There might be an issue because of phases for JACK2

[Ralf Mardorf]

Hi :)

it seems to be, that JACK2 will shift the phases mini, mini, minimal if
I connect a mono signal by system capture_1 to the left and right input
of a recording application. I didn't made any measurements, but I heard
it when I rooted the stereo signal by my analogue mixing console to
mono, while there was no widen effect when listening to it on stereo.

Any other connections seems to be fine.

Maybe someone else can verify or falsify this behaviour.

spinymouse-sudo@64studio:~$ jackd --help
jackdmp 1.9.2

Cheers,
Ralf
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Paul Davis]

On Sat, Aug 29, 2009 at 10:36 AM, Ralf
Mardorf<ralf.mardorf@...> wrote:
> Hi :)
>
> it seems to be, that JACK2 will shift the phases mini, mini, minimal if I
> connect a mono signal by system capture_1 to the left and right input of a
> recording application. I didn't made any measurements, but I heard it when I
> rooted the stereo signal by my analogue mixing console to mono, while there
> was no widen effect when listening to it on stereo.

i can think of no way for this happen. did you try it with jack1 as
well? what recording application?
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

Hi Paul :)

I was using Rui's Qtractor and I asked Rui + the 64 Studio list, before
I wrote to the JACK dev list, see below.

JACK1 on my hardware disconnect clients "without any reason", I'm only
able to make music because there is JACK2 :).

I'm very lazy (at the moment, not only) today. Within the next 2 weeks
I'll make a check up using a phase correlator. Any hints are welcome.
Rui suggested to use JAAA.

I wonder myself, that I don't notice anything bad for the quality of the
sound, when I connect stereo to stereo and listen to it on stereo and
mono, but that there is a loss of sound quality, when I connect mono to
stereo and listen to this stereo signal by mono.

Can I assume that this won't happen if you do run JACK2 on your
equipment and the distro you prefer?!

:S

Cheers,
Ralf

Paul Davis wrote:

> On Sat, Aug 29, 2009 at 10:36 AM, Ralf
> Mardorf<ralf.mardorf@...> wrote:
>  
>> Hi :)
>>
>> it seems to be, that JACK2 will shift the phases mini, mini, minimal if I
>> connect a mono signal by system capture_1 to the left and right input of a
>> recording application. I didn't made any measurements, but I heard it when I
>> rooted the stereo signal by my analogue mixing console to mono, while there
>> was no widen effect when listening to it on stereo.
>>    
>
> i can think of no way for this happen. did you try it with jack1 as
> well? what recording application?

The essential between Rui and me:

    [snip]

    may i ask
    you consider the help of jaaa and ultimately ask how it can help you
    figure whether there's something fishy (or "phasy", my own neologism)
    either in jack or in qtractor.



    > Btw. I'm running JACK2. I have to add that I sometimes run JACK with
    > options that aren't fine with the hardware, but as long as there are
    > no error messages and warnings, e.g. xruns, this shouldn't cause any
    > trouble.
    >

    damn, it ocurred to me that the parallelizing feature of jack2 (formerly
    jackdmp) might be doing it on different period cycles with at least one
    buffer period delay between each channel/port. indeed, i'll suggest you
    have jaaa report some evidence and/or reinstall jack1 and check if
    things sound any different.

    [snip]

    >
    > Okay, this mono issue seems to belong to JACK, but not to Qtractor.
    >

    i'll keep my fingers crossed :)

    [snip]
     

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Fons Adriaensen-2]

On Sat, Aug 29, 2009 at 04:36:21PM +0200, Ralf Mardorf wrote:

> it seems to be, that JACK2 will shift the phases mini, mini, minimal if I
> connect a mono signal by system capture_1 to the left and right input of a
> recording application. I didn't made any measurements, but I heard it when
> I rooted the stereo signal by my analogue mixing console to mono, while
> there was no widen effect when listening to it on stereo.

And how do you hear a 'mini mini minmal phase shift' ?

Any such effects are more likely to indicate a problem
with your sound card.

Anyway it's easy to measure this. Make a stereo recording
of a sineway (1kHz, -10dB), play back the recording to
jkmeter started with the -C option. The correlation meter
will show any phase difference. Repeat at 3,6, 10 and 15 kHz.

Do the same with the signal routed through your sound
card.

Ciao,

--
FA

Io lo dico sempre: l'Italia è troppo stretta e lunga.

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

Fons Adriaensen wrote:

> On Sat, Aug 29, 2009 at 04:36:21PM +0200, Ralf Mardorf wrote:
>
>  
>> it seems to be, that JACK2 will shift the phases mini, mini, minimal if I
>> connect a mono signal by system capture_1 to the left and right input of a
>> recording application. I didn't made any measurements, but I heard it when
>> I rooted the stereo signal by my analogue mixing console to mono, while
>> there was no widen effect when listening to it on stereo.
>>    
>
> And how do you hear a 'mini mini minmal phase shift' ?
>  

It was a mono signal that was connected by JACK2 as a pseudo-stereo
signal. I routed this pseudo-stereo signal to mono by my analogue mixing
console and the signal sounded as two equal signals, but not only that
it was louder, but also that there were some loss in quality.

