Re: [Qutecsound-users] A few issues

[Andres C. <man...@gm...>]

HI Matthew,


On Fri, Jan 30, 2009 at 1:53 PM, Matthew Walworth
<mat...@gm...> wrote:
> Have been playing with qutecsound for a couple of days now. Thank you for a
> wonderful application. It is rapidly becoming my environment of choice. I
> have run across a few issues:
>
> 1. MIDI latency seems quite high. I am trying to play from an external
> keyboard and am getting about 0.5 seconds of delay on each keypress. I've
> tried all the different thread/api options, but they don't seem to make a
> difference. I get little or no latency running the same csd from the command
> line.  Also interesting is that if I use a widget to output information
> about the midi message, the widget seems to update almost instantly on key
> press/release as does the information written to the console.
>

Latency is due to the hardware buffer size. In my limited testing, I
found portaudio to be able to handle low latencies. Try setting the -b
buffer to 128 and the -B buffer to 256. This should give you
acceptable latency.
What you mention about widgets might be an optical illusion, or might
be due to Csound sending the values to the software bus on every
control period, and since ksmps is much lower than the hardware
latency, the values are displayed faster.

> 2. The option to "run in terminal" doesn't seem to do anything. I've checked
> the path to Terminal.app and it is correct. Nothing seems to happen when I
> try it including that no messages are written to the console.
>

That's odd... its working here on 10.4. Are you completely sure the
path is: /Applications/Utilities/Terminal.app?
(case sensitive)

> 3. The option to graph the csd using graphviz has a problem writing its
> output file. I get an error dialog that says, "Cannot create temp dot/png
> file." I'm guessing it is a permissions problem?
>

I also noticed this. The workaround for now is to put the QuteCsound
application in another directory, e.g. the desktop, the documents
directory, etc.


> 4. This is certainly not a big deal, but rendering is cut short by a
> fraction of a second. I noticed this with a simple instrument that had a
> de-clicking envelope applied when it still clicked on the cutoff. The
> fraction of a second seems to vary a little bit around 0.25. It can be
> worked around by adding a time to the "e" statement, but that also must have
> an extra amount added to it to keep the render from being cut short. You can
> see the discrepancy if you check the "Elapsed time at end of performance" in
> the console.
>

Interesting. I'll check.

> My environment is a Mac mini running 10.5.6, 2Ghz Intel Core 2 Duo, 2GB ram,
> various MIDI interfaces including an M-Audio KeyRig 25 usb keyboard and an
> Edirol UM-1.
>
> (funny aside: when spell checking this message, the spellchecker offered
> "godsend" as a substitution for "qutecsound")
>

 =)


Cheers,
Andrés