Understanding Sample Rate

[beerandmusic]

1 hour ago, mansr said:

What a load of twaddle. Everything that guy says is wrong, and I mean everything.

what about the part where he talks over sampling to have the filtering outside the audio band?

 

 

[beerandmusic]

12 minutes ago, Spacehound said:

You asked a question. I answered 100% factually. That is not mocking nor  trolling.

 

 

I asked a simple question...what do you believe since you "inferred" you know how it works.

 

I suggested that most of your comments are just mocking, trolling, or ridiculing, and I just wanted to know if you can say anything besides your stupid one-liners?

 

But apparently not?

 

I ask again, Does most everyone truly believe that sampling higher than nyquist is a possible way to improve SQ?

Please share your knowledge on the topic?

 

Your statement " What you doubt/believe is up to you, as we have seen" is a typical stupid comment that I am referring to....what everyone believes and doubts is up to them, and based on their knowledge and experience.

 

[beerandmusic]

4 minutes ago, pkane2001 said:

 

Read the Wikipedia article on oversampling. Particularly the section titled 'Oversampling in reconstruction'. That's a method of shifting unwanted frequencies and noise out of audio band to enable simpler filtering that doesn't produce as many artifacts. A technique known and used for well over 30 years in CD players and DACs.

 

I already read it and that is what i quoted....

It is still debatable, correct?

My question is the same, does MANSR believe that upsampling can improve the SQ if used in this manner.

[beerandmusic]

1 minute ago, yamamoto2002 said:

 

If bit depth is infinite, it is possible.

 

I calculated real world example about 44.1kHz 16bit PCM data of 1/100th of second (441 samples).

It can store the difference of

1000.0000000 Hz from 1000.0000010 Hz but

1000.0000001 Hz signal is rounded to 1000.0000000 Hz.

Dithering can improve the situation.

By increasing bit depth to 24bit, frequency precision increases by 256 times.
 

 

You totally lost me, but it is always nice to see a fresh perspective (wink)

thanks for sharing.

[beerandmusic]

3 minutes ago, pkane2001 said:

 

It's not debatable that oversampling helps shifts filter artifacts out of the audible range and allows simpler, gentler filters to be used. What is debatable is whether these artifacts were audible in the first place. On that, reasonable people can disagree.

 

that is what i meant....does mansr believe that it can improve SQ?

 

I am really curious if there are smart people that don't believe that anything can be improved beyond pcm 44.1khz?

Regardless of his belief, i consider him very smart.

 

[beerandmusic]

2 minutes ago, mansr said:

Why don't you ask him?

 

I am ...i knew you would be here, and i did ask you specifically about what hans said in the video about increasing sampling for filtering purposes to improve SQ....but firedog made it even easier by suggesting "no one doubts".

[beerandmusic]

3 minutes ago, Fokus said:

 What oversampling does allow is to move the bulk of the filtering action from the analogue domain to the digital domain, where more is possible and at a lower cost.

 

 

In your opinion, Does this oversampling make it possible to improve SQ?
 

[beerandmusic]

6 minutes ago, firedog said:

Beer-

Do you realize we have now changed the topic?

 

i never care if a topic changes...i haven't ever seen a thread on this site that doesn't (wink).  I have already conceded the nyquist theorem.

 

My agenda is the same as when i started the topic, and i have already conceded that in hind site, i would have named the topic differently....and even oversampling PCM isn't where i was going, but since we are here, I would love to see some responses....especially from mansr if he believes anything above pcm 44.1 can improve SQ....

 

[beerandmusic]

9 minutes ago, audiventory said:

 

3. Bit depth, sample rate is potential abilities only. Signal/noise ratio (SNR) is implementation matter.

In some cases, SNR may be better for 16 bit, than for 1-bit/2.8MHz.

Because there may be different quality of sigma-delta modulator.

Also need to consider dithering method for 16-bit.

And sigma-delta modulator at 2.8 MHz is complex device enough.

 

 

So you are unable to state that your belief that SACD can be superior to CD in SQ or not?

[beerandmusic]

2 minutes ago, mansr said:

I switched it off after he opened by collectively insulting all engineers and saying something about "far more clever people" than Harry Nyquist. Nothing good can come after that.