> Any such effects are more likely to indicate a problem
> with your sound card.
>  

But than this effect should be hearable for all connections I do by
JACK2, but any other connections seems to be fine.

> Anyway it's easy to measure this. Make a stereo recording
> of a sineway (1kHz, -10dB), play back the recording to
> jkmeter started with the -C option. The correlation meter
> will show any phase difference. Repeat at 3,6, 10 and 15 kHz.
>
> Do the same with the signal routed through your sound
> card.
>  

I've got a DX7, but it's uncomfortable to use it as a sine wave
generator, my mixing console only provides 1KHz. Is there a sine wave
generator for Linux.

Is this one fine: http://www.lns.com/papers/tonegen/ ?

Cheers,
Ralf
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

Give me one or two days. Jackmeter and Jaaa etc. aren't installed.
Before I'll do anything else, I'll check my equipment and JACK2 ;).
Today I won't listen to any audio ... I guess because you are engineers
too, you understand that after doing some audio work, it won't be smart
to make tests, before having a rest ;). I'm overwrought, not because of
JACK, just because of listening.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[John Rigg-16]

On Sat, Aug 29, 2009 at 05:48:00PM +0200, Ralf Mardorf wrote:
> Give me one or two days. Jackmeter and Jaaa etc. aren't installed.  
> Before I'll do anything else, I'll check my equipment and JACK2 ;).  
> Today I won't listen to any audio ... I guess because you are engineers  
> too, you understand that after doing some audio work, it won't be smart  
> to make tests, before having a rest ;). I'm overwrought, not because of  
> JACK, just because of listening.

Try a null test with a music signal. Set both channels _exactly_ the same,
invert one channel, and sum them. They will cancel if there is no phase
difference.

If you're doing this in the analogue domain, switch off all EQ and any
HP or LP filters, as there will be phase shift differences in these due
to variation in component values.

John
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Jörn Nettingsmeier-5]

Ralf Mardorf wrote:

> Fons Adriaensen wrote:
>> On Sat, Aug 29, 2009 at 04:36:21PM +0200, Ralf Mardorf wrote:
>>
>>  
>>> it seems to be, that JACK2 will shift the phases mini, mini, minimal
>>> if I connect a mono signal by system capture_1 to the left and right
>>> input of a recording application. I didn't made any measurements, but
>>> I heard it when I rooted the stereo signal by my analogue mixing
>>> console to mono, while there was no widen effect when listening to it
>>> on stereo.
>>>    
>>
>> And how do you hear a 'mini mini minmal phase shift' ?
>>  
>
> It was a mono signal that was connected by JACK2 as a pseudo-stereo
> signal. I routed this pseudo-stereo signal to mono by my analogue mixing
> console and the signal sounded as two equal signals, but not only that
> it was louder, but also that there were some loss in quality.

can i ask what mixing console you were using? and does it have an eq
bypass? or was the eq engaged during your test? if the latter, and the
console was anything short of a neve or other gazillion-dollar gadget,
problem solved...
you could measure the phase response of both channels to make sure.


--
Jörn Nettingsmeier

Meister für Veranstaltungstechnik

Audio and event engineer
Ambisonic surround recordings

http://stackingdwarves.net
[phone]

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Fons Adriaensen-2]

On Sat, Aug 29, 2009 at 11:26:24PM +0200, Jörn Nettingsmeier wrote:

> can i ask what mixing console you were using? and does it have an eq
> bypass? or was the eq engaged during your test? if the latter, and the
> console was anything short of a neve or other gazillion-dollar gadget,
> problem solved...
> you could measure the phase response of both channels to make sure.

Mmm, an EQ set to 'flat' visually and not bypassed could have
a few dB of frequency-dependent gain. But since almost all analog
EQ circuits are minimal phase that will not produce any phase
shifts that would really affect mixing two supposedly identical
channels. The EQ errors on each channel would be much more
obvious than any cancellation effects on the mono mix.

For example, 3dB shelf will have a maximum phase shift
(at the 1.5dB point) of 10 degrees. But mixing two signals
with 10 degrees phase shift between them only produces of
loss of less than 0.1dB.

Ciao,

--
FA

Io lo dico sempre: l'Italia è troppo stretta e lunga.

