[keyser]

On most dvd players there is an option of having an output of 96hHz or have it converted to 48kHz?? What is this for and what does it mean. What sources would output 96 and do most amplifiers take that frequency (like the denon 3802)?

Is there anything more I need to know about this frequency phenomenon, like the Philips Q50 has 4x overssamping and makes a 54mhz signal out af 13,5.. is this purely for video?

Can someone give me a 101 on this stuff?

[Reiner]

There is one audio format on DVD Video called 24/96 (24bit/96kHz) - not to be confused with DVD-A though.
Basically it was supposed to be a high-end audio format on DVD Video disc, containing still images or very short video clips.

Initially it was prohibited to send the 24/96 signal down via the digital output for copyright reasons, as well there was a time when no AV amplifier/receiver could actually handle (decode) it. Thus you had to use the decoder in the player and connect it via the analog outs.
The reason for the downsampling was to protect the amplifier and / or speakers from the increased frequency range. Not that it might have been necessary, but better safe than sorry (I guess) ...

Is there anything more I need to know about this frequency phenomenon, like the Philips Q50 has 4x overssamping and makes a 54mhz signal out af 13,5.. is this purely for video?

That's only related to the Video DAC (digital to analog converter).

[keyser]

Do current amplifiers like my denon 3802 accept 24/96kHz signals? Are there any dvd´s that have this format of sound?

[MikeC]

Keyser
The 3802 you have according to the Denon website has 24bit 96Khz Dacs
so theoretically should be able to process the digital signal. Best check the manual tho'.

Yes, you can still buy these audio DVD's - sometimes referred to as DAD's see - Chesky , as well as
Classic Records.

The quality is outstanding but the choice is somewhat limited.
A word of warning, don't be fooled into buying cd's which are 24bit/96Khz mastered. These are only cd's.

HTH.

Mike

[buns]

This is slightly aside from the topic, but seemed a good enough place to ask!

Im doing signal processing as one of my final year degree modules.....now from what im led to believe, if you sample a signal at a certain rate (the nyquist frequency) all information from the original signal can be maintained. Now before i go question my lecturer on whether he's spinning me porkies......what is the point in massively high sampling rates when using the same bandwidth of signal?

ad

[Reiner]

The point is to take more samples and thus allow a more accurate reconstruction of the original signal. If there aren't enough samples you need to interpolate the missing information, possibly resulting in a wrong (or rather less accurate) reconstruction.

The nyquist theorem says you must sample at least at double the frequency of the highest frequency in your signal, so for music (20Hz - 20kHz) this would be at least 40kHz (can't recall why CD actually samples at 44.1kHz - anyone?).

As for DVD-A (and SACD) they sample at much higher frequencies to increase the bandwidth, so if you have e.g. a DVD-A recording of 192kHz the highest frequency re-produced can be 96kHz.
One should also not forget that the low frequencies (say a bass at 65Hz) is still sampled at 192kHz when using PCM! Kind of waste actually ...

[michaelab]

Yes - but why would anyone want to reproduce a sound at 96kHz! Not even dogs can hear that.

Since the CD sampling frequency of 44.1kHz captures all sounds audible by the human ear I really can't see how a higher sample frequency will make any difference.

AFAIK CD DACs don't do any 'interpolating' other than of course smooting the 'stair' shaped digital 'sound wave' into a smooth one. You could argue that with a higher sample rate there would be less 'smoothing out' to do, and you'd be right, but the eventual analog sound wave created wouldn't sound any different because any extra harmonics or fluctuations that a 192kHz sample rate would capture would necessarily be above 20kHz which we can't hear.

I haven't ever heard DVD-A or SACD. I'll believe it's better when I hear it.

Michael.

[NicolasB]

The ear is emphatically not a linear device. If you play a single pure note then the ear will more or less hear only that note; but if you play two or more notes at once then things get much moire complicated. As well as hearing the two notes themselves you begin to hear other notes mixed in - a note whose frequency is the sum of the frequencies of the two real notes, another one at the beat frequency (the difference in frequency between the two real notes) and others too.

What this means is that if you play a number of ultrasonic notes at once then you can end up with beat frequencies that do actually fall within the audible range. So while you don't hear sounds much above 20 kHz (and I might add in passing that some people can hear 22 or 23 kHz) the presence of ultrasonic sound components does actually modify one's perception of the frequencies that you can hear.

I would hypothesise there may also be some effect from how rapidly a note can go from zero to a high amplitude or vice versa - it can't do it in less time than the sampling interval.

[michaelab]

Thanks Nicolas, that's the first believable explanation I've heard about why a higher sample frequency should make any difference. However, if these beat frequencies etc are artefacts created by the human ear (and don't really exist?) then surely we'd need to use speakers with a frequency response much higher than 20kHz to get any benefit from DVD-A or SACD?

