MIDI and the Macintosh

[Protools5LEGuy]

From http://cara.gsu.edu/courses/MIDI/midimac.html


MIDI and the Macintosh
======================

This article originally appeared in three parts in the electronic
newsletter TidBITS, in issues #176/10-May-93, #177/17-May-93, and #178/31-
May-93. The authors have made small changes, corrections, and
clarifications since the original publication. The information appears here
with permission of TidBITS and the three contributors; thanks are due to
them for an excellent and useful article.

TidBITS is a free weekly electronic newsletter that focuses on the
Macintosh and electronic communications. For more information and
instructions on how to subscribe to our mailing list, please send Internet
email to: [email protected]

cheers ... Adam C. Engst, TidBITS Editor -- Sep-93

MIDI and the Macintosh
----------------------
                   by Shekhar Govind -- [email protected]
                   Technical editing by:
                   Craig O'Donnell -- [email protected]
                   and Nick Rothwell -- [email protected]

This Mac-MIDI musical offering is organized in three movements, an
introduction and discussion of MIDI, a look at Macintosh MIDI software, and
finally, some information on MIDI hardware.

1. Introduction to MIDI
    The Antecedents
    The Effects
    How MIDI Works
    MIDI Commands
    A Salute to General MIDI
    Further Readings
2. MIDI Software for the Macintosh
    Applications Software
    Additional System Software
    Gooey Crimes
3. MIDI Hardware
    Interface
    Macs
    Controllers
    Samplers and Synthesizers
    Coda

MIDI and the Macintosh - Part I
-------------------------------

Introduction to MIDI

Picture yourself as a musician, composing and arranging each part for a
musical performance: a string quartet, jazz combo, a rock band, a marching
band, a small orchestra; playing each part to get it just right; printing
the sheet music;, and saving a flawless performance. Did we mention you
could do all this by yourself on your Mac? You are the publisher, the
composer, the band, the conductor, and the sound engineer - all rolled into
one. As Zonker Harris would say "Imagine!" If you'd rather live the
scenario than imagine it, step into the world of MIDI where you can spend
as little as $600 or so for software, an interface, and a used synthesizer,
or as much as $50,000 for a complete MIDI-based production studio complete
with digital hard disk recording which surpasses the sound quality of an
audio CD.

MIDI stands for Musical Instrument Digital Interface. The MIDI
specification  enables synthesizers, sequencer hardware, personal
computers, drum machines, lights, and hard disk recorders, etc. to
interconnect through a standard data protocol via an inexpensive serial
hardware interface. Even though the operating system within each device may
be different, MIDI gives musicians "plug and play" synthesizer-computer
communication as easily as LocalTalk lets Mac owners connect a few Macs and
a laser printer. With a MIDI interface (typically attached to the serial
port) any MIDI-savvy musical instrument can connect to a Mac. With so-
called "sequencing" software running on the Mac, a musical piece played on
the instrument is faithfully "recorded" to a disk file for editing and
playback. (As explained later, the sequencer does not record the audio
sound; it records performance information only, much like a computer
punched-paper-tape reader recorded not sentences that were typed, but which
keys were pressed in sequential order; or think of a player piano, where
the piano roll doesn't contain the sound, it contains information about
what keys should sound together and for how long.)


The Antecedents

It is important to remember that MIDI was created to simplify live
performances. During the 1981 fall convention of the Audio Engineering
Society, Dave Smith and Chet Wood, two engineers from synthesizer
manufacturer Sequential Circuits (creators of the popular Prophet-5
keyboard) proposed an industry standard for an electronic musical
instrument interface. The idea was that performers should not have to
create custom cables and devices to connect synthesizers. Instead, they
should be able to "plug and play" with units from different manufacturers.
(This was not the case before, when Moog synthesizers could not talk to ARP
2600s and neither would talk to Buchla Music Boxes.) Dubbed the Universal
Synthesizer Interface (USI), this draft proposal was modified by the
techies of synthesizer manufacturers Oberheim, Roland, Korg, Yamaha, and
others of their ilk. A consensus was orchestrated on the revised proposal
and in late 1982 (drum-rolls please) the first set of universal MIDI
specifications was adopted.


The Effect

MIDI turned into an unanticipated success, rocketing sales in the
synthesizer category to the top of the musical instrument industry within a
few years. New companies like Opcode and Digidesign appeared overnight in
what had previously been a sedate and technophobic industry. In the early
1970s the best-selling synthesizer keyboard (the MiniMoog) sold only about
12,000 units, and in the late 1970s the best seller (the Korg Poly6) sold
some 100,000 units; the best seller during the dawn of the MIDI age, the
Yamaha DX7, combined new sounds and MIDI to sell at least triple the
previous record (exact numbers are hard to find).


