Sample Rate of a Digital Audio Delivery Medium (i.e., CD)

Les Cargill  <lcargill99@comcast.com> wrote:

> The sheer distortion of transducers overwhelms the rest of the playback
> chain anyhow.

...

>> Are you saying the observations presented here about pre-ringing, pre-echo
>> or delay spread are not technically informed?  (As opposed to, difficult
>> to sell, a point to which I will cotton to.)

>No, I am saying that the perceptual/quality effects of them are 
>questionable at best and very likely moot.
>
>Speakers and rooms are much more of a problem.

In both the above quotes you are assuming that inaccuracies in one
part of the signal chain will mask inaccuracies in another part
of the signal chain.  This does not jibe at all with my own
experiences with audio.  

For example, a music recording made from the audience area of a concert
hall, containing both extraneous sounds and significant concert hall 
room acoustics, will subjectively sound superior when played 
back through high-quality speakers in a good room, as opposed to lower 
quality speakers/poor room.  

This makes perfect sense when you think about it -- the listener's
brain needs to filter out the extraneous sounds and and compensate
for the concert hall's reverberation in order to perceive the
instruments and/or vocals clearly; and this processing is much more 
difficult if the information is further muddled up by poor reproduction.

In audio, the "quality doesn't matter because it's distorted somewhere 
else" argument just does not work as a general argument.  In specific
cases it might be provable.  Maybe speaker transducers always cover
up pre-ringing, but maybe not.

Steve

Randy Yates  <yates@digitalsignallabs.com> wrote:

>Cheap delta-sigma D/As abound, don't they? That is, there is little need
>to STORE stuff at super-high sample rates on the digital audio delivery
>medium.

Unless you believe that the interpolation filters will pre-ring.

Steve

On Wed, 2 Nov 2016 19:24:43 -0500, Les Cargill
<lcargill99@comcast.com> wrote:

>Steve Pope wrote:

>Okay - for *end playback*, 16/44.1k is extremely entrenched and 
>perfectly adequate. 

For the non-demanding ear, 128 bps MP3 is 'perfectly adequate.'  It
depends on whether you're sensitive to the odd-sounding artifacts.

>Those are big things to move to have any sort of  change.

Only if your sole playback is CD format.

>It also doesn't help that you get high-profile folks like Neil Young 
>spouting nonsense about Pono.

That's your own judgement.  Not shared universally.

>> Are you saying the observations presented here about pre-ringing, pre-echo
>> or delay spread are not technically informed?  (As opposed to, difficult
>> to sell, a point to which I will cotton to.)
>>
>
>No, I am saying that the perceptual/quality effects of them are 
>questionable at best and very likely moot.
>
>Speakers and rooms are much more of a problem.

They are completely different problems, and arguably much more
natural-sounding to the human ear.  Pre-ring definitely is not.

It sounds like your mind is already set, but if not, look into
research being done by Meridian, including new formats that are
intended to be more ear-friendly.
 

On Sun, 6 Nov 2016 01:27:03 +0000 (UTC), spope33@speedymail.org (Steve
Pope) wrote:

>Les Cargill  <lcargill99@comcast.com> wrote:
>
>> The sheer distortion of transducers overwhelms the rest of the playback
>> chain anyhow.

>>> Are you saying the observations presented here about pre-ringing, pre-echo
>>> or delay spread are not technically informed?  (As opposed to, difficult
>>> to sell, a point to which I will cotton to.)
>
>>No, I am saying that the perceptual/quality effects of them are 
>>questionable at best and very likely moot.
>>
>>Speakers and rooms are much more of a problem.
>
>In both the above quotes you are assuming that inaccuracies in one
>part of the signal chain will mask inaccuracies in another part
>of the signal chain.  This does not jibe at all with my own
>experiences with audio.  

Interesting comments through this thread, Steve.  I completely agree.
Pre-ring is a very unnatural sounding form of distortion--hence the
latest interest in use of minimal phase filters.  But of course that
results in skewed phase, and the ringing artifacts are still there,
(just assumed to be masked by the tail end of the signal).  It sounds
like we've reached similar conclusions about that.

There's a humorous analogy in one of the DSP books (I wish I could
remember which) about having an army of musicians playing sine wave
instruments for eternity and ending up generating a single narrow
pulse.  And real-world implementation of 44.1k audio doesn't yield
complete fidelity from DC to 22.05k.

>In audio, the "quality doesn't matter because it's distorted somewhere 
>else" argument just does not work as a general argument.  In specific
>cases it might be provable.  Maybe speaker transducers always cover
>up pre-ringing, but maybe not.
>
>Steve

There's a prevalent and mistaken assumption by some in the audio
industry that simply being able to reproduce a certain frequency is an
adequate measure of good sound.  I've seen that a lot even on pro
audio sites--'debunkers' of high bandwidth recording citing Nyquist as
the end-all criterion.

