| A while back, before I ever heard of MIDI, I was having a problem properly
playing a synth line for a piece I was working on. It involved continuous
16th notes for many measures at a time. I could play it on the piano because
I'm used to the action but the mushy keys on my synth were destroying my
technique. So I cheated. I set my tape deck to half speed and recorded
a series of notes, each time adjusting the cutoff frequence on the VCF.
Then I played it back at normal speed, decided which VCF setting sounded
most like what I wanted, and then recorded the whole piece at half speed.
When played at normal speed, it was absolutely perfect.
With MIDI, the speed at which notes are played or recorded does not effect
the pitch or tone. Since MPU-401 time stamps are in terms of MIDI beats, an
obvious solution is for us to allow a tempo scaling factor while recording
which would not change the tempo we have stored for the whole piece. I found
that even the stupid Passport software is capable of this and it works great.
Using this technique, I bet you could "play" 99% of what you want to record
without any problems.
So, now you don't have to be fast, but you still have to be accurate. Suppose
you miss a note or two. Rather than re-record the whole track, have the
computer use the punch-in sequence shown on page 40 of the MPU-401 manual
to re-record say one measure and stop. We might also have it lead in with
a specified number of previous measures so you know where you are in the song.
Note I have been presupposing the computer understands measures. I insist that
it know about time signatures and measures.
Speed and accuracy are now handled by software. We're still left with
coordination and timing. Even a virtuoso can forget to concentrate on phrasing
when all ten fingers are being taxed. Two things can help here. First, make
it easy to record chord progressions, parallel octaves and thirds, etc., by
playing each "voice" on a separate track and merge tracks with a keystroke.
That way you can build up a complicated passage in stages. Second, some
commands to adjust phrasing would smooth things out. You might instruct the
computer to adjust the start and stop times of all notes so that they are
rounded to the nearest fraction of the measure which is 1/4 of the duration
of the note, or a 16th note, whichever is larger. Suppose you were trying
to play a dotted 8th note followed by a 16th note, but you played it more like
an 8th note triplet with the first two tied together. The above algorithm would
correct this automagically and also make smooth legato passages out of sloppy
quarter notes.
If all else fails, you can still edit the score here and there, but you won't
have to spend all day at it. What do you think, John?
|
| Last weekend I coded most of the "merge" algorithm. I have rounding already,
but not punchin-punchout. Maybe next weekend.
I added something using "merge" that I hope will help: when you depress a
key that event is recorded at the current time, when you push the "+" key
time is advanced, and when you release a key that event is recorded at the
current time minus the "gap time", to allow for release. This means that
you can play a line, including chords, using "+" to advance to the next moment
of time when you have gotten the right keys depressed. Then you can use
"-" to back up to the beginning of the phrase and add the next line. I haven't
used it enough yet to get a feel for how it works, but I suspect that the
time spanned by "+" and "-" needs to be more easily adjustable.
I'll try your suggestions, though such experiments take time, since I can
only work on them evenings and weekends.
John Sauter
|
| Hi guys,
My system isn't MIDI, but I have the same performance concerns
as you.
One problem I've encountered with entering music at a super-
slow tempo, is that unless it is a division of 2, slowing things
down only makes it more difficult.
Triplets, quintuplets, etc., all become more difficult as
you slow down the tempo. The 'feel' goes away, and you have to do
some heavy-duty concentration to get the thing to come out right.
John, I do have a suggestion for you; take a ride into Waltham
or Boston and find someplace with a Fairlight synth. Their stave-oriented
music input system has gotten rave reviews. Pretend you are a potential
buyer, and get a demo. Perhaps, they have some elegant solutions to
the problem.
Might also be worth checking out the Synclavier II, though I
know less about that.
- Karl
|
| Although I haven't seen the Fairlight or Synclavier I have used a stave-oriented
input system, with my old ALF. After several months of practice I got to
the point where I could input a sheet of music in 45 minutes. If I get really
ambitious maybe I'll try to reproduce it for my current hardware, but currently
I have even my very simple graphics output routines disabled because they
take a lot of memory and are very slow. Visual feedback is not very useful
unless it is fast, as the ALF was.
Is anybody planning to go visit the Fairlight? I'd like to go with someone,
rather than getting lost in Boston on my own. I have been living in this
area since 1969, but I still cannot abide Boston traffic--I always take the
T when going into the city.
John Sauter
|
| I wouldn't mind checking it out. However, it will have to wait as I am
taking a vacation starting tonight until the 15th.
The MPU-401 manual isn't so bad after you read it 30 times. But your rewrite
was very helpful, John. Thank you. I have many things to discuss about MIDI
software after I get back. See ya.
Ted - Rainbow DECnet gnurd and soon to be XXXxxxx DECnet gnurd
(It's a secret, I think)
|