How MIDI Works

MIDI translates a predefined set of performance events at one instrument,
called the master controller, into digital messages that are sent to other
devices over a low-speed serial link operating at 31.25 kbps - about twice
the speed of a v.32bis modem. To make it easy to keep musical information
going where it should, these events are encoded on any of 16 independent
logical channels within the MIDI data stream. A synthesizer receiving this
incoming data stream responds by playing music.

Imagine playing a series of half-note C major chords on Middle C on a DX7
synth wired to one or more other synths. In this case, the DX7 sends three
"note on" messages, three note numbers, and three "note off" messages each
time you sound the chord - the receiving MIDI device plays a matching chord
in perfect synchronization with the DX7. But (and this is a big but )
depending on its settings, the receiving instrument may use a different
instrument sound, or "patch" (a patch being a particular synth voice -
grand piano, hot guitar, sax, viola, what have you.) The chord is the same,
but the generated sounds within each synth may differ. In other words, MIDI
keeps track of the performance events, and not the audio sounds. (An
apparent contradiction to this rule comes with MIDI drum kit and percussion
sounds - certain MIDI note numbers are usually associated with certain keys
on the keyboard since playing a "cowbell chord" or "tom-tom chord" usually
makes no musical sense.)

A MIDI keyboard can also control a number of sound-producing synths without
any computers involved, and without any recording of the MIDI data. As an
example, consider a DX7 wired up to a Roland Sound Canvas which is in turn
wired to an E-Mu Proteus. (Sound Canvas and Proteus are "sound modules" or
electronic musical instruments with a synth's circuitry but without a
keyboard.) The musician plays a series of half-note C4 on the DX7 keyboard
- patched to sound like a piano. Notes, timing, and other performance
information is transmitted to the keyboard-less Sound Canvas and Proteus
sound modules (which could be patched as, say an organ and strings
respectively).

Schematically, it would look like:

(master)
   DX7 -->MIDI cable--> Sound Canvas -->MIDI cable--> Proteus
   plays C4             plays C4                      plays C4
   as piano             as organ                      as strings

The two sound modules play the same note as the DX7; but the actual sounds
generated within each module use a different instrument, or patch. To hear
all three instruments, you need an audio mixer to blend the three pairs of
stereo outputs into one pair of outputs; this "master output" then goes to
a stereo amp and speakers. (Professional MIDI hardware rarely boasts a
built-in speaker.)

Prior to the MIDI spec, people did their data recording and editing with
special hardware. Some of the most sophisticated pre-MIDI systems came from
Sequential and Oberheim and consisted of keyboards, drum machines and a
hardware recorder (called a "sequencer") connected by proprietary data
links and cabling. Around the same time Fairlight and PPG offered
integrated systems controlled by a piano keyboard, keypad, and CRT to
display waveforms.

Here is an example of a simple Mac-based MIDI setup. A MIDI keyboard (we'll
stick with the DX7) interfaces to a Mac serial port with a $60 MIDI
interface and two MIDI cables:

- one from the keyboard's MIDI output to the interface MIDI input
- one from the interface MIDI output to the keyboard's MIDI input

Schematically, MIDI data travels like this:

[DX7 MIDI out]-->>MIDI cable 1-->>[interface MIDI in    ]
[            ]                    [interface serial port]<->|
[DX7 MIDI in ]<<--MIDI cable 2<<--[interface MIDI out   ]   |
                                                            |
[Mac serial port]<->---------<->serial cable<->----------<->|

The MIDI data links are unidirectional to keep everything simple and
inexpensive. The two MIDI data links convert to a bidirectional serial
signal inside the MIDI interface. The interface also locks the serial port
to its own special 31.25 kHz clock.

Consider this. You launch an inexpensive sequencer program like Opcode's
EZVision and tell it to record incoming MIDI data. When you play a note on
the synth, a message is sent to the Mac identifying the key, how hard you
struck it, for what duration you held it down, etc. The software stores
this information in a file. Once all performance information has been
recorded, you can edit individual musical events on the screen in much the
same way you edit text in a word processor. Saved MIDI events are time-
stamped to insure that they happen in the proper order on playback; in
other words, when a piece of MIDI music begins, the sequencer starts
beating out an ever-increasing number of timing pulses and keeps ticking no
matter what. Even if the piece includes a minute of silence in the middle.
Even if the piece, a la John Cage, is simply 4 minutes and 33 seconds of
silence.