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

John Rigg wrote:

> On Sat, Aug 29, 2009 at 05:48:00PM +0200, Ralf Mardorf wrote:
>  
>> Give me one or two days. Jackmeter and Jaaa etc. aren't installed.  
>> Before I'll do anything else, I'll check my equipment and JACK2 ;).  
>> Today I won't listen to any audio ... I guess because you are engineers  
>> too, you understand that after doing some audio work, it won't be smart  
>> to make tests, before having a rest ;). I'm overwrought, not because of  
>> JACK, just because of listening.
>>    
>
> Try a null test with a music signal. Set both channels _exactly_ the same,
> invert one channel, and sum them. They will cancel if there is no phase
> difference.
>
> If you're doing this in the analogue domain, switch off all EQ and any
> HP or LP filters, as there will be phase shift differences in these due
> to variation in component values.
>
> John

Thank you John :)

this is really a good idea. I don't need to install anything and to
learn how it works. I should have think about it myself. Anyway, I'm
(nearly) sure there will be phases. I'll do it Sunday or Monday in the
evening.

Cheers,
Ralf
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

R Parker wrote:

>> It was a mono signal that was connected by JACK2 as a
>> pseudo-stereo signal.
>>    
>
> What you call pseudo-stereo is mono split to both Left and Right. I doubt if summing the mono Left and Right back together is a valid test of anything. In real world applications, I can't think of any reason to do that.
>
> Summing stereo to mono via patch cables is problematic but happens because we run out of inputs on consols.
>
> I suspect you built a route that shouldn't be used. If correct, the good news is you have discovered something that should be avoided.
>
> I'm only speaking as an audio engineer.
>
> rtp
>  

It's a colourless story why I did it. You're right, normally I don't do
it. But I did it and there seems to be this issue. I'll do the test John
recommended.

Ralf

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

Jörn Nettingsmeier wrote:

> Ralf Mardorf wrote:
>  
>> Fons Adriaensen wrote:
>>    
>>> On Sat, Aug 29, 2009 at 04:36:21PM +0200, Ralf Mardorf wrote:
>>>
>>>  
>>>      
>>>> it seems to be, that JACK2 will shift the phases mini, mini, minimal
>>>> if I connect a mono signal by system capture_1 to the left and right
>>>> input of a recording application. I didn't made any measurements, but
>>>> I heard it when I rooted the stereo signal by my analogue mixing
>>>> console to mono, while there was no widen effect when listening to it
>>>> on stereo.
>>>>    
>>>>        
>>> And how do you hear a 'mini mini minmal phase shift' ?
>>>  
>>>      
>> It was a mono signal that was connected by JACK2 as a pseudo-stereo
>> signal. I routed this pseudo-stereo signal to mono by my analogue mixing
>> console and the signal sounded as two equal signals, but not only that
>> it was louder, but also that there were some loss in quality.
>>    
>
> can i ask what mixing console you were using?

Better not :D. It's a Behringer UB2442FX-Pro ;), but even if it's a
Behringer it's okay. Note that I'm an audio engineer and that I don't
have this problem if I send a mono signal by stereo to JACK. It only
happens when I spilt the signal by JACK.

> and does it have an eq
> bypass? or was the eq engaged during your test? if the latter, and the
> console was anything short of a neve or other gazillion-dollar gadget,
> problem solved...
> you could measure the phase response of both channels to make sure.
>  

What ever the reason will be, JACK or not JACK, it's not the mixer.

Ralf
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

Fons Adriaensen wrote:

> On Sat, Aug 29, 2009 at 11:26:24PM +0200, Jörn Nettingsmeier wrote:
>
>  
>> can i ask what mixing console you were using? and does it have an eq
>> bypass? or was the eq engaged during your test? if the latter, and the
>> console was anything short of a neve or other gazillion-dollar gadget,
>> problem solved...
>> you could measure the phase response of both channels to make sure.
>>    
>
> Mmm, an EQ set to 'flat' visually and not bypassed could have
> a few dB of frequency-dependent gain. But since almost all analog
> EQ circuits are minimal phase that will not produce any phase
> shifts that would really affect mixing two supposedly identical
> channels. The EQ errors on each channel would be much more
> obvious than any cancellation effects on the mono mix.
>
> For example, 3dB shelf will have a maximum phase shift
> (at the 1.5dB point) of 10 degrees. But mixing two signals
> with 10 degrees phase shift between them only produces of
> loss of less than 0.1dB.
>
> Ciao,
>  

Full ACK :). It's a very bad mixing console I'm using for home
recording, I don't think inside will be any discrete circuit ;), anyway,
I'm using JACK2 since month and never heard any thing when I check the
stereo signals in mono, but some day ago I did this mono to stereo
connection by JACK and checked the signal by listening to the sum in
mono and noticed that there is something wrong.

I'll do John's test and additional run jackmeter -c.

It can't harm if anybody else running JACK2 will do the same ;).

