moe midi timing testing

[Len Ovens]

Started this on irc, copied it here:

ailo, The only thing I am seeing with midi timing is a general spread of
notes that are played simultaneously. A lengthening of notes. I was using
drum track with noteon/off pretty much the same time. (zero tics as far as
qtractor could tell) all of the notes were at least .003 - .006 (3 to 6
tics) after going out the midi port and back. With rtc root.root it was
mostly .008 or more.
I think default tic is 1/96 quarter note.
Three beats/notes that started out on top of each other, ended up beside
each other... close to .020 difference with no rtc. about .010 with. This
is with clocks at stock 64.
I think the reason my demo sounded so bad is that it is quite busy. I
think you may find testing with 10 or 20 simultaneous notes will give a
clearer picture.
Single note stuff was very good.

Because my usb midi IF has male plugs, I have been using a through port
which should only add delay and not jitter, however, I found an old switch
unit I put together for changing routing. It is all hard wire, I will try
using that to see if there is any difference.

I am going to make up a qtractor test track with all 1/8th notes 10 notes
for each 1/8th note (note a pretty sound :P ) each 10 tics (.010) wide and
try again. It isn't worth trying tweaks without being able to reproduce
the badness I have heard with my ears time after time.

10 notes at a time may be too small, it is not hard to think 3 drum hits
(6events) and one keyboard part with two bass notes and a four note chord
(Am7 for example though I normally play a C over A bass) Which is 12
events all together. Add strings.... an arpeggio on top of that could
sound drunk.


-- 
Len Ovens
www.OvenWerks.net