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

[elcorso]

The number of options is bewildering.

 

To simplify, my approach is:

 

1) Try to obtain recordings/works in the native format of the mastering engineer

 

2) Convert these files to the native format and resolution preferred by the particular DAC

 

+1

 

That's what I do.

 

As I always listen to music in DSD format, I do it through HQ Player, even if the music recorded in PCM is sold in DSD too, because I like HQ Player algorithms more.

 

Roch

[Jud]

If you read Lavry's papers he isn't hiding anything. He talks about how his own converters are delta sigma and run in mhz range. Even then what he talks about as getting less accurate sampling at too high a speed and fewer effective bits is still in effect. He warns the reader to keep both sampling rates clearly separate in their minds when trying to understand how this works. Doesn't sound like someone pulling a fast one if you read it rather than what others say about the paper.

 

http://lavryengineering.com/pdfs/lavry-sampling-theory.pdf

 

 

Perhaps it's just the way Lavry's writing strikes me, but on re-reading I once again feel his arguments are far more slippery than clear.

 

He implies the decimation from the original SDM processed bitstream at MHz rates to RedBook at 16 bit 44.1KHz is needed for "accuracy," when in fact it is marketing-driven by the need to convert to CDs for sale. In doing this he completely bypasses the issue of whether the distinction he repeatedly makes between "audio sample rates" (the rate of the audio data) and "other sample rates" (such as the rates used by DACs internally) needs to exist at all.

 

The paper states (at page 23 of 27) "In the case of DA converters, the data is interpolated to higher rates which help filtering and response. Such oversampling and up sampling are local processes and tradeoff aimed at optimizing the conversion hardware." [Emphasis added.]

If the original "audio sample rate" doesn't go through the decimation step but stays at the original higher rate, then one gets to "help filtering and response" and "optimiz[e] the conversion hardware" without requiring the "local processes" of oversampling/upsampling. And if these higher rates do indeed "help filtering and response," then where exactly is the tradeoff between speed and accuracy he's trying to sell?

[Miska]

If the original "audio sample rate" doesn't go through the decimation step but stays at the original higher rate, then one gets to "help filtering and response" and "optimiz[e] the conversion hardware" without requiring the "local processes" of oversampling/upsampling. And if these higher rates do indeed "help filtering and response," then where exactly is the tradeoff between speed and accuracy he's trying to sell?

 

Yes, that's why we should be using hires as much as possible. Be it PCM or DSD. Much less back and forth massaging of the data. Filters can be much more relaxed, for DSD, simple 1st order antialiasing filter is enough at ADC side. For DXD it can also be pretty simple. RedBook needs the most massaging at both ADC and DAC side to get something proper.

 

If all content would be native DSD256 recordings...

[Ralf11]

re: it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible frequencies.

 

I would say (based on a post somewhere up above) that it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible [sine wave] frequencies.

 

i.e. out to 20 kHz

 

if there is an issue with the 1930s research that established that limit, I bet that is for transients

 

 

as for bewildering options - I find it pretty simple: I start with redbook and avoid lossy compression (maybe the psychoacoustic models for lossy are fine, but the cost to me is about zip to have non-compressed redbook).

 

Now, once I have the music I want on redbook; I look at the particular things that others say sound a lot better than what I already have. I don't buy it if it costs $100 for a CDRSACDHDCDwhatever, but if it is something I really want and costs me $20 or $30 then why not?

 

I figure (as per some above posts I made) that I am most likely buying better recording techniques, remastering, etc. but I don't care. If it really IS bit rate or bit depth, that's fine too and eventually the world will be re-Ponocized.

[Jud]

re: it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible frequencies.

 

I would say (based on a post somewhere up above) that it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible [sine wave] frequencies.

 

i.e. out to 20 kHz

 

if there is an issue with the 1930s research that established that limit, I bet that is for transients

 

Careful. I believe the Sampling Theorem works for non-periodic signals just as well. They must still obey the rule that the sampling rate must be more than double the highest "frequency of interest," which in the case of transients would require sampling at more than double the rate of change of the transients. But I am not aware of undisputed research that we can hear transients with rates of change greater than 20,000Hz. If there is such research, I'd be grateful to see it or a link to it.

[Jud]

Hi Jud,

 

Good post, thanks for the explanation (TBH... I get a bit lost when talk moves to impulse / ringing / etc... that's way over my pay grade).

