From: Glyn A. <gl...@ad...> - 2012-01-08 12:20:28
|
On 07/01/12 21:38, D. Michael McIntyre wrote: > On Saturday, January 07, 2012, Glyn Adgie wrote: > >> The timing tests I have done so far indicate that the delay between the >> first NoteOn and its related PatchChange is about 10 msec. >> >> Any thoughts on this? > Does the file have program change events in its data stream, or does this > happen with any random file that sets up programs using the top level > controls? > > Were you using the free-standing Fluidsynth, or the DSSI plugin version? The only PatchChange events are specified in the "Special Parameters" for each channel. I will test PatchChange events embedded in the score later. I used standalone Fluidsynth. Curiously, the problem did not seem to occur when I used Rosegarden to record a trumpet piece played with Fluidsynth. It looks like a thread synch problem in Rosegarden, which might not always be reproducable. I am guessing that the initial setup of each channel when playing starts is done in a different thread from the score events, and that the processing of score events is not waiting for the setup to finish. I will try putting a PatchChange at the very start of the score, using the Event Editor. That should fix it, as the patch change is part of the score, and must occur before the first NoteOn. Thanks for all your help. |