 

Forgetting Hans or your opinion of him, do you believe that either DSD or upsampled PCM can improve SQ in any way, or are you of the opinion that nothing can improve SQ above a standard CD (or non-upsampled PCM 44.1 or however you want to phrase...i think you understand my question?)

 

Truly curious...

 

[beerandmusic]

3 minutes ago, Blackmorec said:

It suddenly dawned on me why sampling at double the highest frequency is sufficient. May seem obvious to many but initially it wasn't to me.....All sound waves are sinusoidal. All you need is the amplitude and the number of oscillations per second (frequency) to perfectly describe the signal...whether you're recording (A to D) or replaying D to A.

 

I learned everything about ADCs and DACs from signal processing, and there you need at least 5 data points to describe a peak, a lot more would be better. But if the peaks are always sinusoidal i.e have the same, defined shape, all you need to know to describe it perfectly are amplitude and frequency.

 

Given the above, there's no reason why higher sampling rates should better describe a particular sound wave within the defined spectrum.  Ha! Got there in the end.

 

So, why is it the that higher sampling rates do sound better? A few months ago I had a demo of some YG Carmel IIs driven by AVM electronics.  I was listening to standard CDs and wasn't happy with the sound, which lacked air and acoustic resonance, replaced instead by the typical digital sting in the tail.  I complained to the dealer, who pushed a button on the amp, after which all was sweetness, light and all the air and acoustic resonance you could want, with no trace of digititis. Clearly it was unresolved components of the music that had caused the problem. As soon as they were properly resolved, away went the problem.  

So what had the dealer done? Switched to the amp's upsampling mode.

 

just curious, what amp had this "upsampling mode switch" and do you know specifically what it did?

[beerandmusic]

quote by unknown source:  nyquist never proved that you can perfectly bandlimit a signal and in practice you can't.

 

http://www.wescottdesign.com/articles/Sampling/sampling.pdf

[beerandmusic]

18 minutes ago, mansr said:

An sinc filter of infinite length does exactly that. In practice, you can get as close as you need to.

I don't know what a sinc filter is or how you get one of infinite length, or what any of this means (grin), but

I would equate "you can get as close as you need to" to a statement similar to "little audio difference"?

 

 

[beerandmusic]

another white paper on nyquist and the theorem was for telegraph systems and should not be taken at face value...as it is for bandlimited signals.

 

The more i read, the more i believe that i would want to sample more than just twice the highest frequency.....unless you just want "good enough".

 

http://www.analog.com/media/en/training-seminars/tutorials/MT-002.pdf

[beerandmusic]

1 hour ago, mansr said:

Digital oversampling is a convenient way of constructing an accurate DAC. It's an implementation detail, nothing more.

 

I would hate to think things are implemented for any reason other than SQ?

So based on that statement, i will assume that oversampling can improve SQ....

 

[beerandmusic]

9 minutes ago, jabbr said:

 

That’s fine. A certain degree of over engineering is reasonable. So let’s say record at 24/96 or 24/192 ... but recording beyond that is really really hard pressed to justify any benefit (unless you were doing a ton of post-processing — even then)

 

On the other hand, upsampling for the purpose of improving the DAC is totally reasonable and an excellent technique — in fact the argument that you don’t need it would be the harder one to justify.

 

I know many people here upsample everything to quad dsd and suggest it is notably better.

 

I really don't understand what band limited is, but even when i gave the example of the 10000 generators, MANSR stated a needed criteria of upsampling higher that i didn't really understand.

 

Regardless, I think i will upsample all my pcm (which i rarely listen to anyway) to 192K...its free to do anyway, so why not...

 

My main objective for this thread really is a totally different issue anyway, and really had to do with why an SACD sounds better than a CD, and i thought everyone believed that, but apparently not.......i guess i need to go back to living in my thread "is everything debatable". 

 

 

 

 

[beerandmusic]

1 minute ago, mansr said:

Note also that this is slightly incorrect. The sampling frequency must be greater than twice the highest frequency component in the signal. Exactly equal doesn't cut it.

criteria of "bandlimited signal" (like telegraph system, which the theorem was initially designed for)

[beerandmusic]

3 minutes ago, mansr said:

So band-limit the signal before sampling. The theorem is generic. It applies to everything, including things that haven't been invented yet.

 

so you can band-limit (whatever that means -grin), or you can just increase the sample rate?