 

 

 

Ringing has to do with something called the Gibbs Phenomenon, which I don't understand either , and which again has to do with math: https://en.wikipedia.org/wiki/Gibbs_phenomenon .

 

There are a couple of funky things about the ringing associated with the Gibbs Phenomenon -

 

- For the filters used in digital audio, the ringing occurs at ultrasonic frequencies, so audible effects, if any, must be in the time domain. Miska and others have described the effect as "smearing" the signal in time.

 

- Ringing occurs both before and after the main impulse as long as the filter is linear phase. Even though the notion is a bit brain-twisting, this does not violate cause and effect, but it does mean the impact of the ringing (e.g., the "smearing") will occur both leading up to and after the main impulse. Minimum phase filters move the ringing/smearing mainly or exclusively after the impulse. This is one of the reasons minimum phase filters are used, to minimize or eliminate "smearing" prior to, and thus the sonic impact of, the leading edge of sharp impulses/transients, on the theory that "smearing"/ringing on and after the trailing edge of the impulse/transient won't sound terribly unnatural.

[Abtr]

re: it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible frequencies.

 

I would say (based on a post somewhere up above) that it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible [sine wave] frequencies.

 

i.e. out to 20 kHz …

The problem may be the relative timing of tones. You can distinguish e.g. multiple HF tones based on their relative timing or relative phase. The smallest audible timing differences are measured in microseconds and simply cannot be encoded using a 44.1KHz sample rate. Some analog recordings might do better..

[Ralf11]

could be - my point is simply that we don't know for sure (not that it matters much for the reasons I posted above)

[esldude]

The problem may be the relative timing of tones. You can distinguish e.g. multiple HF tones based on their relative timing or relative phase. The smallest audible timing differences are measured in microseconds and simply cannot be encoded using a 44.1KHz sample rate. Some analog recordings might do better..

 

Please..not this bull carp again. Redbook can encode timing to within about 55 picoseconds accuracy. It is not limited by the time between samples.

[esldude]

 

 

Perhaps it's just the way Lavry's writing strikes me, but on re-reading I once again feel his arguments are far more slippery than clear.

 

He implies the decimation from the original SDM processed bitstream at MHz rates to RedBook at 16 bit 44.1KHz is needed for "accuracy," when in fact it is marketing-driven by the need to convert to CDs for sale. In doing this he completely bypasses the issue of whether the distinction he repeatedly makes between "audio sample rates" (the rate of the audio data) and "other sample rates" (such as the rates used by DACs internally) needs to exist at all.

 

The paper states (at page 23 of 27) "In the case of DA converters, the data is interpolated to higher rates which help filtering and response. Such oversampling and up sampling are local processes and tradeoff aimed at optimizing the conversion hardware." [Emphasis added.]

If the original "audio sample rate" doesn't go through the decimation step but stays at the original higher rate, then one gets to "help filtering and response" and "optimiz[e] the conversion hardware" without requiring the "local processes" of oversampling/upsampling. And if these higher rates do indeed "help filtering and response," then where exactly is the tradeoff between speed and accuracy he's trying to sell?

 

Well this was written in 2004 by someone who mades ADCs and DACs for pro audio use. At that time CD was the only pertinent market to orient toward. SACD was not big enough to worry about. Downloads of hirez material weren't available, and the market for such recordings on DVD based discs never took off either.

 

Even if now there is no limitation on the accuracy at the high rates it is Dan Lavry's opinion anything over 96 khz is unnecessary. At one time 96 khz was less capable even over sigma delta chips. I have a Focusrite recording interface which uses a balanced DAC chip able to work up to 192 khz. The device is only setup to work to 96 khz rates. Focusrite said at the time this was because the DACs didn't work as well at 192 khz and measured results suffered. They recently released mark II version that do work at 192 khz. Same DACs. Same specs. No specs specific to 192 khz other than latency. So it is possible nothing was changed other than turning on 192 khz operation. And 192 khz may still not work as well. I don't know for certain of course.

 

The idea to keep pushing sample rates up both in the operation of delta-sigma designs and in the PCM or DSD available rates is an easy selling point. Just like megapixels for cameras. I would be more convinced if someone had rational explanations for what point it sounds so good no improvement can be detected with higher sample rates. So far it is just higher is better as if the improvement can be extended forever with more more more.

[Abtr]

Please..not this bull carp again. Redbook can encode timing to within about 55 picoseconds accuracy. It is not limited by the time between samples.

