When Is A Computer? (Part 5)
Why is a MIDI?
Sometimes "One, Two, Three... GO" is not enough. You need communication. This month Andy Honeybone looks at how micros and synths talk to each other, weighs the importance of MIDI, and predicts the success of MSX, its computer equivalent.
Sheet music was once very popular. Prior to the 1940s, a Number One was awarded for music rather than record sales. The reason was simply that most parlours were furnished with a piano, but a phonograph was the plaything of a select few.
The format of popular vocal sheet music has changed over the years. Gone are the hieroglyphics of the tonic sol-fa (doh re me etc) system which enabled suitably practised persons to sight-sing the melody. Vanished are the ukulele tunings and chord boxes, now replaced by their guitar equivalents. The guitar is the latest in the succession of 'most popular' instruments, it having toppled the accordion several decades ago. Fodder to the guitarist is of course the chord symbols and these form an essential part of any present day song-copy.
The reason for this notational preamble is to provide some analogy for the important topic of communication – including the transfer of musical information from performer to instrument and from instrument to instrument. Verbal communication has been covered fully in a previous One Two (Jan 84 p58) wherein the eloquent Reg speaks out. For this article the less florid technical aspects will be considered.
Sheet music is an example of a means of communication. Its content has changed to accommodate new developments and fashion, and yet despite certain deficiencies it has survived for centuries and transcends all language barriers — not bad for just five lines and some dots. Further, the same copy can provide chords for a guitarist (reading the symbols), lyrics for a vocalist, a bottom line for a bass player, the number of measures for a drummer, or a complete harmonised arrangement in the hands of a good' pianist.
An attempt to provide a similar, multi level, common communications medium between electronic musical instruments has resulted in the Musical Instrument Digital Interface – MIDI for short. By this connection, MIDI-compatible synthesisers, sequencers, drum machines and computers can be linked to allow synchronisation, sound layering, remote control, memory dumps, and sharing of common peripherals (printers, disk drives, plotters and so on).
In the same way that the guitarist ignores the musical notation when reading chord symbols from sheet music, so a MIDI connected drum machine will ignore any pitch information which is sent to it. The guitarist is only interested in chord symbols, the drum machine is only interested in tempo (real time) information.
The most important question about MIDI is not 'what?' but 'why?'. The answers are lurking behind a smoke-screen of intention. The system does allow for MIDI-compatible mono and polyphonic synthesisers from different manufacturers to be slaved together, and for home micros like the Commodore 64 to be pressed into service as effective sequencers. Drum machines can also be synchronised to sequencers (but this has been possible for some time without MIDI). The current message seems to be 'wait and see' and as with all computer applications the hold up is from software development.
Another stumbling block is that one layer of the MIDI code is intended only to be read by instruments of the same brand as the transmitter (system exclusive). This is great in that it will allow changes in technology and fashion to be incorporated, but it also means that the standard exists only at the most primitive levels of pitch and timing. This greatly reduces the scope of 'one size fits all' software as a music composition tool.
For software creators the problem is the large number of permutations of synthesiser and home computer, each requiring a specific solution. Help is at hand in the shape of three letters: MSX. This is a standard for a home micro computer drawn up by Microsoft (of BASIC fame) primarily for the Japanese. It specifies not only the dialect of BASIC but also the type of microprocessor, video and sound chips, amount of memory, pin connections for plug-in program cartridges, sockets for joy sticks and musical keyboards, and everything else to ensure compatibility across various manufacturers' products. It is also permissible within the standard for a manufacturer to produce just one 'maxi-chip' for the whole computer as long as compatibility is maintained. This could mean some very cheap micros. When MSX becomes widely available, MIDI software will be reduced to MSX – Prophet T8, MSX – JX3P, MSX — DX7, and so on.
Yamaha have now entered the home computer business with an MSX machine: type in 'CALL SOUND' and plug in a mini keyboard and you have a poly synth; plug in a card-reader cartridge and run a 'Playcard' through the slot and an on-screen menu (list of options) allows you to edit the voices and change the pitch and tempo of your chosen song. The Playcard is another example of communication, having the tune written in musical notation on the upper part of the card and a magnetic stripe at the bottom carrying the same information in coded form. Other add on modules for this computer include a MIDI interface and an FM synthesis unit. A whole host of magazines devoted to MSX software can also be expected.
