Author |
Message |
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Wed Jan 31, 2007 3:06 pm Post subject:
|
|
|
polychoron wrote: | I'm measuring latency | What procedure do you use for that ? I'm a bit puzzled as to how one would be able to measure delay times of half the sample time.
And I was thinking that maybe there is some asynchroneous aspect involved here due to the way things are being measured.
The really odd thing here seems to be though that a device can be used to measure it's own behaviour, as if it's on a couch, talking to a therapist
Well not quite, but still. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
chresan
Joined: Jul 11, 2006 Posts: 150 Location: Karlsruhe, Germany
|
Posted: Wed Jan 31, 2007 3:39 pm Post subject:
|
|
|
Hi Rudi,
I assume much time is wasted by creating the patches and setting the module indices. Perhaps you are faster if you have a program to automatically generate the patches from templates. If it helps, I will write this program for you.
Christian _________________ ----------------------
http://nmedit.sf.net |
|
Back to top
|
|
|
polychoron
Joined: Dec 12, 2006 Posts: 16 Location: Eindhoven, the Netherlands
|
Posted: Thu Feb 01, 2007 2:35 am Post subject:
|
|
|
Quote: | The really odd thing here seems to be though that a device can be used to measure it's own behaviour, as if it's on a couch, talking to a therapist
|
.. or is it the couch that's talking
I also had doubts whether the NM could be used to check up itself, but there are some things you can do to make sure that what you measure makes sense, e.g. placing two copies of the modules you are testing in series (the total latency should then double) etc. I trusted the method I used completely until i saw the 4-1switch's inner latency.
Just after I posted yesterday I edited my post to include the module `couch' for the 4-1switch. So there you can see how NM's own delay module is used for generating delay to measure against latency. The delay module is perfectly capable of making delays that are not integer multiples of the sample duration --- it probably does interpolation of some kind.
Christian, thanks for the offer, I will think a little about a sensible module couch template.
---Rudi |
|
Back to top
|
|
|
chresan
Joined: Jul 11, 2006 Posts: 150 Location: Karlsruhe, Germany
|
Posted: Fri Feb 02, 2007 10:32 am Post subject:
|
|
|
polychoron wrote: | ...
Christian, thanks for the offer, I will think a little about a sensible module couch template.
---Rudi |
Hi Rudi,
I have hacked together a simple patch file generator app (I have also use for it). You can specify a patch configuration with modules and cables and additionally place holder variables for a class of modules (for example modules in category Audio).
The application will create a patch file for all combinations of the place holder's value. And you have full control over module indices.
Well these explanations are always a bit abstract, so here is how a template file could look like:
Code: |
modules:
dummy = category:Audio
out = 2Output
mixer = Mixer
osc = OscA
inv = InvLevShift
cables:
out:input:1 - out:input:0
mixer:output - out:input:0
osc:output - inv:input
osc:output - dummy:input
inv:output - mixer:input:0
dummy:output - mixer:input:1
|
Inside the section 'modules:' you specify the modules like:
variable '=' <module>
or
variable '=' 'category:' <category>
In the 'cables:' section you define the wiring (I'll explain that later if necessary).
In this example the 'dummy' variable acts as place holder and will be replaced by each module in the audio-category .
However, I hope this helps to get an idea how a template file will look like and if you have use for it. If you need some special features, you are welcome to ask for.
Sincerely,
Christian _________________ ----------------------
http://nmedit.sf.net |
|
Back to top
|
|
|
polychoron
Joined: Dec 12, 2006 Posts: 16 Location: Eindhoven, the Netherlands
|
Posted: Fri Feb 02, 2007 1:05 pm Post subject:
|
|
|
Hi Christian,
This will be very, very useful. I sort of worked by trusting that after deleting a module and a newly added module would fall in the `hole'. So I was paranoid all the time about the module order (fortunately I had Jan's tool to check it).
In a module couch template, I would like to make two `columns' of modules that are identical and that both contain the same module X that is being scrutinized. So it would be nice if placeholders can appear more than once and be substituted simultaneously with the same module (that is just too many words but you know what i mean). Perhaps this is already a feature of your generator.
It would also be nice if I could connect up all the inputs and all the outputs of module X in each column in a similar manner. So some kind of enumerator of audio in/outputs would be nice, one that survives references to nonexisting in/outputs (so that e.g. the AudioIn can be substituted although cables for its nonexistent inputs are specified in the template).
I think I solved the 4-1switch mystery, that is, i found new evidence that is consistent with it having a delay of half the sample duration. Such a delay can only be created by interpolation, say maybe plain averaging a sample with its predecessor. In any case, there would be a loss of information. And indeed there is information loss in the 4-1switch. If you feed a saw through 64 4-1switches in series, the saw is no longer a saw. I cannot hear it but it is clearly visible on the oscilloscope. It has lost amplitude, and the sharp corners are gone. Exactly what you would expect from averaging.
---Rudi |
|
Back to top
|
|
|
chresan
Joined: Jul 11, 2006 Posts: 150 Location: Karlsruhe, Germany
|
Posted: Fri Feb 02, 2007 1:45 pm Post subject:
|
|
|
Hi Rudi,
I will add the desired features. I think I can finish the tool somewhen this weekend. By the way, do you have any preferences about how the generated patch files should be named? For example should it be possible to name them after the tested module, or is an ordinary numbering scheme sufficient like in couch1.pch, couch2.pch, ... ? _________________ ----------------------
http://nmedit.sf.net |
|
Back to top
|
|
|
polychoron
Joined: Dec 12, 2006 Posts: 16 Location: Eindhoven, the Netherlands
|
Posted: Fri Feb 02, 2007 2:14 pm Post subject:
|
|
|
Hi Chistian,
naming after the module would be nice. There would be a large number of these patches sitting in a directory--- so I could use a reminder . Module numbers are also OK.
Thanks,
Rudi |
|
Back to top
|
|
|
chresan
Joined: Jul 11, 2006 Posts: 150 Location: Karlsruhe, Germany
|
|
Back to top
|
|
|
|