It seems rather like giving someone a color-blindness test to
determine whether they'll appreciate fine art.  It may be a component,
but it's hardly a complete view on the subject.  The problem is that
perceived reaction to phasing/time-alignment, pre-ring, or even
'listener fatigue' is difficult to gauge or describe.

I've heard some systems set up by people who really know what they're
doing, playing well-recorded high-bandwidth audio, and the results are
impressive.  Not so much in-your-face striking in an audiologist
sense, but just very natural and more involving. Much easier to
appreciate the performers and musical events in the recordings.

On Thu, 27 Oct 2016 19:59:13 +0000 (UTC), spope33@speedymail.org
(Steve Pope) wrote:

>Tim Wescott  <tim@seemywebsite.really> wrote:
>
>> And second, because really steep filters ring.
>
>Bingo.
>
>> and my boys can hear sounds significantly above 20kHz, which is probably 
>> around where the ringing would happen if you're aiming to sample at 44kHz.
>
>This part I'm not following -- the filters presumably are linear
>(so the filter itself adds no new frequency components) and
>neither are they underdamped (existing frequency components
>are not amplified by the filter).
>
>The ringing artifact shows up as delay spread, as I noted above.
>Sounds components that were once separated along the time axis
>now overlap.  
>
>Steve

I always thought that this should be intuitive.  You send a pulse into
a steep filter (as in a CD player), and you get the impulse response.
No one argues that.  So what happens with a sharp snare drum hit?
I've heard some demos where this is very audible.  

Of course high bandwidth doesn't eliminate that. But it shortens the
time frame, makes it possible to use less aggressive filters, and
lessens the degree of phase skew when minimal phase filters are used.

Steve, thanks so much for your interesting thoughts. I always
enjoy reading your stuff.

Steve Pope wrote:
> Les Cargill  <lcargill99@comcast.com> wrote:
>
>> The sheer distortion of transducers overwhelms the rest of the playback
>> chain anyhow.
>
> ...
>
>>> Are you saying the observations presented here about pre-ringing, pre-echo
>>> or delay spread are not technically informed?  (As opposed to, difficult
>>> to sell, a point to which I will cotton to.)
>
>> No, I am saying that the perceptual/quality effects of them are
>> questionable at best and very likely moot.
>>
>> Speakers and rooms are much more of a problem.
>
> In both the above quotes you are assuming that inaccuracies in one
> part of the signal chain will mask inaccuracies in another part
> of the signal chain.  This does not jibe at all with my own
> experiences with audio.
>

All experiences are different. But mine are 180 degrees off from yours :)

An artifact that's a couple orders of magnitude more than another
artifact will very nearly always mask the lower artifact from what I
can tell. And specific to transient-oriented artifacts, the sheer
mechanical  slew rate of transducers overwhelms any mess from playback.

Technically, you're exactly right, though - it all adds up.

I cannot say, however, what pre-ringing might sounds like. It'd take
some serious consideration to design a test to find out.

> For example, a music recording made from the audience area of a concert
> hall, containing both extraneous sounds and significant concert hall
> room acoustics, will subjectively sound superior when played
> back through high-quality speakers in a good room, as opposed to lower
> quality speakers/poor room.
>

Unless the hall is a very good hall, it won't sound that good anyway
no matter what it's played back on.

Practitioners deal with this with spot mics. We're just not
acclimated to a lot of room sound any more.

> This makes perfect sense when you think about it -- the listener's
> brain needs to filter out the extraneous sounds and and compensate
> for the concert hall's reverberation in order to perceive the
> instruments and/or vocals clearly; and this processing is much more
> difficult if the information is further muddled up by poor reproduction.
>

I've heard demos of things like this on a Blue Sky setup in a good room. 
What I can't say is that I could generalize from that.

Blue Sky is expressly designed to minimize distortion. But they're not
exactly common - I can't afford them myself.

> In audio, the "quality doesn't matter because it's distorted somewhere
> else" argument just does not work as a general argument.


Mmmmmmmm.... it does when you have a human in the loop. It's all
peeling the onion from a perceptual standpoint. So very much of
our hearing is in processing.

It's almost a "causal hallucination". Hearing is extremely unreliable.
Don't get me wrong; it's miraculously good - but from day to day,
the same thing sounds different.

>  In specific
> cases it might be provable.  Maybe speaker transducers always cover
> up pre-ringing, but maybe not.
>

Agreed. All things considered, I think the lower-hanging fruit
is in speakers, myself. I'd be really excited to hear a set of speakers 
that could show pre-ringing clearly enough . Assume a pathological
transient that highlights it, of course.

