digital faders distort sound

[Guest]

Archived from groups: rec.audio.pro (More info?)

"U-CDK_CHARLES\\Charles" <"Charles Krug"@cdksystems.com> wrote in message news:<Ay8ed.6$803.1@trndny04>...
> On 21 Oct 2004 11:00:06 -0700, Gary <midicad2001@yahoo.com> wrote:
> > You're not getting Doppler distortion caused by ramming your car
> > repeatedly into your stereo are you?
>
> Oh THAT brings back memories! The final exam for Engineering Physics II
> oh . um . .*COUGHS* years ago:
>
> "You are driving towards a perfectly reflective brick wall at 100m/s.
> The temperature is 35C and there is no wind.
> Your blow your horn, which sounds a pure tone at 200Hz.
>
> What pitches do you hear?
>
> Note all the "Only in physics class" contrivences.

200 Hz (direct from the horn) and 313.4 Hz reflected from the wall.

35C means that sound travels at 352.5 m/s. If a stationary listener
were to hear a stationary horn playing at 200 Hz, the listener would
hear 200Hz, regardless of temperature. However, the source (horn) is
moving toward the wall at 100 m/s, and the listener is approaching the
wall at 100 m/s, so the compounded rate is 200ms (listener moving
toward the source). 200 m/s is .567 the speed of the sound itself *at
this temperature* (horrible to think what would happen to the listener
if the car hit the wall at .283 the speed of sound!) so the resulting
wave is shifted up by 56.7% or 113.4 additional cycles per second,
resulting in the final 313.4Hz.

Karl Winkler
Lectrosonics, Inc.
http/www.lectrosonics.com

 

[Guest]

Archived from groups: rec.audio.pro (More info?)

On Fri, 22 Oct 2004 09:30:34 -0700, maxdm wrote:

> philicorda <philicordaNOOSPAM@azriel.tydrwg.com> wrote in message news:<pan.2004.10.23.02.55.55.42508@azriel.tydrwg.com>...
>> On Thu, 21 Oct 2004 05:39:39 -0700, maxdm wrote:
>
>> I tried this with whatever was in the multitrack at the time... vox,drums
>> and bass I think. Anyway, if it nulls, it nulls. The signal is identical.
>>
> Could you describe what you did? I didn't get the technical details.
> keep in mind that two signals that distort in the same way when summed
> out of phase will null anyway.

Ok. I last did this ages ago, without paying much attention, so I re-did
this test today in cubase SX, and found some interesting results.

When I say 'nothing' in the following text, I mean absolutely nothing,
digital zero, nada. When I say 'fader at zero', I mean I am not boosting
or reducing the signal at all.

This is the setup.
1 stereo track with a 16/44.1k song mixdown. Fader at zero.
1 stereo track with the same mixdown, but out of phase. Fader at zero.
Both tracks are routed to the master outs.

So, as you might imagine, nothing at all comes from the speakers when I
press play. (The in and out of phase signals totally null.)

Next, I added ten group channels, and connected the out of phase mixdown
track to the first group channel instead of the master outs. I connected
the output of that first group to the input of group two, and so on up to
group ten. Group ten's fader goes to the master outs. The original mixdown
is still going direct to the master outs, with it's fader at zero, and
remains so for all the rest of the tests.

I then pressed play, and, as expected, nothing came out.

Now it gets interesting.
I set the levels of the group channels with the faders as follows. (dbs)
+1
-1
-3
-4
+6
+1
-3
-3
+5
+1

And... SHOCK HORROR! It does not null! The faders are doing something they
are not telling me about!
Analysis of the 'nulled' mixdown shows...

Min sample -0.001, Max sample 0.001
Max RMS (Left) -64.56
MAX RMS (Right)-64.75
This is very bad. The mixdown file contained the original song,
just very, very quietly, though not distorted as far as I can tell.

So, suspicious of the CubaseSX faders, I redid the test using the
'tools-1' vst plugin on each of the groups (It's a simple stereo
fader+phase reverse button plugin). All the Cubase faders on the groups
were reset to zero.
The levels in the 'tools-1' plugin are as follows (dbs)....

-1.5
+1.5
-10.5
+10.5
-17.0
-7.0
+12.0
+12.0
+3.0
-3.0

And I get a total null on the nulled mixdown.
Min sample 0, Max sample 0
Min/max RMS -infinity.
Putting the signal though all those faders had not made any change to it
whatsoever.