Michael.

[NicolasB]

I don't know about any benefit, but I suspect you're basically right, yes. Some speakers do do this, of course. The KEF "Reference" series, for example, uses a "hypertweeter" which is rated at up to 50 kHz and has some response as high as 70. I think it's safe to assume that KEF wouldn't have bothered with that if it didn't make a perceptible difference to the sound quality.

Some headphones have quite a wide frequency response too. My faithful Sennheiser HD600 headphones are rated at 14Hz-39 kHz.

[MikeC]

Don't also forget that if you sample at 44.1Khz, you have to have incredibly steep low-pass filter set at just above 20Khz to remove ultrasonic junk.
These filters almost certainly affect frequencies below 20Khz ie can be audible.
Therefore, by sampling at the much higher frequency of 96/192 Khz, these filters can be
made less steep and at a higher frequency and will consequently not affect the
audible frequencies anywhere near as much.
Oversampling I beleive also has the same effect.

Mike

[buns]

so basically nyquist doesnt sample the signal without loss, rather it is an approximation?

ad

[NicolasB]

buns:

What Nyquist states is that in order to reproduce a component of the original signal that has frequency x you have to sample at a frequency of at least 2x.

So to reproduce a 20kHz signal you have to sample at >40kHz.


MikeC:

Good point.

[aardvark179]

Originally posted by Reiner
(can't recall why CD actually samples at 44.1kHz - anyone?).

When stored digital audio research was being done in the 70s the digital audio data was encoded onto a video signal that could then be stored on a video tape, the sample rate was one that happened to be 40Khz + some headroom, and allowed easy encoding onto the video signal. Until CDR and the like came along CD images were still sent to manufacturers encoded on video tapes.

If you didn't need to store the audio data and were just transmiting it then you could choose a nice round number for the sampling frequency, which is why almost everything else has a sample rate different from CDs.

[michaelab]

Check out this web page:

http://www2.cs.ust.hk/faculty/layers...waveforms.html

..quite a good basic intro on sampling frequencies, Nyquist and aliasing (which happens if you try to sample a frequency higher than the Nyquist frequency, ie more than half the sample frequency).

Michael.

[Reiner]

so basically nyquist doesnt sample the signal without loss, rather it is an approximation?

Basically correct but you can't blame him for that. No matter how high the sampling rate, it will always be an approximation only.
Then again there comes a point where an extremely high sample rate wouldn't make sense anymore.
Thus Nyquist's theory is good enough for most applications (ISDN *) is based on that, too) but perhaps not hifi.

*) ISDN samples at 8kHz with a resolution of 8bit = 64kbit/s

[buns]

ah ok thanks. Just like to get the fine details of things correct in my mind so i dont go blundering in with the wrong ideas!

ad

[bjd]

QUOTE:

Yes - but why would anyone want to reproduce a sound at 96kHz! Not even dogs can hear that.

Maybe if you look here:

http://www.ocf.berkeley.edu/~ashon/a...trasonics2.doc

That's just a summary.....the links at the bottom are worth reading. Could be one of the reasons people like myself who were brought up on high end vinyl systems find the cd medium so unsatisfying

Brian

[NicolasB]

The frequency cap is not the only reason why CDs have trouble competing with vinyl. The dynamic range is also an issue. The sound of a snare drum goes to something like 145 dB for the tiniest fraction of a second after it's hit, so any system which limits you to a 96dB dynamic range won't be able to reproduce this authentically.

Also, IIRC, the difference between intensity levels represented by consecutive bit values in CD audio is constant regardless of loudness, i.e. the difference in intensity between what is represented by the number 60,000 and the number 60,001 is the same as it is between 12 and 13. As the ear is not a linear device but (approximately) a logarithmic one, and as 16 bits limits you to only 65,535 possible values across the entire range, that means that means that towards the quiet end there simply aren't enough possible values to give an audibly smooth signal.

The difference between a good CD player and a bad one is not that the good one is reproducing the CD recording more accurately, it's that it does a better job of working out what the original recording must have been like in order for the CD recording to eventually end up in the mangled state that it's in.

I think it's a great shame that all they seem to be doing with modern audio formats is messing about with multichannel sound. Nobody listens to music sitting in the middle of the orchestra, that's just silly. If they concentrated on providing a 24-bit 96kHz recording medium in stereo then that would do wonders to the actual quality of the sound. That probably would challenge high-end vinyl.

Mind you, I think binaural recordings are a good idea, so what do I know?

[buns]

i agree, forget the silly multichannel audio and get the best from stereo!

ad