your tweeter in the process. Not to mention you just be introducing alot more noise to the mix.

I do not know where you got your information, but you need to give it back. It is wrong, and it will surely damage ones speakers if implemented.

DO NOT TRY THIS AT HOME!!!

Oh, Terrance, brick wall filters haven't been used for a long time, very long time. Oversampling is the reason.
Please, check it out.

[skeptic]

In one regard, you are correct, oversampling means sampling the playback at a higher rate than the recording was made at. This is a small technical point. However, increasing the recording sampling rate to extend the captured analog frequency bandwidth will not improve the regenerated analog signal in the audible spectrum in comparison to a recording made at the lower sampling rate and oversampled at the same higher rate. I don't think there is much point in discussion of frequencies beyond the audible spectrum. Some audiophiles will never concede what scientists and engineers already know and that is that their inclusion adds nothing perceptable in terms of accuracy to humans and at worst can actually detract from accurate reproduction by introducing all kinds of new distorton. It is also pointless to argue what mathematicians, engineers and scientists already know and that is that 44.1Khz sampling on recording is sufficient to accurately capture the entire 20Khz audible analog frequency spectrum that humans can hear (if their hearing isn't imparied.)

"Secondly, an EQ CANNOT do what sampling at a higher rate can do."

I never said it could. What I said was that many so called high end cd players merely fiddle with the analog frequency response after D/A and call the altered sound a "breakthrough." On direct comparison, it will obviously sound different but the difference is easily obtained far more cheaply than the hundreds and even thousands of dollars extra they normally charge for these players. Once the analog signal is perfectly regenerated which is not merely possible but now virtually universal with 1 bit oversampling players, NO FURTHER IMPROVEMENT IN THE PROCESS IS NECESSARY OR POSSIBLE. That is one reason why SACD audio is doomed unless it is just as cheap, totally compatable, and gradually replaces RBCD. There is simply no technical reason for it but electronics companies are always looking for new classes of products to trick audiophiles into thinking they need something better than the fully developed offerings already cheaply available on the market.

[Feanor]

... Is the sampling rate the biggest advantage DVD-A has over CD audio?

... I'll never really understand the sampling rate issue despite many explainations.

No doubt the theory is sound that you can perfectly record a sound a sample rating rate of twice its frequency, e.g. 20KHz, (the human limit of audibility), can perfectly reproduced a rate of 40KHz.

There are couple problems in the real world:
1) If you're sampling at 40KHz and a 25KHz sound comes along, any attempt to record it will cause errors, that is, distortion, called "aliasing", below the 20KHz, hence audible. I don't understand why this is the case, but you should understand that sampling sounds above 20KHz must be avoided.
2) Since for each sample recorded, there is a given number of bits representing the amplitude of the sound, an actual sound level might be slightly higher or lower than can be represented by the available bits. This need to round causes "quantization error". For some reason I don't clearly understand, this distortion is correlated to the overall sound level, hence is not random, (like clicks & pops on vinyl), and sounds very objectionable. Stuff must be done to conceal quantization error, usually by added ramdon "noise".

Higher sampling rates and more bits per sample can help to remove these problems "at source" so to speak. But it isn't the higher frequencies that can be recorded that matter make higher rez better -- because we can't hear these frequencies.

Another problem that afflicts digital recording is "jitter", that is, the effect of no taking the sound sample, and/or not reproducing the sampled sound, at exactly the right moment in time. It seems small amounts of jitter can sound quite unpleasant. I don't know whether higher sampling rates inherently help solver the jitter problem, but maybe not!!

OK you experts, go to town.