Magazine Archive

Home -> Magazines -> Issues -> Articles in this issue -> View

Recording Techniques (Part 10)

MIDI & Multitrack

Part 10: David Mellor looks at ways of getting the best of both worlds by combining a MIDI system with a multitrack tape recorder.

MIDI systems will never replace multitrack recording. Well, that's a bold statement to open with and one which is guaranteed to raise the hackles of committed MIDIphiles worldwide. But at least given the present state of the art, any MIDI system, no matter how sophisticated, can have its capabilities increased a thousand fold by the addition of even a modest 8-track tape recorder. Let me explain...

Suppose you have a sampler or multitimbral synth with eight voices together with a sequencer to drive it. This means that you have the possibility of having eight different sounds on the go at any one time. Add an 8-track recorder, which is reduced to seven tracks due to the necessity for a sync track, then you can have all eight voices simultaneously sounding on all seven available tape tracks. 8x8x8x8x8x8x8 = 56 voices. See what I mean? And if you could synchronise the sequencer to the tape, giving an extra eight voices for the final mixdown, then you would have a grand total of 64 possible different sounds playing at the same time — all originally from the one instrument. With a bigger MIDI system, the possibilities are even more vast (far too big for my calculator to cope with).

Of course, multitrack recording will not always mean tape. Soon it will mean magnetic hard disk recording; at a later date it will mean optical disk recording, and possibly complete integration into a sampling and synthesis system under MIDI control. But multitrack recording is a powerful technique now and, however it is achieved, always will be.

Supposing that the end product of our endeavours is going to be music, which is usually thought of as a creative and artistic thing to do, then what we require of our recording system is that it should be as versatile as possible, allowing creativity free rein with very few restrictions. The combination of MIDI and multitrack can do this. As we shall see, a MIDI system in itself is good for musical experimentation and development. Adding a multitrack not only enlarges the artist's palette, where colours are mixed and worked together, but also provides the canvas upon which a sound picture may be gradually built up.


As I have already outlined, a MIDI system is inherently limited by the fact that whatever it is being asked to do, it must do everything at the same time. If you are using Program 1 on your effects unit for the synth pad, then you can't simultaneously use Program 99 for the snare drum (not until they develop 'multitimbral' effects units that is). With pure multitrack recording, then it is no problem at all to use one particular effect (eg. reverb) when you record on Track 1, and quite another effect (eg. flange) from the same effects unit on Track 2. But multitrack, by itself, has severe limitations...

Since multitrack tape is a linear medium, ie. it starts at one end and plays through to the other, it allows very little scope for experimentation with the structure and the changes of tempo in a piece of music. These things have to be worked out thoroughly in advance, because once the basic tracks are recorded, then it's a long way to go back and start again if things are not quite right. In some ways, this is not a bad thing, because it encourages the use of the most powerful system for music editing there is — the human brain. If you can hear the music in your head, and perhaps strum a guitar or plonk away on a keyboard, then the structure of a piece can be worked out to the minutest detail before a note is recorded.

But while this way of working is good for musical geniuses, and for ordinary musicians who don't mind taking the risk of getting trapped into following familiar patterns of thought, the alternative method of improvising, then making that improvisation better by polishing, restructuring, and making additions, is a powerful way of coming up with new ideas, developing them in new directions, and generally making the best of one's musical abilities.

MIDI systems which incorporate a sequencer are ideal as an aid to musical creativity, because they allow you to start from the germ of an idea and build it up gradually into a finished piece of music, while all the time being available to play back the piece in its current state. It's as though Beethoven had a team of copyists and an orchestra of musicians at the ready while he was writing his symphonies — well almost. You can begin with something as simple as a two bar drum loop, expand that into an eight bar section, add a bass line and chords, copy that into a new section and alter the parts slightly, create a chorus in the same way, shuffle the sections around, add bits, take away what you don't like... the sequenced MIDI system is as representative of freedom in musical thought as a chunk of concrete hacked off the Berlin Wall is of political freedom.


MIDI sequencers and multitrack tape recorders, of course, need to be synchronised together if the music isn't going to be turned into the aural equivalent of scrambled eggs. It almost seems like the dim and distant past when this was thought of as a real problem, but it really is only within the last two or three years that MIDI sync has got itself properly sorted out. There are basically two options available — two sensible options that is.

The best way of synching a sequence to tape is to use a sequencer which has a built-in SMPTE/EBU timecode capability. Take my word for it, with one of these you will have no problems (well, not very many). Possible candidates include C-Lab's Creator/Notator and Steinberg's Cubase for the Atari ST computer, which both have accessory units which plug into the ST and the multitrack, and integrate totally with the software. Akai's ASQ10 hardware sequencer has SMPTE built in and is one of extraordinarily few fully featured sequencers that come in one box.