Ralf
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

Fons Adriaensen wrote:
> jkmeter started with the -C option.

1.

Qtractor > imported a sine at 440Hz mono

Qtractor Master/out_1 connected to jkmeter in-1 and in-2 > phases are
fine, at +1
Qtractor Master/out_1 connected to jkmeter in-1 and Qtractor
Master/out_2 connected to jkmeter in-2 > phases are fine, at +1

2.

Audacity > imported a sine at 440Hz mono

Audacity out connected with Qtractor in (not visible by any JACK
connection GUI) and Qtractor Master/out_1 connected to jkmeter in-1 and
Qtractor Master/out_2 connected to jkmeter in-2 > phases are fine, at +1

3.

Anyway, an Oberheim Matrix-1000 connected to the sound card channel 1
and than connected by JACK system capture_1 to Qtractor Master/in 1 and
Master/in_2, while Qtractor Master/out_ was connected to system
playback_1 and Qtractor Master/out_2 was connected to system playback_2
the sound WAS bad, when I rooted the subgroop by my mixing console to
mono AND it was not the mixing console, for other sources doing this the
sound was fine.

4.

Oberheim Matrix-1000 connected to the sound card channel 1, system
capture_1 connected to Qtractor Master/in 1 and Master/in_2, Qtractor
Master/out_1 connected to jkmeter in-1 and Qtractor Master/out_2
connected to jkmeter in-2 > phases are fine, at +1 but without playing
the Oberheim phases aren't at 0 or +1, but a little bit beneath 0 at +.

Checking it by headphones on mono the sound is fine :).

This is very strange! It wasn't fine some days ago, but it seems to be
not JACK, but some kind of voodoo :D.

Now everything is fine :).

Cheers,
Ralf
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Ralf Mardorf]

PS:

The day when I had this trouble I used extreme "Frames/Period" and
"Periodes/Buffer" settings, sometimes there were asyncs, xruns (without
hearable glitches), but not while listening and getting this bad sound,
there were no messages by JACK. Today I used normal settings.

I repeated it, running jack with this extreme critical settings,
Oberheim --> system capture_1 --split--> jkmeter left and right, phases
are fine and without playing the Oberheim a little bit less between 0 at
+, than when using Qtractor.

Even this bad settings don't cause phases for jkmeter BUT when I
listened to it by headphones I can hear that the sound is bad on mono.
So the voodoo isn't voodoo, but wrong settings for JACK, even if there
are no messages by JACK and even if jkmeter didn't notice it, it's
hearable. Anyway, this dosen't happen when using good settings for JACK :).

> Fons Adriaensen wrote:
>> jkmeter started with the -C option.
>
> 1.
>
> Qtractor > imported a sine at 440Hz mono
>
> Qtractor Master/out_1 connected to jkmeter in-1 and in-2 > phases are
> fine, at +1
> Qtractor Master/out_1 connected to jkmeter in-1 and Qtractor
> Master/out_2 connected to jkmeter in-2 > phases are fine, at +1
>
> 2.
>
> Audacity > imported a sine at 440Hz mono
>
> Audacity out connected with Qtractor in (not visible by any JACK
> connection GUI) and Qtractor Master/out_1 connected to jkmeter in-1
> and Qtractor Master/out_2 connected to jkmeter in-2 > phases are fine,
> at +1
>
> 3.
>
> Anyway, an Oberheim Matrix-1000 connected to the sound card channel 1
> and than connected by JACK system capture_1 to Qtractor Master/in 1
> and Master/in_2, while Qtractor Master/out_ was connected to system
> playback_1 and Qtractor Master/out_2 was connected to system
> playback_2 the sound WAS bad, when I rooted the subgroop by my mixing
> console to mono AND it was not the mixing console, for other sources
> doing this the sound was fine.
>
> 4.
>
> Oberheim Matrix-1000 connected to the sound card channel 1, system
> capture_1 connected to Qtractor Master/in 1 and Master/in_2, Qtractor
> Master/out_1 connected to jkmeter in-1 and Qtractor Master/out_2
> connected to jkmeter in-2 > phases are fine, at +1 but without playing
> the Oberheim phases aren't at 0 or +1, but a little bit beneath 0 at +.
>
> Checking it by headphones on mono the sound is fine :).
>
> This is very strange! It wasn't fine some days ago, but it seems to be
> not JACK, but some kind of voodoo :D.
>
> Now everything is fine :).
>
> Cheers,
> Ralf
> _______________________________________________
> Jack-Devel mailing list
> Jack-Devel@...
> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
>

--
Secret of Tux: http://images.wallaceandgromit.com/user_uploads/forum_thumbnails/5/75/355.jpg
"Gromit bit me" says HMV dog: http://img.dailymail.co.uk/i/pix/2007/03_03/GomitHMVPA_468x319.jpg

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org