T.R | Title | User | Personal Name | Date | Lines |
---|
563.1 | marginal | SAUTER::SAUTER | John Sauter | Fri Nov 07 1986 08:10 | 15 |
| I have an Apple II-MIDI setup. There is software available, using
either the Roland MPU-401 interface or the Passport interface.
The problem with using an Apple II is its lack of memory. Even
without pitchbend or release velocity you can't get more than
a few minutes of piano music into memory.
At the time I got my system there was no decent software available,
so I wrote my own. It uses the Roland interface and provides a
very poor interface to the user, sort of like TECO without echoing.
I am also looking for a (better) PC to do music with. My first
choice is the Amiga, but I am waiting for a hard disk and some
music software. Second choice is the Atari ST: a smaller system
but significantly less expensive.
John Sauter
|
563.2 | See KEYBOARD | ECADSR::SHERMAN | | Fri Nov 07 1986 10:08 | 8 |
|
KEYBOARD's MIDI issue lists about seven software packages available, ranging
from 3000 to 8000 notes, most at about $100 to $250, and available for a
variety of interfaces. If I were to pick what I would consider the "best",
I'd go for Master Tracks by Passport with a Passport interface. The software
would be in the $200 range, and the interface would be about the same. I'd
shoot for an interface that has tape sync.
|
563.3 | will it crawl enough ? | ADVAX::T_ROBERT | | Sun Nov 09 1986 14:17 | 12 |
|
"few minutes of piano music" ?? So clarify for me, if I only
have 64K, would I be able to record midi info for two keyboard parts,
and a drum machine, say for about a four minute composition ?
Seeing that Master tracks has eight tracks available, and
it's in the 3000-8000 note range (depending on memory I assume)
so even 3000 notes, it would seem that should do what I want.
But then again, I'm not very famialar with the way MIDI is recorded,
is space (memory) eaten up even when no events are occuring, or
does it just grab single events as they come through ?
|
563.4 | sequencer time estimating | SAUTER::SAUTER | John Sauter | Mon Nov 10 1986 07:55 | 51 |
| re: .3--Your question does not have a precise answer, but here are
some details. In the MIDI data stream, a note consists of a "note
on" and a later "note off" message. If you are not recording
aftertouch, pitch wheel, any pedals or release velocity each of
these messages takes 2 bytes: the note number and a byte which either
specifies the "on" velocity or is 0 to specify "note off". If you
do anything complex each message takes 3 bytes, but let's assume
the simplest case, requireing 4 bytes per note.
However, 4 bytes per note isn't enough for storage inside a sequencer.
The duration of a note isn't represented as data bytes in the MIDI
stream but as the interval between messages. In a sequencer it
must be represented as data bytes. How this is done varies between
sequencers, but mine uses one additional byte per message to record
the time since the last message. If there is lots of silence it
uses about one byte per second to record long intervals. Thus,
a minimum of 6 bytes per note is required.
Now lets start from the other side: you have a 64K Apple II. Some
of that is taken up by program. The Apple's CPU is a 6502, a processor
with a very primitive instruction set, so lots of code is required
to produce a good program. In general, the better the program
(in terms of useful features and good human engineering) the more
memory it will take. I'm not familiar with Master Tracks, but let's
assume it takes 16K of memory, leaving 48K for buffer space.
In the best case, 48K/6 = 8K notes, so their promise of 8000 notes
seems reasonable, but only in the simplest case. If you are using
multiple instruments you will frequently need the full 3-byte MIDI
message, or 8 bytes per note, which reduces your storage to 6000
notes.
Assuming you like drums as much as my daughter does, you will get
about twice as many "notes" from the drum set as from any one keyboard,
so the 6000 notes should be allocated as follows: 3000 to the drums
and 1500 to each keyboard.
Now we come to the part that depends very much on the kind of music
you are playing, so my estimates may be way off. Some music plays
lots more notes per second than others. Remember that each note
in a chord counts as a note. "Moonlight Sonata" is less than 10
notes per second, but it's a pretty slow piece. If you're doing
anything modern you should probably figure on at least 15 notes
per second. At 15 notes per second per keyboard you can count on
about 100 seconds of recording time, or just over a minute and a
half.
