How many bits, how fast, just how much resolution is enough?

[esldude]

Bummer you feel that way. This is a logical question given what he said.

 

 

 

1. 192kHz considered harmful192kHz ... The effect is very slight, but listening tests have confirmed that both effects can be audible.

 

 

2. Empirical evidence from listening tests backs up the assertion that 44.1kHz/16 bit provides highest-possible fidelity ... Not one listener throughout the entire test was able to identify which was 16/44.1 and which was high rate [15], and the 16-bit signal wasn't even dithered!

 

I understand your complaint. However, item 1 is can be. It can be demonstrated. Does not mean it will be or always is. I do think he is over-hyping this problem. I have found it to be a non-issue with any music I ever cared to listen to in fact.

 

Item 2 is no one heard a difference they could demonstrate. And in fact one of the complaints about that test is the hirez samples either weren't sourced from hirez or didn't have enough high frequency content to show differences. So another group of people trying to have it both ways.

[esldude]

Ah, yes, wrong example. Make that two 10KHz tones with a 1 (or 99) microsecond relative time delay. Would 48K - and (e.g.) 96K samples/second PCM recordings be indistinguishable in the analog domain?

 

 

Yes.

[esldude]

Can you prove it?

 

I would say yes. Instead let me ask, what would convince you?

 

I can send you files as you described and let you listen to see if they sound different. They won't.

 

There are two things you have in mind which are making you think they would be different which are misconceptions about how digitally sampled audio works.

 

First that you can't discriminate in time between sample points. You are missing that if a wave starts at a slightly different time, even in between samples, it generates a different value in all subsequent samples than if it started right on a sample. The reconstructed wave is continuous and will reconstruct the waves offset by the correct amount of time even between samples.

 

The other is that higher sample rates somehow are more accurate on below 20 khz signals because they have more sample points. That isn't how it works. The higher sample rate gets you more bandwidth. Both 48 khz and 96 khz will recreate a 10 khz signal accurately enough you won't see an analog difference.

 

So what would convince you?

[esldude]

Thanks, I was hoping for mathematical proof I'll consider your arguments..

 

Shannon-Nyquist theorem. That is the mathematical proof.

 

View the video linked in my signature. You can skip to the 20 min 50 sec. mark, and watch about two minutes. Using analog sources and analog monitoring gear with AD/DA in between he shows you can move a band-limited square wave thru various amounts of delay between sample points and see the wave shape you get on the analog o-scope is exactly the same other than moving in time relative to a second squarewave. What more proof could you want? You have the theorem predicting something, and an analog monitoring system showing the theorem works as advertised.

[esldude]

I mean two simultaneous analog tones.

 

I still think you are missing his question.

 

Which of these are you thinking about:

 

 

A: one 10 khz tone in the left channel, combined with one 10 khz tone delayed by 15 microseconds also in the left channel.

 

 

B: one 10 khz tone in the left channel and one 10 khz tone in the right channel delayed by 15 microseconds in the right channel.

 

 

A: isn't going to work like you think. Identical tones out of phase with each other super-imposed partially cancel out leaving a single reduced level 10 khz tone. This can be done, but all you see if a lower level 10 khz tone with different zero crossing points. If such tones get 180 degrees out of phase they fully cancel out. If completely in phase they add together.

 

B: can be done, has been done, they do something like it in the video I suggested. I have done it, others have done it, so what would convince you it works?

[esldude]

...that you cannot implement exactly in real world. It only conveniently assumes infinitely long signals with filters that are infinitely long ,so that nothing ever comes out of the filter, because it also has infinite delay. And assumes infinite precision of timing and resolution of the samples. But other than that, yes, it works nice...

 

 

 

IIRC, pretty heavily oversampled converters. Which makes pretty big difference...

 

Here's 19k sine wave from a NOS DAC running at 44.1k sampling rate:

[ATTACH=CONFIG]31158[/ATTACH]

 

And the same source data, same DAC, but now upsampled to 384k sampling rate before sending to the DAC:

[ATTACH=CONFIG]31159[/ATTACH]

 

So you certainly want to have the conversion running at higher rate than 44.1 kHz...

 

 

P.S. DPO-scope is good way to see how stable the waveform is, which also tells quite a bit about the reconstruction that you don't see in old-school scope.