To reiterate, a MIDI sequencer file is only performance information, not
the sounds themselves. The universal acceptance of MIDI has made it
possible to use software sequencers instead of the earlier proprietary
hardware sequencers.

The raw MIDI performance data can be edited, looped, reversed, the tempo
can be changed for playback (most sequencer software supports an untold
number of tempo changes), and any part or the entire piece can be
transposed to any key. In short, the data can be processed separately and
in a manner quite different compared to anything in the audio domain. If
the software has a notation editor, the MIDI data can also be displayed on
screen as music notation, and printed as sheet music. Finally, the file may
be output as MIDI commands back to the synth for flawless playback.

The "Switched-on Bach 2000" audio CD showcases what a talented performer
can do with MIDI. Wendy Carlos's 25th anniversary re-recording of Walter
Carlos' hit classic(al) album "Switched-on Bach" was produced on a Mac
IIfx. Wendy Carlos owns a stunning array of advanced synth gear, however,
so remember that the Mac isn't making the sounds; the synths are. By way of
comparison, the original album used multitrack tape and primitive
sequencing hardware.


MIDI Commands

MIDI commands are 8-bit (sandwiched between 1-bit start and stop messages)
binary serial messages with 16 encoded channels. A master controller, one
cable, and a slave device make up the simplest possible MIDI network. Once
a computer is connected to the MIDI network, messages captured by a
sequencing program may be saved as a Standard MIDI file, a cross-platform
standard. This means that MIDI music is, to a great extent, device-
independent. A Standard MIDI file created on a DX7 and a Mac can play back
on a different synth connected to a PC clone.

While most synths respond to the complete set of MIDI messages, some older
and cheaper models don't. The scope of this article does not permit a
detailed explanation of the various MIDI messages.


A Salute to General MIDI

Many late model synths understand "General MIDI," a new subset of MIDI
specifications from the MIDI Manufacturers' Association. In a nutshell,
General MIDI specifies a 128 pitched instruments (sound patches) and 47
percussion instruments which all General MIDI synths must play.

Why the need for General MIDI? Well, to start with, for years and years,
each synth manufacturer invented a proprietary "map" of patches, or voices.
As an example, a Roland synth and a Korg synth would both have at least one
Grand Piano, usually several.

The problem:, the Grand Piano "address" in the synth ROM would be different
for each - or put another way, the two synths would assign different patch
numbers to the Grand Piano sound.

To put it in a Macintosh-type perspective, it would be as if every word
processor developer arbitrarily chose what Command-Q, Command-S, Command-A,
and Command-Z, -X, -C and -V would do. *Something* will happen but you
can't predict what.

Or think of it as WYPIWYG: what you played is what you get.

Furthermore, one synth might have 48 different Grand Piano sounds and
another might have four. An expensive synth might have 256 preset patches
plus hundreds of user memory locations, and a cheap one, 16 presets and 16
user memory slots.

This free-for-all made it impossible to take a fully-orchestrated MIDI file
created on a Korg M1, load it into a computer, and play the music as the
composer intended it to be heard on an  E-Mu Proteus. You'd get music all
right, but instead of violins during the intro, you might hear bassoons. Of
course, the way chords are voiced within an arrangement is a paramount
concern for composers. None wants a lyrical flute passage with sustained
strings in the background to be played by fuzz guitar with cymbal-crash
"chords".

So for the music to sound as originally intended, someone would have to
revoice (or "remap the patches") the arrangement for every new output
device. So we lied to you a little bit. MIDI files aren't strictly device-
independent when it comes to playing the **original** sounds. All they can
say is "use patch 35, velocity 102, for this part". They don't specify
which sound is associated with the patch number. Think of it in this way:
Adobe Type Manager and TrueType both specify which font outline to scale to
what size. But what would a printed page look like if the font information
only consisted of "use a serif font, 36 point"?

To use MIDI in multimedia, and to put MIDI chips on inexpensive sound
cards, there has to be agreement on what musical sound corresponds to which
patch number. Remember, MIDI is tone-deaf and doesn't know a Hammond Organ
from a Tam-Tam. MIDI just broadcasts signals such as: "Yo! Synth on Channel
1! Set Patch 45! Now play these chords!" Unfortunately, with complex
orchestrations, the results can be unintentionally hilarious. A piece of
well-crafted music ends up sounding more like the Portsmouth Sinfonia,
Spike Jones, or Peter Schickele.

General MIDI solves this because it specifies instruments and patch numbers
which synths can share as a least common denominator. Of course, any
manufacturer is free to go beyond General MIDI in a given product, or just
ignore it.

