Re: [Denemo-devel] A Good Idea

[Richard Shann]

On Thu, 2010-03-04 at 15:05 -0600, Jeremiah Benham wrote:
> 
> On Mar 4, 2010, at 6:46 AM, Richard Shann <address@hidden>  
> wrote:
> 
> >
> > J. would you like to look at importmidi.c with a view to separating  
> > out
> > the file open/new score/new staff stuff from the core function that
> > takes an smf_track and puts the smf_events into a given staff?
> 
> I am a bit confused here. You want seperated functions. The first  
> loads the midi into smf structure in memory.
given a filename, yes.


>  A second function that  
> creates a new score?
well we already have that - you are calling something, newscore() or
why.

> Another function that adds a new staff.
again we already have such a thing, you don't need it for your current
MIDI import because the newscore creates one staff for you.


>  Or are  
> you looking for me to import the midi appending it onto new staffs.  
> Then a seperated function that adds the imported midi as a new  
> movement. Then lastly one that will replace the current score.
No. Just breaking up your current file import so that it uses a function
that takes an smf_track_t and puts the smf_event_t's that are in that
track into the current staff starting at the current position.

(OK, I admit, I am slightly glossing over the details of information,
(meta data I think) that will come at the start of the track, giving
timesignature etc. You have the usual things that you can do for this -
crudest would be just a flag for if you are doing import from file or
just adding smf_event_ts that have been recorded off someone's MIDI
controller.)

On quantization I have some good (I hope) ideas, but for the moment just
fix a value. The nice thing is that we don't destroy the recording when
we try out a quantization value, and, as I say, I have some ideas that
might actually make import midi work as a useful thing to do (basically
the idea of rhythmic cliches and, ultimately, a neural net classifier,
but this maybe would be farmed out to a separate program).

Richard


> 
> Importmidi either needs a GUI to set quantization parameters.  
> Basically what would need to be set would be the minimal granule size.  
> This granual could be an sixteenth, 32snd, 64th etc. If a GUI is not  
> desirable perhaps we can do what rosegarden does and have a very small  
> initial granule setting and then have a scheme script that can do  
> further quantization.
> 
> Jeremiah
> 
> >
> > Richard
> >
> >
> >
> >
> > _______________________________________________
> > Denemo-devel mailing list
> > address@hidden
> > http://lists.gnu.org/mailman/listinfo/denemo-devel