[linux-audio-user] recording keyboard works (e.g. rosegarden)
j4strngs at bitless.net
Wed Dec 22 18:35:56 EST 2004
On Wednesday 22 December 2004 04:14 pm, pirrone wrote:
> Hans Fugal wrote:
> >I'm interested in rendering some organ works electronically. The first
> >step of course is to get the music in the computer, e.g. in MIDI
> >The piece I'm working on you can see at
> >I can play this piece well enough. I don't have a MIDI pedalboard but
> >the pedal line is easy enough to add later. But as you can see the
> >music is in multiple voices, and I would like to instrument each voice
> >differently. I have tried recording them one at a time, in an ensemble
> >style, but even when I follow a metronome it ends up sounding like
> >trash. I have decided that either I need to work on my ensemble
> >playing or I am going about it the wrong way. Maybe I should just
Do you practice with a metronome? If not, that would be my first guess
as to what you're seeing when you sequence.
You _could_ turn the click off, crank up the ticks per quarter to the max,
jack the tempo up and just record the guide track rubato (fancy talk for
"close enough for rock&roll" (I'm figuring you know that)). Not good in terms
of a pretty display, using the file to generate notation or a tidy SMF, but
it's approximating recording the performance on tape as far as the sequencers
temporality limitations goes. Again, I'm not sure of the exact reason you're
doing this. These things may be important, in which case you have to take a
Step recording is another possibility, but it's tedious. You didn't say if
your sequencer is auto-quantizing the input, but that can have an effect.
If your hearing playback that doesn't jibe temporally with what it sounded
like when you laid it down, i.e. tuplets are messed up, then it's either
auto-quantizing (either in or out) or you have insufficient resolution ticks
per quarter wise. Consider a tuplet that doesn't divide evenly and what
happens to the mantissa.
> >record the manuals and try to separate out the parts into separate
> >tracks as a post-processing task. That's going to be a problem when I
> >try to do something else that requires more than one manual since I
> >only have one MIDI keyboard. Also, as much as I'd like to, I don't
> >have the time to learn each piece I'd like to render to the point of
> >perfect performance like I have this one.
> >What do you all do in this situation? The same issues naturally apply
> >to other styles of music as well, e.g. choral or orchestral.
Depends on the chart and what you need for an end product. Lilypond
will produce output from a chart, but it's got it's learning curve. If you
just need the sequencer to "overdub" and an audio rendering is what you're
after, ditching the click is the fastest way to get there.
Um.. all this assumes you're using a sequencer with stable time to begin with.
> This may not be too helpful with your particluar needs, but all complete
> compositions I've recorded with MIDI have been done by necessity one
> voice at a time. With Jazz, Blues, and Rock the key is putting down the
> timekeeping track(s) first. For example, drums and rhythm instruments
> over which melody and color instrumets are added. This approach allows
> both good timing and good expression for anyone at all comfortable with
> automated programs that provide accompaniment for either practice or for
> building up a complete recording.
> I assume your timing is working well but expressiveness is lacking with
> the pieces you're trying to build. If that is so, then it's not a
> simple matter of getting more used to following the metronome, or
> quantizing after recording.
> One possibility would be to record under a duet situation where you get
> someone to play a critical timekeeping part on a keyboard or other
> instrument (guitar) acoustically while you record your first parts
> digitally. That way you'd have the ensemble feel, and also the creative
> interplay with another live musician, both of which should help you get
> the feel down that you'd like to have.
> Others with more experience or even a technological solution to offer
> will be more helpful, so good luck.
More information about the Linux-audio-user