 

I must say I am surprised and dismayed at you Miska. If that is from the Musette at 44.1 it simply indicates a bad DAC. I don't have any images, but early DACs have displayed 19 khz with good waveform shape long ago in the early days of CD because I have seen them do it on my oscope. If your NOS DAC needs 384 khz to do a 19 khz wave, you should forget it as a poor design.

 

The video uses an inexpensive ADC/DAC from around the year 2002. So probably a sigma delta chip. If such chips give PCM 44.1 khz results with a nice clean wave and the NOS can't then it suggests problem for the NOS, not some blanket dismissal of results using modern DAC chips that aren't handicapped. Earlier in the video he steps 1 khz at a time from 15 khz to 20 khz and you can see the image Miska posted has no bearing on what is possible with modern DACs. (well calling a 14 year old low end DAC modern)

 

Here is the analog scope image of 18 khz from that video. He moved the camera, but you can see 19 khz and 20 khz is just as nice.

 

 

Actually looks nicer than your 384 khz image. Though I understand that is just an artefact of the display.

 

Really intentionally deceptive on this one Miska. Shame on you.

 

Edit to add: It appears the Musette (assuming that is the NOS you are showing) doesn't use output filtering at all. Not digital and not analog. So you need to run 384 khz to let the rest of your gear, your speakers, and the 6th order filter of your ears perform the filtering. Terribly deceptive to use that as an example. Another name for a DAC with no output filtering is a broken DAC. Most certainly completely ignores bandwidth limiting required by Shannon-Nyquist.

[esldude]

It's beside the point you are making (that more samples aren't needed to accurately represent the waveform), but is possibly still quite relevant to the OP's question about whether a higher sample rate makes for better digital audio.

 

Can we agree that the engineers who designed 8x oversampling into DAC chips decades ago did so for solid engineering reasons? And if that's granted, then can we go from there to saying there is no engineering reason for a decimation-interpolation sequence in the middle of the recording/playback chain, but this exists solely as an artifact of the way the music industry has evolved?

 

Well if PCM 48/24 (or 96/24) is enough, then bits and sampling at that are enough whether via some purist version that performs as it should or whether you have in between DSP. If we go straight to some higher rate and higher number of bits transmitted that sounds like an unengineered method that will work. An engineered method is more efficient.

 

That is before we mention recording, mixing, mastering, and processing being easily doable in PCM. Then upon playback digital volume control and Room EQ that sort of thing also being easily handled in PCM formats. If the cost is some measured inaudible improvement and simplicity at the cost of those things plus a higher bit thru put, doesn't sound like a clear cut great trade off to me.

[esldude]

Re: comparing classical... see if you can do the same experiment on an Auralic VEGA?

 

... if someone want to make available 2 files of classical 16/44 and hires, I'll try it out.

 

Luckily the UK is also a free country.

 

:-)

 

You could try some of these:

 

2L High Resolution Music .:. free TEST BENCH

[esldude]

Which two files would you recommend I compare?

 

(I only want to do this once... comparing files isn't what music enjoyment is about)

 

:-)

Don't have any suggestions. Download all the 44 khz versions as they are smaller. Pick a couple you like and pick one of the highest resolutions you can play.

 

Sent from my Nexus 6P using Computer Audiophile mobile app

[esldude]

Wow, talk about making a virtue of necessity! The engineers who developed 8x oversampling within a very short time after digital audio was first marketed were not operating from an "unengineered method," they were engineering the solution that gave acceptable results at greatest cost and resource efficiency. 352.8/384KHz sampling rates are not "some purist version," but what was acceptable performance to non-audiophile engineers working for major consumer corporations trying to sell to the mass market decades ago.

 

Of course you're correct that currently at least engineering of recordings is more easily done in PCM. That's the 8x oversampling I was mentioning. So the "debate" would be between maintaining DXD (8x) rates all the way through and DSD conversion, which allows less expensive filtering (that's why delta-sigma has become ubiquitous, it's cheaper for what we'll call, without wishing to get into the old PCM-DSD debates, a substantially equivalent result).

 

I live in a semi-rural setting a few miles from a town of 4,000 people, and the download speed from my ISP (a cable company that is a relatively small local family owned operation) is 50mbps, soon to double to 100. I can buy a 4TB external HDD for a little over $100. When you talk about storage space and bandwidth, you're really reprising my argument that the RedBook standard is an artifact of history - digital audio was developed when storage space and bandwidth were much greater limitations than they are today for most people in a position to be concerned about things like the sonic impact of boutique power supplies.