 

I agree with JABBR, WHY NOT upsample?

 

[beerandmusic]

3 hours ago, yamamoto2002 said:

 

If bit depth is infinite, it is possible.

 

I calculate real world example about 44.1kHz 16bit PCM data of 1/100th of second (441 samples).

It can store the difference of

1000.0000000 Hz from 1000.0000010 Hz but

1000.0000001 Hz signal is rounded to 1000.0000000 Hz.

By increasing bit depth to 24bit, frequency precision increases by 256 times.
 

 

why would you need 44.1khz sample if only frequency is 1000hz?  wouldn't you just need a sample rate of 2.1K?

and why would you need an infinite bit depth?

[beerandmusic]

17 minutes ago, Fokus said:

 

The theorem predates Shannon and Nyquist with many decades, putting it firmly before the era of electrical communications. It belongs to the area of pure mathematics and was not designed or invented. It was discovered, then proven.

 

proven for band limited signal

[beerandmusic]

6 hours ago, firedog said:

He apparently has an obsession with his idea of resolution and accuracy and has almost dropped it, and then goes looking on the net for material he doesn't understand so that he can come back and say what we've been telling him is wrong. 

 

It's pretty much time to give up responding to him. 

 

And yes, beerandmusic, there are conditions for the Shannon -Nyquist theorem that are never actually met in real life. That means nothing as we can get so close to the conditions that the gap is meaningless.

If you don't believe that, I suggest you stop using every bit of technology you own, and never get in a car or on a plane or train. It's all based on "almost" getting there in calculations.

NASA got to the moon using calculations made by a slide rule, because it was "accurate enough" that in real/practical terms you could call it 100% accurate.

 

Actually, i accepted that the nyquist frequency can capture accurately all possibilities,...

 

but shortly after accepting it, i later rejected it.

 

After thinking about it more, I am back to my original thinking that more samples will bring more accuracy to the point where technology is not able to process without error...i keep thinking about in real life about the infinite possibilities and infinite time slices and the inability to playback or even record with 100% accuracy.  It came to me while i was laying in bed and hearing my wife moving around and cooking in the kitchen and outside background noise, and my ears as microphones able to distinguish location and details that i know that NO reproduction is capable of.  There is no way that anything less than an infiinite sampling and infinite time is able to reproduce more accurately.

 

No theorem will convince me otherwise.

[beerandmusic]

Just now, mansr said:

I'm so sorry.

I am too...men are so feeble

[beerandmusic]

1 hour ago, mansr said:

Yes, DSD256 is better still.

better than DSD64 but not better than pcm 44.1K?

[beerandmusic]

3 minutes ago, pkane2001 said:

 

Happy for you. Resist all science, it is made up by EEs who pretend to know stuff just to demonstrate their superiority.

 

EE's yawn...they are still mortal

[beerandmusic]

34 minutes ago, yamamoto2002 said:

 

44.1kHz sample rate is not necessary to capture 1kHz signal. This is just a test setup. 2.1kHz sampling PCM can capture every detail of 1000Hz signal if AD converter is well designed.

 

Infinite bit depth is often assumed to understand a effect of sampling of AD converter, on the desk, when considering the nature of the signal ②:

 

(Analog input signal) → lowpass filter → ① → sampling → ② → quantization → ③(PCM output)

 

This thread is about sampling so engineers talk about sampling. "sampled signal" is ②. ② has infinite bit depth and ② can express continuous frequency, any rational and irrational number frequency exactly.

 

Because PCM data ③ is sampled and quantized, PCM data has artifact caused by quantization bit depth, it is called quantization error.

Rounding happens and quantization error occurs. On musical signal input, quantization error becomes persistent whitenoise-like hiss noise. Persistent hiss is not heard from every CD, this means noise caused by quantization error is sufficiency low and 16-bit bit depth is sufficient for listening.

 

I tested reducing bit depth of 44.1kHz 16bit PCM (ripped from music CD, Stravinsky Le sacre du printemps) and found, reducing bit depth to 13 bit (without dither) causes very subtle, constant hiss. Reducing to 12 bit causes more obvious hiss.

 

Thanks for sharing.

Just curious, in your opinion can any format above 44.1kpcm 16 bit able to provide better SQ, whether it be higher rate pcm or dsd, given all possible complex waveforms where maximum frequency is less than 20khz?