General MIDI also addresses a question that's a shade more esoteric - "What
do I do with the drumkit?" (Musicians who play live would probably phrase
this as "What the h*** do I do with the drummer?") In MIDI, multiple
drumkits could possibly be contained in a single patch with individual
drums and cymbals assigned to different notes on the piano keyboard. For
example, a drum patch on your keyboard might map C2 to bass drum, C#2 to a
rim shot, D2 to a snare drum, D#2 to a china cymbal etc. (Yes, you can play
drums from the keyboard!) Different drumkits could be different patches.
You might have:

Patch #    Drumkit
  45       light jazz kit
  46       rock kit
  47       electronic rock kit
  48       orchestral percussion
and so it goes.

A general-purpose synth needs to listen for drum commands on a given MIDI
channel so that the notes come out as hi-hat and snare instead of as
flugelhorn. We have already discussed that General MIDI specifies a
standard patch number for a particular instrument (including drums). But
which of the 16 possible channels could possibly be broadcasting the drum
events? Well, prior to General MIDI there was no default channel number for
drums that everyone agreed on. Now there is - Channel 10 is reserved for
the patch with drum kits.

In a certain sense, General MIDI is restrictive in that it makes
instruments conform to a limited set of sounds and a minimum polyphonic
capability (polyphony is the ability to play multiple notes simultaneously,
for example, a string quartet would be four-part polyphonic music. A piano
is polyphonic, but the number of voices could be as many as 88 if the
sustain pedal is on and every key is struck somehow). This standard is not
necessarily the future of MIDI and synthesis; it is merely a robust lowest
common denominator for people who want to orchestrate music using a
predefined sonic palette. General MIDI music can be saved as MIDI files and
will continue to sound similar on different General MIDI savvy synths (for
example, for multimedia applications) without requiring patch remapping.


The MIDI Specification
-------------------------

The MIDI specification can be purchased from International MIDI Association
(which is just that - a worldwide MIDI user group):

International MIDI Association
5316 West 57th Street
Los Angeles, CA 90056
(310) 649-6434.

Other technical information about MIDI is available on the Internet via FTP
from, among other places, , , ,
and .


Further Readings

Don't be lulled into a false sense of complacency. Like any computer
communications language, MIDI becomes complex once you move beyond a simple
setup with a couple of synths and a Mac. (Just as integrating Macintoshes
into a PC network is more challenging than setting up a couple of computers
at home with System 7 File Sharing.)

For further edification, you may want to delve into MIDI reference books.
Steve De Furia has authored (and coauthored) several informative general
and Mac-specific MIDI books. Keyboard Magazine has published several useful
volumes and "Special Focus Guides" for a detailed look at MIDI and synth
basics. Craig Anderton's readable "MIDI for Musicians" is a classic. Most
libraries (and fine bookstores) offer at least a dozen other publications
about using MIDI and creating MIDI software. Like most things technical,
MIDI is a moving target and new books appear each year. Leading MIDI-
related magazines continue to be the well-known Keyboard and Electronic
Musician.


MIDI and the Macintosh - Part II
--------------------------------

MIDI Software for the Mac: Application Software
For simple purposes, MIDI application software can be categorized into one
of two main types - a recorder/player (sequencer) with non-real-time MIDI
editing tools, or a music-notation editor to create printed scores, or
"notation,"  from a MIDI file or a performance. (Other classes of MIDI
software esoterica will be dealt with later.)

Until a few years ago, the Amiga and the Atari ST boasted some of the best
MIDI software. However, highly acclaimed MIDI sequencers and notation
editors are now available for the Mac, and many people believe it will
continue to be the professional's computer of choice. As more MIDI software
is ported to (or created for) Windows, the balance may change. But MIDI
editing remains fundamentally a graphic process, a task at which Macs
continue to edge out PCs.

Sequencing packages are geared to "conventional" music making. They have
multiple recording tracks, let you display and edit notes, and so on. They
often present a tape-recorder screen metaphor, with music being recorded
and played from beginning to end, linearly. Notation editors provide
seamless translation of standard music notation into MIDI files (and vice
versa). A good analogy for a notation editor and a synth keyboard would be
a word processor and a QWERTY keyboard.


Mac sequencers and notation editors

A partial list follows. Apart from a few exceptions (such as Metro,
arguably "the biggest bang for the buck"), the list price is a fair
reflection of the software's capabilities. The high- and mid-range packages
are for the professional musician, while the low-end programs are aimed at
the semi-pro or casual musician or composer.