Of course it is. Imagine two equivalent 10 KHz pure analog sinusoidal tones, one started 100 microseconds before the other. A digital (PCM) 44.1/48 KHz sampled recording of this will sound different from a higher resolution digital recording..

[mansr]

Of course it is. Imagine two equivalent 10 KHz pure analog sinusoidal tones, one started 100 microseconds before the other. A digital (PCM) 44.1 KHz sampled recording of this will sound different from a higher res digital recording..

 

What makes you say this?

[Ralf11]

this timing argument is interfering with the other arguments here

 

you guys have 55 picoseconds to wrap it up or start your own delay phase thread

[Abtr]

What makes you say this?

Try it..

[gmgraves]

Neither it seems can my ears come up with a sonic comparison.

 

I wonder if someone will chime in with something really solid that'll start me wondering if my brain has decided that hires is pointless and expectation bias is confirming my preference for Redbook?

 

What kind of music do you listen to? I ask because I notice the difference mostly (but not exclusively) on low level material. If one listens mostly to pop and rock where the dynamic range of the music itself ranges from loud, to louder and loudest, or, as in most modern CDs has had what dynamic range there is compressed to within an inch of its life, there isn't much in the way of low-level detail or large room ambience. Here, the difference between 24-bit and 16-bit is really profound. While my listening choice is mostly classical, I also like jazz and listen to it a lot. I have to admit that on most jazz recordings, I find it often difficult to impossible to tell 16-bit Red Book from 24-bit. There just isn't a lot of low-level material or hall ambience in that genre either to show off the advantages of the higher bit rate. Of course, if one compares the Red Book version directly with a real 24-bit (or DSD) version, you should notice a general increase in cleanliness of the higher sample rate version that one might not notice without a direct comparison. Even then, there are a lot of variables in a commercial release of either, that some might argue could account for that difference.

;-)

... seriously though, if someone gave me my entire library in 24 bit (if that was possible) I'd obviously accept the gift and use it... but I'd never contemplate doing any work myself to get (any of) my library in hires. On my system that would seem a waste of time.

 

Well, the difference between a Red Book CD of a recording and a high-res version of same certainly is not on the order of the difference between a master tape and the LP made from it, so the high-res version most assuredly doesn't make the Red Book version unlistenable. Enjoy.

[esldude]

Try it..

 

Think about those numbers you picked for a moment please. 10,000 hz tones offset from each other by 100 microseconds.

 

It makes no difference really, but are you talking one tone in each channel or both in the same channel?

 

Hint:___this isn't going to work out the way you think it will.

[esldude]

this timing argument is interfering with the other arguments here

 

you guys have 55 picoseconds to wrap it up or start your own delay phase thread

 

Come on now. Resolution in time is what higher sample rates give along with wider bandwidth. I don't think it is important, but that is the difference.

[Ralf11]

it was a joke

[gmgraves]

Thx, unfortunately only one of those is a peer-reviewed publication. They could still be correct, but your claims are not supported, esp. by the Stereopile 'thing'. I'll take a look at the one AES later on...

The links were not provided to sway you either way. As my last line says "This series of links should give you enough info on both sides of the argument to make an informed decision on the value of the Meyer/Moran study."

Is that the same Bob Stuart that everybody dumps on for trying to foist MQA on us? ;]

Yes.

Similarly, there some studies showing that mp3 files cannot be distinguished from redbook. It costs me nothing to use Apple Lossless instead tho (well, ok - I guess it cost me a little more for a higher capacity iPhone).

 

Actually, at bit rates of 192 kbps and higher it is extremely difficult to tell the difference between Red Book and MP3. Listen, some Saturday evening, to the live Boston Symphony Orchestra webcast from WCRB in Boston or listen, on demand to the past concerts as listed on the page referenced below:

 

The Boston Symphony Orchestra | 99.5 WCRB

 

These are 192 kbps and sound damn good on speakers. I do hear occasional compression artifacts on headphones though, but then, I know what compression artifacts sound like. These broadcasts are virtually indistinguishable from 16-bit LPCM as on a CD.

 

I'm fine with using higher bit rates or depth as long as I don't have to pay a lot for it.

 

Reasonable response

[jabbr]

@elsdude is correct, if the system is only responsive to 20khz then 44 khz is enough to sample the relevant transients. Conversely, if the system is sensitive to faster transients then it is *by definition* responsive to > 20 khz.

