Re: Real-time category

[Jeff Larson <Jeffrey.Larson@Sun.COM>]

bruce tovsky wrote:
> i do believe that hardware devices exist that can spit out midi clock
> derived from a real-time tempo input such as a drum track or live 
>drummer.

Yes, but that is a different issue, we were discussing just the feasibility
of tapping a loop of the correct size.

Once the loop is created, if you were using such a tempo extraction device
then the problem is one of external synchronization, how well does the 
looper
running as a slave adapt to changes in the MIDI clock rate.  I don't think 
an
RTOS provides any particular advantage here.  Again we're dealing with MIDI
clocks of unknown origin so everyone will do smoothing and lag a bit before
recognizing a change in the distance between clocks is actually a tempo
change and not just jitter.

The bigger problem for any audio looper is what you do in response to the
rate change.  You can't just change the rate at which you play the 
"sequence"
like a drum machine, you either have to make periodic adjustments by 
jumping
the playback position backward or forward, or you have to use a time
stretching algorithm to make the loop fit the new tempo.  To my knowledge
the only hardware looper that attempts time stretching is the Repeater
and it doesn't always handle it well.  It is a very hard problem, and an
RTOS doesn't help you solve it.

Periodic jumps are easier, but it only works if the tempo is drifting
by a BPM or so.  If the drummer dropped from say 120 to 87 BPM, any
looper that does not do time stretch will either start ignoring the clock
or start retriggering at seemingly random locations.  An RTOS doesn't
do anything for you here either.

Jeff