Understanding Sample Rate

[adamdea]

2 minutes ago, jabbr said:

That’s what you call a proof? Hmm ...

It is at least a very strong indication. 

[jabbr]

5 minutes ago, adamdea said:

I'm talking about the quantisation noise that has to be shaped by the noise shaper. Sorry if that was not obvious. You have 1 bit of resolution spread over 64fs that still only 4 bits in the audioband isn't it. Where do we put the other 12 bits of noise to get 16 bit performance in the audioband?

http://bitperfectsound.blogspot.ca/2014/09/noise-shaping.html

 

I'll defer to @Miska who may have some nice graphs of the effects of his actual noise shaping as HQPlayer upsamples from DSD64 to DSD512 and the effects on this "quantisation noise" -- and I routinely upsample from both PCM and DSD source to DSD512 to feed into the iFi Micro, for example.

 

So your argument assumes a specific implementation which is, I'd say irrelevant and proves nothing.

 

Not a proof, not even close. Check your assumptions at the door. I am not saying that there is proof either that ultrasonics have an effect on SQ but according to your own logic, the fact that DSD256-DSD512 upsampling sounds better than DSD64 would be "proof" that ultrasonics matter? According to your logic, you could "prove" whatever you felt like on any day of the week.

[adamdea]

1 hour ago, jabbr said:

http://bitperfectsound.blogspot.ca/2014/09/noise-shaping.html

 

I'll defer to @Miska who may have some nice graphs of the effects of his actual noise shaping as HQPlayer upsamples from DSD64 to DSD512 and the effects on this "quantisation noise" -- and I routinely upsample from both PCM and DSD source to DSD512 to feed into the iFi Micro, for example.

 

So your argument assumes a specific implementation which is, I'd say irrelevant and proves nothing.

 

Not a proof, not even close. Check your assumptions at the door. I am not saying that there is proof either that ultrasonics have an effect on SQ but according to your own logic, the fact that DSD256-DSD512 upsampling sounds better than DSD64 would be "proof" that ultrasonics matter? According to your logic, you could "prove" whatever you felt like on any day of the week.

Upsampling has nothing to do with this (especially not pcm upsampling). It doesn't have much to do with individual implementation either. I am talking about dsd 64.

 

 

Really what I'm saying about the noise problems of dsd 64 is not really controversial. The precise conclusions one draws are of course debatable, but the fact that dsd 64 produces a lot of noise just above 20kHz is not. Clearly I am not suggesting that a formnal logical or mathematical proof has been shown, but it is tricky to make a case for the region above 20Khz mattering if one is happy to pollute it too heavily. Like so many things in audio, it's only a problem if you think about it.

[jabbr]

14 minutes ago, adamdea said:

but it is tricky to make a case for the region above 20Khz mattering if one is happy to pollute it too heavily.

But conversely if I or most people who listen to DSD are noise shaping this region above 20 kHz, and notice a preference when doing so,  then does that suggest the region is important?

 

How many people who use HQPlayer have a preference for DSD64 over DSD128? How many people feed DSD64 into an NOS DAC?

 

Your argument is spurious.

 

[Miska]

2 hours ago, adamdea said:

Really what I'm saying about the noise problems of dsd 64 is not really controversial. The precise conclusions one draws are of course debatable, but the fact that dsd 64 produces a lot of noise just above 20kHz is not. Clearly I am not suggesting that a formnal logical or mathematical proof has been shown, but it is tricky to make a case for the region above 20Khz mattering if one is happy to pollute it too heavily. Like so many things in audio, it's only a problem if you think about it.

 

I'd be much more concerned about polluting it with leaky oversampling filters, such as used by MQA, that produce directly correlated images there. And for DAC chips, additional problem are the limited computational resources, so for most cases oversampling filters are anyway limited to 352.8/384k rate, thus producing strong images around those frequencies.

 

What happens with the modulator noise is up to the DAC to real with it. It is certainly not passed untouched from the DAC output. For DACs this is much more easier to deal with. In this scope, I'd be more concerned about class-D amplifiers that operate in similar way but where output filtering is much more challenging.

 

But from theoretical point of view, a perfect modulator at 64fs - one that is able to push all noise above the 1x band, would have 385.32 dB dynamic range in the 1x band. In practice, DSD64 modulators can reach about 192 dB dynamic range in audio band, equivalent of 32-bit PCM. Depending on implementation.

 

But I would be happy to see real-life measurements where "noise problems of DSD64" are demonstrated. I've been doing a lot of measurements of real world DAC outputs.

 

In addition, you can alternatively deal with DSD64 also by upsampling it to DSD256 or DSD512 before sending to the DAC. That is what I do (just like for all PCM content too).

 

And for the record, noise output from most new ADC chips look a lot like noise output of DSD128...

 

[Miska]

18 hours ago, mansr said:

That's unlikely. Audio equipment tends to use either linear phase (no distortion) or minimum phase (delayed high frequencies). What you suggest would mean some step in the processing used a filter skewed towards maximum phase. I've never heard of anything doing this, and I can see no reason why anyone would want it.

