I haven't had a look at the patches, but did you separate audio and gem in any way? there are different strategies on how to do that, explained here: http://web.uni-weimar.de/medien/wiki/Audiovideo
m.
Am 03.07.2010 um 11:43 schrieb Darren Kelly:
PD version: Pd-0.42.5-extended-20100501 Mac OS X: 10.5.8, 4GM, 2.5GHz Intel Core 2 Duo
Hi PD List,
I'm replying to my own thread, started a couple of years ago, concerning crackles and pops in VFOs driven by modulating accelerometer signals (such as from a Wiimote).
I've been creating some new 3D GEM visuals (and making videos from them), however the quality is compromised by pops, crackles, and glitches in the audio.
One can hear the sound problems (and see the nice videos) at:
Gallery: 3D Drancel RGB visuals and sound synthesis with a PureData/GEM Prototype of DranceWare http://www.webel.com.au/node/2592
Some videos that demonstrate the sound hiccups include:
Video: The Drancel RGB virtual triaxial accelerometer: colour=sound synthesis principle in 3D (with 3 VFOs and simulated sinusoidal signals) http://www.webel.com.au/node/2617
Video: Drancel RGB accelerometer "light horns" sound=colour synthesis principle (with 3 Variable Frequency Oscillators): driven by Wiimote http://www.webel.com.au/node/2616
Video: 1 Drancer (acceleration avatar) with 5 Drancel RGB virtual triaxial accelerometers: colour=sound synthesis principle in 3D (PureData/GEM) 3 VFOs per Drancel and simulated sinusoidal signals http://www.webel.com.au/node/2618
Video: 2 Drancers (acceleration avatars) each with 5 Drancels: in PureData/GEM Prototype of DranceWare: rotating camera: driven by Wiimote http://www.webel.com.au/node/2614
I can conclude so far:
It happens whether or not I show the 3D GEM visuals (I can remove the [gemhead] and it makes no difference to the sound).
It happens even for the simplest of my systems with only 3VFOs (and does not seem to get proportionally worse with number of elements).
It happens whether or not I use smooth simulated (sinuisoidal) or real accelerometer signals driving VFOs
(I can safely exclude the accelerometer sensor signal quality).
- It also happens even when I do not monitor signals in widgets
(I've come up with a nice little checkbox/spigot-based bypass system for number and slider widgets, because they can definitely eat a lot of CPU, and they can cause disturbances).
It happens even when I am not changing parameters in the graphical interface. It "just happens" sometimes.
It happens no matter what time-smoothing scale I use to smooth changes in the driving frequency of the VFOs.
I don't think it is a CPU problem, it happens even when the CPU usage is modest.
And, by comparison, it definitely does not happen on an older Java/JSyn version of my system, with very similar parameters and synthesis, so I don't think the algorithm is to blame.
I am clueless as to why I am still getting these pops and crackles in PureData+GEM on Mac OS X.
I would be most grateful for suggestions of any other tests I might perform to isolate the problem.
Q: Is there a standard patch I can use to test/benchmark performance of PureData on my maschine ?
Yours Gesturally Accelerated,
Dr Darren Kelly (Webel IT Australia)
Darren Kelly wrote:
Hi PD list,
I have uploaded an example of the problem as about 20 rather unmusical seconds of me waving 1 Wiimote as accelerometer driving frequencies of 3 [osc~], almost nothing else running, and no visuals, sliders, atoms:
http://www.webel.com.au/project/drancing/puredata/audio/drancing.VFO.3osc.cr...
It is panned quite left (when I use 2 Wiimotes I pan them leftish and rightish).
Will keep breaking it down further to isolate the problem.
The patches including visuals (now disconnected) can be viewed at:
http://www.webel.com.au/project/drancing/puredata/index.jsp?page=DranceWareP...
Darren
Collin Oldham wrote:
I think you are getting those crackles from the atoms (number boxes) that are trying to redraw every time they get a new number from the accelerometers. (which is probably pretty often!) That would explain why it only happens you are using the sensors, and why your very sensible solutions didn't work. Just a hunch! I'd try hiding it all in a sub patch, and if that works, you could use something like [speedlim] (from cyclone?) to limit the refresh rate on the atoms.
Collin
-- Webel IT, "The Elements of the Web" Specialists in Drupal CMS web engineering, UML, SysML, Java and XML, Darren Kelly, BSc, PhD phone: +61 (2) 9386 0090 mobile: 0405 029 008 post: PO Box 1816, Bondi Junction, NSW 1355, Australia
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list