 

I'm not sure the 20khz limit is set in stone. That could be the limit for detecting isolated tones but since the (human auditory) system is highly nonlinear, that doesn't mean the entire system's response is limited to 20khz.

 

Consider the retina: although not sensitive to infra-red, the skin is :-)

 

It is possible that 30 khz frequencies are interpreted as "irritating" or other nonlinearities such as the response of amplifiers/electronics to "ultrasonics". Consider why we need to filter out > 20khz from DAC outputs: the system itself is sensitive to ultrasonics. These ultrasonic effects on electronics (that manifest themselves in the audible range) are examples of nonlinearities in the electronics -- similarly the auditory system itself may have (other) nonlinearities.

 

As they say, when you assume ...

 

In any case we really aren't doing science here so folks should judge for themselves. I find that DSD256 goes well with scotch.

[gmgraves]

re: it's a fact that a 44.1KHz sample rate is mathematically adequate to reproduce all audible frequencies.

No doubt about that. But it isn't all about frequency response. There are other criteria that define fidelity.

[jjjjusta]

I wonder, too, if much of the perceived benefit of DSD comes from 1) identifying artifacts and knowing you can reliably hear them and then 2) hearing artifacts as evidence of inferior fidelity. So you're listening for artifacts instead of music *most* of the time. As a musician and engineer I'm always having to zoom in and out to have a faithful idea of what the thing actually sounds like. There is a point in hi-fi where you are missing the message.

 

Sent from my SAMSUNG-SM-G891A using Tapatalk

[r_w]

LOL, that covers Steve Gadd's drumming then.

 

Please..not this bull carp again. Redbook can encode timing to within about 55 picoseconds accuracy. It is not limited by the time between samples.

[r_w]

Hi George,

 

The main comparison I did was Weather Report Heavy Weather 24/176 vs my own CD rip... both versions sound perfectly enjoyable to the extent that I wouldn't use up extra disc space for a noticeable difference. Both versions sound tonally very similar - that album was always a little bright and thin (certainly not fat and buttery) - but it does sound very airy.

 

I listen to lots of classic rock, jazz fusion, some propg rock, good 80s rock/pop (Tears For Fears, etc).

 

I don't do classical very often, but given how everything else sounds I don't see why it would be represented poorly.

 

I think the key thing appears to be how well the DAC does redbook... the VEGA seems to excel in this area and that's why I'm enjoying it so much - and as I've voiced before, the system just seems to be so much more forgiving that any other system/config that I had before.

 

I will continue to enjoy... but won't rule out hires in case I hear something I've been missing.

 

 

;-)

 

What kind of music do you listen to? I ask because I notice the difference mostly (but not exclusively) on low level material. If one listens mostly to pop and rock where the dynamic range of the music itself ranges from loud, to louder and loudest, or, as in most modern CDs has had what dynamic range there is compressed to within an inch of its life, there isn't much in the way of low-level detail or large room ambience. Here, the difference between 24-bit and 16-bit is really profound. While my listening choice is mostly classical, I also like jazz and listen to it a lot. I have to admit that on most jazz recordings, I find it often difficult to impossible to tell 16-bit Red Book from 24-bit. There just isn't a lot of low-level material or hall ambience in that genre either to show off the advantages of the higher bit rate. Of course, if one compares the Red Book version directly with a real 24-bit (or DSD) version, you should notice a general increase in cleanliness of the higher sample rate version that one might not notice without a direct comparison. Even then, there are a lot of variables in a commercial release of either, that some might argue could account for that difference.

 

 

Well, the difference between a Red Book CD of a recording and a high-res version of same certainly is not on the order of the difference between a master tape and the LP made from it, so the high-res version most assuredly doesn't make the Red Book version unlistenable. Enjoy.

 

[r_w]

I agree mate, same as you.

 

... not criticising anyone but that's why I always bang on about building systems that make as much of the library sound great as possible... not just a few golden recordings - that pursuit really is missing the message in the music.

 

:-)

 

I wonder, too, if much of the perceived benefit of DSD comes from 1) identifying artifacts and knowing you can reliably hear them and then 2) hearing artifacts as evidence of inferior fidelity. So you're listening for artifacts instead of music *most* of the time. As a musician and engineer I'm always having to zoom in and out to have a faithful idea of what the thing actually sounds like. There is a point in hi-fi where you are missing the message.

 

Sent from my SAMSUNG-SM-G891A using Tapatalk