 

OTOH, most loudspeakers where tweeter is mounted on the same baffle with mid/bass driver, the tweeter leads the mid/bass in terms of timing. IOW, they are not timing coherent...

 

Planar electrostatics and speakers with coaxial elements (KEF, Tannoy) usually don't have this time coherency issue.

 

And some traditional designs offset this by moving tweeter further back and using waveguide on front. Which also improves the horizontal dispersion pattern with more consistent frequency response as the angle increases. Otherwise there's a sharp change in the directivity pattern at cross-over frequency. (bass is practically omnidirectional while highest frequencies are very directive)

 

Some of this can be offset with digital correction filters...

 

[adamdea]

3 hours ago, Miska said:

 

In addition, you can alternatively deal with DSD64 also by upsampling it to DSD256 or DSD512 before sending to the DAC. That is what I do (just like for all PCM content too).

 

And for the record, noise output from most new ADC chips look a lot like noise output of DSD128...

 

So you convert it to another digital format and then filter out the noise and then remodulate?. Do you do that via dsd wide or by converting to pcm on the way? 

 

Sorry I didn’t follow the last bit. Do you mean the output from multi level dac chips is still the same as dsd 128? 

[buonassi]

20 hours ago, buonassi said:

Wait a minute... I have this backward?  Minimum phase delays the highs relative to the lows?  So the lower frequencies arrive slightly before the highs?  I realize this happens on such a fast level that it's barely (if at all) perceivable.  But I thought minimum phase did the opposite - allowed the highs to arrive first, followed by lows - giving it the attack I think I'm hearing.

 

I can see where your statement is correct however.  Looking at @Archimago's phase graphs, the highs take on a positive phase shift, which would mean moving the sine ahead in time.  Is that a correct interpretation?

 

My narcissism showing here by quoting myself, but I did spend some time on Archimago's blog last night after signing off of CA.  It seems the phase shift is NEGATIVE for higher frequencies relative to low with minimum phase filter.  Third pic on this page:  http://archimago.blogspot.ca/2015/04/internet-blind-test-linear-vs-minimum.html

 

This would place HFs arriving first at the listener's ear (for single transducer headphones at least).  This is probably why I prefer some implementation of minimum phase.  I actually enjoy the attack of the higher frequencies, followed by that of the lows.  Being a musician and having listened to drums in real life, this is what I'm used to hearing:  a clear smack of the drum head when the stick hits it, followed by the fundamental tone of the drum, then the bass wave.  All very fast, mind you.

 

It makes sense to me in theory that this explains my preference, but, like most of my theories in audio, I'm open to other explanations.  

[buonassi]

12 hours ago, psjug said:

I've never tried to tell the difference like you are doing, but I'm pretty sure I would have a hard time telling the difference.  Maybe with headphones, but I never listen with headphones.  Loudspeakers mess with phase too.  Then there are room reflections, and probably other things to consider.  Since I'm an engineer, the fact that linear phase reconstruction filters are better at reconstructing the waveform appeals to me, so I'd probably choose linear phase if I make a choice of filter.  But really I don't care much about it.

Yes, I believe you'd be able to pick it out on headphones, and earphones as well, unless they are multi BA transducers that aren't phase aligned.  This is why I believe that the dragonfly red is so popular - being primarily a dac/headphone amp.  It's also why the ios devices have a minimum phase filter - headphones!

 

While differences are subtle, they become easier to pick out with very congested music with lots of drum work (prog, metal).  With linear phase, some people hear a smoother presentation, and that's all.  Others would hear better instrument separation as well.  With minimum phase, some would claim the sound becomes duller, with less precise imaging, less live sounding.  Each filter has its own characteristics and one isn't better than the other.  They are different, and appeal differently to people.  

 

Personally, I find the value in both depending on the type of music I'm listening to, and it's nice to be able to change it via upsampling (defeating my DACs oversampling filters).  It's almost a curse at the same time, because you have a near infinite number of options given all the parameters you can change.  I sometimes find myself obsessing over the filters instead of enjoying the music!  Even on the fly I'm subject to the torture of choice as my DAP has selectable filters.

[bogi]

On 3/1/2018 at 12:51 AM, adamdea said:

So you convert it to another digital format and then filter out the noise and then remodulate?. Do you do that via dsd wide or by converting to pcm on the way? 

 

With DAC chips that support the direct DSD path option typically no remodulation occurs. Remodulation of DSD input can be an option if volume control is required.

 

On 3/1/2018 at 12:51 AM, adamdea said:

Sorry I didn’t follow the last bit. Do you mean the output from multi level dac chips is still the same as dsd 128?

 

Miska mentioned ADC (not multilevel DAC) in his last sentence.

[Miska]

On 3/1/2018 at 1:51 AM, adamdea said:

So you convert it to another digital format and then filter out the noise and then remodulate?. Do you do that via dsd wide or by converting to pcm on the way? 

 

It goes just through rate multiplying remodulator that has suitable transfer function.

 

On 3/1/2018 at 1:51 AM, adamdea said:

Sorry I didn’t follow the last bit. Do you mean the output from multi level dac chips is still the same as dsd 128?

 

I'm talking about ADC side, not DACs.