But speakers are due to get better. The problem is that it's a
messed-up marketplace. If you could bild much better speakers, how
would you market it? Firms like Meyersound deal with people who are
much more knowlegeable for live SR systems, and there's a lot more
margin.

Whereas they can't even sell me a nice 2.1 Blue Sky setup. :(

> Steve
>

-- 
Les Cargill

Max wrote:
> On Wed, 2 Nov 2016 19:24:43 -0500, Les Cargill
> <lcargill99@comcast.com> wrote:
>
>> Steve Pope wrote:
>
>> Okay - for *end playback*, 16/44.1k is extremely entrenched and
>> perfectly adequate.
>
> For the non-demanding ear, 128 bps MP3 is 'perfectly adequate.'  It
> depends on whether you're sensitive to the odd-sounding artifacts.
>

I can't consider 128k MP3 adequate for anything. You can diff the
waveforms; the difference is a lot of signal. 320k even has *some*
difference. It's more subtle.

>> Those are big things to move to have any sort of  change.
>
> Only if your sole playback is CD format.
>

I suppose there could be linear, download-centric material, but
it's pretty hard on the content provider.

>> It also doesn't help that you get high-profile folks like Neil Young
>> spouting nonsense about Pono.
>
> That's your own judgement.  Not shared universally.
>

Young is legendarily bad at talking about this stuff. He simply
doesn't understand. PONO could be the greatest thing since sliced
bread, but he's not the guy to have as spokesman.

Young is a sharp guy but this is just outside his envelope. I'm a
pretty big fan.

>>> Are you saying the observations presented here about pre-ringing, pre-echo
>>> or delay spread are not technically informed?  (As opposed to, difficult
>>> to sell, a point to which I will cotton to.)
>>>
>>
>> No, I am saying that the perceptual/quality effects of them are
>> questionable at best and very likely moot.
>>
>> Speakers and rooms are much more of a problem.
>
> They are completely different problems, and arguably much more
> natural-sounding to the human ear.  Pre-ring definitely is not.
>
> It sounds like your mind is already set, but if not, look into
> research being done by Meridian, including new formats that are
> intended to be more ear-friendly.
>
>

I will - I didn't mean to sound closed-minded. I have read some of their 
stuff. It's a bit of a slog.

-- 
Les Cargill

Max wrote:
> On Thu, 27 Oct 2016 19:59:13 +0000 (UTC), spope33@speedymail.org
> (Steve Pope) wrote:
>
>> Tim Wescott  <tim@seemywebsite.really> wrote:
>>
>>> And second, because really steep filters ring.
>>
>> Bingo.
>>
>>> and my boys can hear sounds significantly above 20kHz, which is probably
>>> around where the ringing would happen if you're aiming to sample at 44kHz.
>>
>> This part I'm not following -- the filters presumably are linear
>> (so the filter itself adds no new frequency components) and
>> neither are they underdamped (existing frequency components
>> are not amplified by the filter).
>>
>> The ringing artifact shows up as delay spread, as I noted above.
>> Sounds components that were once separated along the time axis
>> now overlap.
>>
>> Steve
>
> I always thought that this should be intuitive.  You send a pulse into
> a steep filter (as in a CD player), and you get the impulse response.
> No one argues that.  So what happens with a sharp snare drum hit?
> I've heard some demos where this is very audible.
>

I have a cheap USB D/A-A/D setup, and what a single-sample impulse does 
is about what you'd expect.

> Of course high bandwidth doesn't eliminate that. But it shortens the
> time frame, makes it possible to use less aggressive filters, and
> lessens the degree of phase skew when minimal phase filters are used.
>

-- 
Les Cargill

> 
> I cannot say, however, what pre-ringing might sounds like. It'd take
> some serious consideration to design a test to find out.
> 
this is an interesting point

pre or post ringing at 22kHz would be difficult for many people to hear.

So as an experiment, scale everything (the filter cutoff freq) down by a factor of 5 or 10.

Ringing at 2 or 4 kHz should be easy to hear.

It would be an interesting experiment to see if you can hear it at 10 kHz even.

m


m

Most of the concern over pre-ringing occurs in the studio, where you may ha=
ve 5 or more dsp sound processing boxes connected in series using analog I/=
O.  This increases the pre-ringing period from about 0.5 ms for a single bo=
x up to 5ms for a cascade of 10 units.=20
Also some DAW software may optionally provide super-steep filters for sampl=
e-rate conversion, which may put you in the danger category. =20

Some a/d converters now use minimum-phase decimation filters instead of lin=
ear-phase filters, which reduces the pre-ringing at the expense of more pos=
t-ringing which tends to be temporally masked. A good idea IMHO; it's too b=
ad the industry adopted the term "linear-phase" because who doesn't want li=
near ?

Bob