> Further to the above, RoseGarden can use the Hydrogen drum synthesizer so
> if you are looking for just randomization of rhythm that would be a way to
> go.  On physical pitched instruments real humans don't actually make
> "random" errors but instead tend to particular errors due to physical
> difficulty of executing that part of the performance.

I think there is a difference between systematic errors (e.g. a hard to 
play part, physical constraints of an intrument) or intentional 
deviations from what a sequencer reproduces when perfectly quantizing 
(e.g. rallentando, crescendo, sforzando...) and the 'natural' randomness 
in tempo and velocity deriving from a human playing.

Ideally the former should be intentially 'programmed' in the MIDI 
writing on a sequencer. The second can be addressed by adding some 
randomness in tempo (note onsets, duration) and velocity.

Hence the use of
> "Amateur" soundfonts when one wants to simulate a high school band or
> drunken performers.

I think that it's much easier to do some randombess by hand in the 
matrix editor in Rosegarden than editing a soundfont to get that effect. 
It wouldn't be that hard to implement a live 'randomizer' e.g. in Pure 
Data but then you'd have to playback the midi and re-record it in 
Rosegarden which would be a bit cumbersome.
 From a meta-programming/logical point of view randomization isn't that 
hard, once you establish the max randomisation (maybe a percentage of 
something) you just cycle through all notes and change e.g. note onset. 
I know.... easier said than done, but I'm sure some of the code which 
already does bulk operations on selected notes (e.g. velocity changes) 
could be reused? :)

Of course a humanizer/randomiser could be part of a wider 'groove 
quantize' feature for Rosegarden, but I imagine that would be rather 


Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
Rosegarden-user mailing list
Rosegarden-user@lists.sourceforge.net - use the link below to unsubscribe

Reply via email to