DevNotes:04.06.2007

From Resound-wiki

Jump to: navigation, search

The first live set up of a Resound-based system, Culture Lab (Newcastle University) 4th June 07.

1 Hardware
2 Software
3 Setup
4 Testing and Observations
5 Images

[edit]

Hardware

RM Expert 3010 PC
Focusrite Saffire Pro 26 IO Audio interface
8 x Genelec 8050A monitors
Mackie Onyx 1640 mixing desk
Smartfade control surface (lighting desk)

[edit]

Software

Resound Client and Server 0.0.29
jackd 0.103.0
qjackctl 0.2.22
ardour 2.0.2

[edit]

Setup

One stereo track in Ardour -> Ardour Master buss -> Resound inputs (x2)
Resound outputs -> audio interface analog outputs (x8)
Saffire Pro analog outputs (x8) -> mixing desk (for in-line attenuation) -> active monitors

[edit]

Testing and Observations

The system was tested by with 2 inputs (stereo track from Ardour) and 8 outputs.
Matrix parameters were assigned directly to on-screen physical controls in stereo pairs
The Mexican Wave effect was also tested
JACK settings: 48kHz sample rate (but the source file played back from Ardour was 44.1kHz -- assume this means that the FreeBoB driver was (unnecessarily) re-sampling on the fly?), 3 periods per buffer.
Various JACK buffer sizes (frames per period) were tested:
- With a buffer size of 1024 samples the system operated very smoothly, but delay times between interface movements and audible results were unacceptably high, especially with fast fader movements.
- With a buffer size of 32 samples, there was a lot of spiking and glitching.
- With a buffer size of 128 samples some glitching was still present. With rapid changes in parameters (e.g. with the Mexican Wave effect, especially at higher frequencies) there was much glitching and drop-outs in the automation of parameters.
- With a buffer size of 256 samples, some glitching was still present, and the response time is noticably sluggish.
- With a buffer size of 512 samples, there were occasional glitches (say once every 30 seconds or so, sometimes much less) but a very sluggish response time to changes in parameters.
JACK sample rate was then changed to 44.1kHz (matching that of the source file).
- At 512 samples occasional glitches were still present. Could this be to do with graphics updates, perhaps?
- Changes to the "Priority" setting in JACK were followed by strange cyclic buffer errors and artefacts in the audio stream. More thorough testing is required.
These tests have been carried out with ardour2 and a web browser running in the background... A more 'light-weight' media player (or external audio input to the sound card) would probably yield acceptable results with smaller buffer sizes.