So, to conclude...
Some digital faders can attenuate or boost a signal, without adding
distortion, assuming the mixer and faders have a high enough bit depth to
not lose any information.

The CubaseSX faders ain't doing what they say. They are not distorting the
signal, but they are not boosting or attenuating by the amount you set
them to either. I don't know how I managed to get them to null the first
time I tried, but I suspect that a bug has crept in as the program has
been updated.

> Sine waves in digital come out fairly well. up until near sampling
> frequency.
> the more complex the wave the more difficult it is to reproduce.

Digital faders operate on a single sample at a time, so they don't care
what the waveform is. It's a multiply or divide on a single number,
nothing more complex.

> THD is not the kind of distortion I am talking about, I am talking about
> loss of imaging, depth and transient distortion. Someone posted that a
> flute is like a sine wave, but the flute only resembles a sine wave, I
> think most musicians and listeners will agree that there is a lot more
> to the flute sound than a sine wave!

You have to find a way to measure this distortion you are hearing.

>
> the ear is very sensitive to transients and phase/time distortion.

I very much doubt a digital fader would do either of these, but I welcome
further investigation.

 

[Tony]

Archived from groups: rec.audio.pro (More info?)

On Fri, 22 Oct 2004 19:18:27 GMT, philicorda
<philicordaNOOSPAM@azriel.tydrwg.com> wrote:

>On Fri, 22 Oct 2004 09:30:34 -0700, maxdm wrote:
>
>> philicorda <philicordaNOOSPAM@azriel.tydrwg.com> wrote in message news:<pan.2004.10.23.02.55.55.42508@azriel.tydrwg.com>...
>>> On Thu, 21 Oct 2004 05:39:39 -0700, maxdm wrote:
>>
>>> I tried this with whatever was in the multitrack at the time... vox,drums
>>> and bass I think. Anyway, if it nulls, it nulls. The signal is identical.
>>>
>> Could you describe what you did? I didn't get the technical details.
>> keep in mind that two signals that distort in the same way when summed
>> out of phase will null anyway.
>
>Ok. I last did this ages ago, without paying much attention, so I re-did
>this test today in cubase SX, and found some interesting results.
>
>When I say 'nothing' in the following text, I mean absolutely nothing,
>digital zero, nada. When I say 'fader at zero', I mean I am not boosting
>or reducing the signal at all.
>
>This is the setup.
>1 stereo track with a 16/44.1k song mixdown. Fader at zero.
>1 stereo track with the same mixdown, but out of phase. Fader at zero.
>Both tracks are routed to the master outs.
>
>So, as you might imagine, nothing at all comes from the speakers when I
>press play. (The in and out of phase signals totally null.)
>
>Next, I added ten group channels, and connected the out of phase mixdown
>track to the first group channel instead of the master outs. I connected
>the output of that first group to the input of group two, and so on up to
>group ten. Group ten's fader goes to the master outs. The original mixdown
>is still going direct to the master outs, with it's fader at zero, and
>remains so for all the rest of the tests.
>
>I then pressed play, and, as expected, nothing came out.
>
>Now it gets interesting.
>I set the levels of the group channels with the faders as follows. (dbs)
>+1
>-1
>-3
>-4
>+6
>+1
>-3
>-3
>+5
>+1
>
>And... SHOCK HORROR! It does not null! The faders are doing something they
>are not telling me about!

I may be misdirected here, but are you assuming that because all the
"dBs" sum to 0 that the mix should be equivalent to summing 1 channels
all at 0dB (which of course is not the case, since dB is a log
function)?

Tony (remove the "_" to reply by email)

 

[Tony]

Archived from groups: rec.audio.pro (More info?)

On Sat, 23 Oct 2004 09:05:58 +1000, Tony <tony_roe@tpg.com.au> wrote:

>On Fri, 22 Oct 2004 19:18:27 GMT, philicorda
><philicordaNOOSPAM@azriel.tydrwg.com> wrote:
>
>>On Fri, 22 Oct 2004 09:30:34 -0700, maxdm wrote:
>>
>>> philicorda <philicordaNOOSPAM@azriel.tydrwg.com> wrote in message news:<pan.2004.10.23.02.55.55.42508@azriel.tydrwg.com>...
>>>> On Thu, 21 Oct 2004 05:39:39 -0700, maxdm wrote:
>>>
>>>> I tried this with whatever was in the multitrack at the time... vox,drums
>>>> and bass I think. Anyway, if it nulls, it nulls. The signal is identical.
>>>>
>>> Could you describe what you did? I didn't get the technical details.
>>> keep in mind that two signals that distort in the same way when summed
>>> out of phase will null anyway.
>>
>>Ok. I last did this ages ago, without paying much attention, so I re-did
>>this test today in cubase SX, and found some interesting results.
>>
>>When I say 'nothing' in the following text, I mean absolutely nothing,
>>digital zero, nada. When I say 'fader at zero', I mean I am not boosting
>>or reducing the signal at all.
>>
>>This is the setup.
>>1 stereo track with a 16/44.1k song mixdown. Fader at zero.
>>1 stereo track with the same mixdown, but out of phase. Fader at zero.
>>Both tracks are routed to the master outs.
>>
>>So, as you might imagine, nothing at all comes from the speakers when I
>>press play. (The in and out of phase signals totally null.)
>>
>>Next, I added ten group channels, and connected the out of phase mixdown
>>track to the first group channel instead of the master outs. I connected
>>the output of that first group to the input of group two, and so on up to
>>group ten. Group ten's fader goes to the master outs. The original mixdown
>>is still going direct to the master outs, with it's fader at zero, and
>>remains so for all the rest of the tests.
>>
>>I then pressed play, and, as expected, nothing came out.
>>
>>Now it gets interesting.
>>I set the levels of the group channels with the faders as follows. (dbs)
>>+1
>>-1
>>-3
>>-4
>>+6
>>+1
>>-3
>>-3
>>+5
>>+1
>>
>>And... SHOCK HORROR! It does not null! The faders are doing something they
>>are not telling me about!
>
>I may be misdirected here, but are you assuming that because all the
>"dBs" sum to 0 that the mix should be equivalent to summing 1 channels
>all at 0dB (which of course is not the case, since dB is a log
>function)?

That should have been "10" channels. And PS, everything has a
tolerance, even the "dB" setting of a fader, but that's not distortion
in my world.

Tony (remove the "_" to reply by email)

 

[Guest]

Archived from groups: rec.audio.pro (More info?)

"philicorda" <philicordaNOOSPAM@azriel.tydrwg.com> wrote in message

> And... SHOCK HORROR! It does not null! The faders are doing something they
> are not telling me about!
> Analysis of the 'nulled' mixdown shows...

So if your methodolgy is correct, you are actually suggesting that Cubase SX
has a bug, rather than 'digital faders' being inherently flawed ?

geoff

 

[Guest]

Archived from groups: rec.audio.pro (More info?)

On Tue, 26 Oct 2004 17:15:12 +1300, Geoff Wood wrote:

>
> "philicorda" <philicordaNOOSPAM@azriel.tydrwg.com> wrote in message
>
>> And... SHOCK HORROR! It does not null! The faders are doing something they
>> are not telling me about!
>> Analysis of the 'nulled' mixdown shows...
>
> So if your methodolgy is correct, you are actually suggesting that Cubase SX
> has a bug, rather than 'digital faders' being inherently flawed ?

Yes, and I think it's an interface bug rather than one in the audio engine..
I bet if I entered the fader levels by hand in the number boxes without
ever touching the faders, it would null. Just a hunch, I reackon there is
a tiny rounding error cropping up, and have a vague memory of someone else
finding the same bug...

I haven't a cubase SX box here right now, so it may have to wait a while
before I can test that theory.

>
> geoff

 

[Guest]

Archived from groups: rec.audio.pro (More info?)

On Sat, 23 Oct 2004 09:13:15 +1000, Tony wrote:


>>>And... SHOCK HORROR! It does not null! The faders are doing something they
>>>are not telling me about!
>>
>>I may be misdirected here, but are you assuming that because all the
>>"dBs" sum to 0 that the mix should be equivalent to summing 1 channels
>>all at 0dB (which of course is not the case, since dB is a log
>>function)?

Ohh. That does seem blindingly obvious now you have pointed it out, but
why did the 'tools-1' faders null? (And I don't think it was blind luck as
I changed the settings a few times and it still nulled.).

>
> That should have been "10" channels. And PS, everything has a tolerance,
> even the "dB" setting of a fader, but that's not distortion in my world.

Surely, with a digital fader, there is no problem with tolerance? At least
as far as being 'reversible'. Or is it because it's a log function, it has
to be an approximation (finding a point on a curve)?

>
> Tony (remove the "_" to reply by email)