On the subject of getting musical information into our computer instruments we should not forget bar codes. These are the blocks of black and white lines found on everything from cans of baked beans to songbooks for the Casio CT 701 and MT 70. Several codes are in use, the most popular and complex being the Universal Product Code (UPC). Some codes use the white gaps just to separate the black bars, some use up to three line widths of each shade for encryption. The reading error increases as the bar lines become closer and so print quality has to be carefully watched. This results in bar codes being rather bulky for the amount of data they can hold. Nobody wants to scan a wand over five sides of eye-boggling lines to load a piece of music. Unless you have a dot-matrix printer, bar codes are a read only medium. But they can be practical, as the Casio machines prove.
One tetchy point: a bar code scanning wand is not a light pen. A wand is a combined light source and photocell with some optics to bend the light on and off the printed codes. A light pen is a photocell mounted in a tube which can be touched on the screen of a visual display unit. Because the light pen's electronics are much faster than the eye it can see the single electron beam which is writing the display. When the beam coincides with the position of the pen, the photocell sends a signal to the computer which records the row and column numbers corresponding to that point. The light pen 'draw your own waveform' idea of the Fairlight has its roots in the chain and sprocket 'Oramic graphic system' invented a good while ago by Daphne Oram. On this machine the envelope wave forms were drawn on clear film and cranked over a photocell.
Dialogue between computer instruments involves the passing of digital information. Digital words generally come in multiples of eight bits known as bytes, so a method of communication over eight wires could be envisaged. Problems would arise in knowing when to send the next word to be transmitted — and how would we know if some other device also connected was trying to talk? The answer lies not in the soil but in the provision of some extra 'bus control lines'. The signals on these wires supervise data direction and transfer by a series of 'hand shakes'. Without getting too involved, this means that a line is either busy or ready. If the computer finds a line busy then before it can send another word it must contemplate its navel until the control line gives the OK to go ahead.
This side-by-side bit transfer is known as a parallel communications link; the eight data lines and their control signals are collectively known as a bus. The parallel arrangement gives greatest speed but runs into problems with cost, interference, and driving long runs of cable. An example of this kind of connection is found on Hewlett-Packard and Commodore computers under one or another of its many pseudonyms (HP-1B, IEEE-488, GP-1B, 1EC-625). If you can imagine a digital word as a line of eight wooden cubes on a table, then serial communication would be the process whereby a push on the block furthest from the table's edge would cause each block in turn to fall off the table. In other words, the byte is transmitted bit wise. Serial communication has the advantage of two-wire transfer over reasonable distances with good immunity to interference. The 'bacon-slicer' which separates each digital word into its component bits for transfer is a chip which goes by the name of a UART (Universal Asynchronous Receiver/Transmitter). The UART tells the computer when it is ready to shovel out another data word, and also if anything has been received. The rate at which the bits are sent (the Baud rate) is set by a crystal clock; both receiver and transmitter must be set at the same rate for correct transfer to take place.
The most famous (perhaps infamous) serial standard goes by the name RS232. The BBC micro has the compatible RS423 interface which uses lower voltage levels and can drive further and faster. Of most interest to ourselves is the serial communication standard MIDI. For diverse reasons, MIDI does not conform to any existing standard. The time taken to transmit data serially is much longer than by a parallel bus, and even though the MIDI transfer rate is fast (31.25 thousand bits per second), mumbles are coming through that it is not fast enough. One nice touch is the provision of opto-isolation. This is a technique akin to sending Morse code between ships using a shuttered (Aldis) lamp. Electronically, current from the serial signal lights an LED encapsulated in a package also containing a phototransistor. The transistor 'sees' the pulse of light and converts it back to a digital signal. By this method there is no electrical contact and noisy (hum) ground loops are avoided.
So far, all communication has been assumed to have been through wires. But before long this practice will seem antiquated. Fibre optics have moved in great strides from the Tomorrow's World mare's tail ornaments and can now offer very high bandwidth transmission. Using laser drivers and audio multiplexing it is possible to send all the inputs to a remote mixer, down just one slim optical fibre. Mixer makers Neve do just this with the digitised microphone signals between their analogue-to-digital input channels and the main desk of their DSP (Digital Sound Processing) mixer.
Feature by Andy Honeybone
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!