My estimates may not be correct for the kind of music you are doing,
of course, but I think I have provided enough information that you
should be able to do the estimates yourself, using better parameters.
John Sauter
|
563.5 | drummers are privileged... | BARNUM::RHODES | | Mon Nov 10 1986 11:19 | 15 |
| I might add that I believe that some drum machines don't send note-offs,
so that is one less byte (word?) per note than a musical keyboard note.
This probably means that the drum machine will take up approximately the
same number of bytes per unit time as the keyboard. Len probably knows more
about this than I do, tho...
As soon as I get a MIDI interface for my C64, I will do some recording with
the Dr. T sequencer that I have. I will be happy to post the results of
many of the experiments that I plan on undertaking aimed at the space
limitation problem that the C64 has (which is similar to that of the Apple
II [64k memory, 6502 processor]).
Good luck,
Todd.
|
563.6 | Run, run, run, butcha just can't hide... | JAWS::COTE | Luckless pedestrian... | Mon Nov 10 1986 11:37 | 3 |
| Do drum machine use running status?
Edd
|
563.7 | Note Off = yes, running status = ? | SAUTER::SAUTER | John Sauter | Mon Nov 10 1986 14:14 | 16 |
| re: .5, .6--My experience with drum machines is limited to the
Roland RX15 and the Octapad, but I've been able to use them both
with my software, so they must send Note Offs. One of them (I think
the Octapad) has a parameter called "gate time" which controls the
interval between Note On and Note Off.
Not sending Note Off would be a violation of the MIDI specification.
I wouldn't fault any device which refused to receive such a message
stream--it would look like a very busy chord!
Even if a particular device doesn't use Running Status a seqeuncer
can still use it for its internal storage. Of course, Running Status
doesn't do you much good if you have multiple instruments, or pitch
bend, or after touch, or release velocity, etc, etc, since the status
byte will change on just about every note.
John Sauter
|
563.8 | Drum Machines and Note Off Etc. | ERLANG::FEHSKENS | | Mon Nov 10 1986 17:11 | 24 |
| Drum machines *do* send note offs (the gate time, or time to the
note off message, corresponds to the step size) when used as
sequencers, but they ignore them (and don't require them) when they
receive. *If* the sequencer assumes that the drum machine uses one
and only one channel, then the sequencer can use running status, i.e.,
not have to send the channel number/command byte and just send lots of
note numbers.
In any case, it need not *store* those bytes in its internal memory,
as it can "reconstruct" them at output time. This is one of the reasons
the Roland MC500 has a rhythm track - it allows *much* more efficient
use of memory than storing drum parts as fully general MIDI messages.
In fact, the MC500 can only control 32 voices on its rhythm track,
so it only needs 5 bits to identify the voice (which it probably
uses as an index into a table of note numbers, and possibly channel as
well), and could have 3 bits left over for something else.
The Octapad (*great* gadget) has a gate time parameter because it
has no notion of step size. The step size is the time resolution
for programmable events - on Roland drum machines it is determined
by the selected "scale" (12/16/24/32 steps per 4/4 bar).
len.
|
563.9 | 5500 Notes Not Nearly Enough | ERLANG::FEHSKENS | | Mon Nov 10 1986 17:16 | 9 |
| My old Roland MSQ-100 purported to hold up to 5500 notes, and I
filled it to the brim with the first 51 bars of the first movement
of the 6th Brandenburg concerto (4 parts of mostly 16th notes, and
2 parts of mostly 8th notes). If I add that all up, that's no more
than (51*16*4+51*8*2) or 4070 notes, so there's some overhead in
there too.
len.
|
563.10 | Is a megabyte on a IIe enough? | VRTMP1::ROBBINS | Worth Robbins | Sun Oct 18 1987 09:47 | 8 |
| How about an update on this Apple II note...
I have an Apple IIe, with a megabyte of RAM. My son just bought
a CZ1000. Given that "installed base", is there a reasonable "upgrade
path" that would give him (me) sequencer capabilities that are decent?
Has anyone produced a quality piece of software that takes advantage
of a IIe with lots of RAM?
|