 

 

 

 

Indeed they may not be best practices today. Do you think those practices would have evolved in the direction of *more restricted* capabilities than a couple of decades ago, or greater capabilities?

 

Your only making our point for us Jud. I didn't imply that going to 8x oversampling filters was un-engineered. Given the 48/24 PCM format and doing what it takes to reach results at the very edge of theoretical perfection is excellent engineering. They built black boxes that take the relatively low bit rate in, and put out near perfect results. What happens in between, delta-sigma etc etc. is the engineering part.

 

Now using DXD at 384 khz/24 bit is the sledge hammer approach. And for what can only be called minimal or perhaps non-existent benefits. The mis-guided purist idea that since we end up at 384 somewhere anyway, we might as well just do 384 khz is the issue.

 

Do the saved bits matter? I live in a suburban area with 6 mbps internet near a town of 150k people. You only get close to that between 1 am and 6 am. Otherwise about half. On Sundays and holidays you get sporadic 400 k service. So no at least for some of us redbook is not yet an artefact of history. And if redbook or slightly beyond redbook is done well enough there is nothing audible to gain, then there is nothing audible to gain. An idea that doesn't market well.

 

What about multi-channel? I don't find multi-channel an incredible advancement. It is however a good solid genuine improvement which one has no trouble whatsoever hearing as such. Should I go 384 khz for stereo or with similar bandwidth uncompressed 7.1 multi-channel with 48 khz for each channel? Heck we'll just do 384 multi-channel or DSD 256 as Miska has already told us it works fine. In fact I see no reason not to just bump to DSD512, you know, just to be sure. DSD1024 will be relatively available in 3 years time I believe. Maybe a bit soon for that yet. It will take time, but we'll get to that 20 ghz I mentioned earlier.

[esldude]

Britten: Frank Bridge Variations - Romance TrondheimSolistene

 

Compared:

CD 16BIT/44kHz - 8 MB

24BIT/192kHz - 42 MB

DSD 1285.6448Mbit/s - 132 MB

 

Me: Difference is so miniscule it's hard to even describe and therefore in my system not worth any extra file size

 

Wife: "All sounds the same to me"

 

... end of experiment (and any fear of Redbook not being able to supply me with fabulous music (on the Auralic VEGA)).

 

;-)

 

Same results I get and when I play it for others or myself.

 

I've done some recordings of musician friends at both 48khz and 192 khz. No processing done. As pure as can be done. Played both, and asked which was better. Played over a good system too mind you. I got sort of blank stares with comments like, "both sound fine", "what's supposed to be the difference", "I don't know, you listen to that kind of stuff more than we do, which is the best one to you". If there were large benefits I don't think you would get those responses. You didn't get that kind of reaction when you first played DVD to someone who only had VHS tape. You didn't get that response when you played Blurays to people with DVD. You don't get that response when people with HDTV see 4K.

[esldude]

Ah, MQA is going to be your favorite thing!

 

Only kidding folks. We now resume the originally scheduled programing.

 

Well even MQA isn't enough to fix my situation. You are right however, if MQA can really make a difference at the smaller bandwidth it fixes a real problem. Of course there is more going on than just that with MQA. Maybe I need MinQA.

[esldude]

Oh, I've seen many claims that human eyes are not accurate enough to tell difference between 1080p and 4K...

 

Many people just haven't learned how to listen, they are much more trained in using their eyes than their ears. Just ask any blind person...

 

Specious reply Miska.

[esldude]

While the topic has arisen, maybe now is a good time to say something about phase. With delta-sigma DACs and digital filtering you really don't have phase issues with the great majority of DACs out there today. At least not to 20 khz you don't. So worry about phase at 44 khz or 48 khz rates is a non-issue. When it can become an issue is boutique DACs that decide to use a different filter that alters the flat response below 20 khz or has some other alleged benefit.

 

Here is an article from way back in 2004 showing that phase is fine right up to 20 khz.

 

digital

[esldude]

As Abraham Lincoln famously said, "I read it on the Internet, so it must be true."

 

You know, I have read that before. Who would argue with old Abe other than pro-slavery folks like Stephen Douglas? So I'll just apply this idea to the CA forum so everything will be copacetic.