The other option is to use a MIDI/timecode synchroniser, which converses with the sequencer via MIDI sync pulses and song position pointers. The FriendChip SRC/AT is an excellent example of the type and can be used with virtually any sequencer. However, the problem with this type of unit is that vital information — such as the SMPTE start time and the tempo of the music — have to be kept in a notebook rather than being stored with the sequence on floppy disk. The method works, but for that and other reasons it's not nearly as elegant as having a sequencer with its own SMPTE/EBU. (A third method of sync involves converting SMPTE/EBU timecode to MIDI Time Code, but since this use of MTC isn't as common as it one day might be, I'll do nothing more for now than mention it as a possibility).

For the purposes of this article. I'll assume that the sequencer has SMPTE built in. There's more to say on synchronisation, but that will come in due course.


The first step in creating a piece of music by the MIDI/multitrack method is to get hold of some equipment. But what equipment and how much of it? On the MIDI side, what you need is enough sound producing machinery to get a composition started purely by sequencing, with a wide enough range of sounds to be able to finish it off in the multitrack domain.

Rather than dive in and think about the exotic equipment we might like to have, let's think about the music first. What does the music need? I might say that the minimum would be a drum part, bass line, chords, and melody. That implies at least three MIDI synths and a sampler (or fewer if you have a multitimbral synth which offers some drum sounds and is easy to use). But you can do a lot with less than this.

With a good sampler, such as the Akai S950, and a modest synth it's quite possible to sketch out a composition with just drum sounds and a single synth sound on two MIDI channels. Using the equipment like this, in a very simple way, gives the brain a chance to work out the music and plan how it will develop. Then, at a later stage, this initial sketch will be gradually coloured in and recorded line-by-line onto multitrack tape.

When the outline of the composition is complete, with a finished structure and all tempo changes decided, it's time to get on with some real work — you've had your fun for the moment.

Assuming you are starting from square one with your system, then it's time to take out that free reel of tape that came with your multitrack recorder. Since the tape is completely blank, you'll need to stripe it with some timecode (I wonder whether manufacturers will ever offer us pre-striped tape?). Striping simply means recording 30 or so minutes worth of timecode all the way along the tape, usually on the highest numbered track — 8, 16, or 24. To do this, proceed as follows:

Connect a cable from the timecode output of your sequencer to the appropriate input of the multitrack.

Select the type of timecode that you wish to use — 25 frames per second is the EBU (European Broadcasting Union) standard, so we might as well be good Europeans and use it. Unless you take the tape to another studio, it really doesn't matter which timecode type you use. The generator can be set to start from any time below 23:26:40:00, but for our purposes 00:00:00:00 is quite good enough.

With the tape recorder set to 'record ready' and any dbx noise reduction turned off on the timecode track (Dolby C is fine with timecode), start the timecode generator and check the record level. It should read around -7dB to -5dB below the zero mark. If it isn't, then you'll have to take steps to adjust the level. If you feed the timecode through your mixing console to do this, make sure that there is no EQ used and that you disconnect anything concerned with timecode from the mixer before you start recording any music. Timecode has a habit of leaking into channels where it isn't wanted, so it's best not to feed it into the mixer unless you really have to.

Having set the correct record level, set the tape into Record mode and start the timecode generator. You now have just over half an hour to enjoy a cup of studio coffee.

Once the tape is striped, the next thing to do is to hook up the output of track 16 (or 8 or 24) directly to the timecode input of the MIDI sequencer, keeping well away from the mixing console!. On the sequencer, you'll have to set a Start Time. 00:00:45:00 is about right, a few metres away from the start of the tape. Now, with your sequencer set to SMPTE/EBU timecode sync, whichever part of the reel you play, the sequencer will read the timecode value on the tape and convert that to a bar and beat number, according to the tempo you have set. If you play the tape from just before the start point, then as 00:00:45:00 is passed, the sequence will start as if by magic!

Before we record any sounds onto tape, we must do one thing to the sequence: add two empty bars at the beginning. This is because most sequencers will take a little time to get going, and this would throw the rhythm if it happened in the first bar of the music. The second empty bar is so that you can have a good four beat metronome count-in each time you start, as the first bar will be a bit off.


Assuming that the drum sounds you used in the sequence are the ones that you want to use on tape, these will be the first items to be recorded. But in all probability, they will all be on the same track of the sequencer, because they were all recorded using the same MIDI channel. So they will have to be separated out. Clever sequencers will do this automatically, but if you have to do it by hand, as I do with my ASQ10, then take care — it is very easy to lose something.

Once you have each drum sound on a separate sequencer track, preferably with a name, then it's a simple matter to solo each track and record that onto the multitrack, one track at a time. I always have bass drum on Track 1, snare on Track 2, hi-hat on Track 3, toms on Track 4, etc. If your sampler or drum machine has separate audio outputs which are convenient to use (I find it more of a slog than it's worth on my Akai S900) then you can allocate each drum to its own output and record them on to the multitrack in one pass. Doing it this way can save a lot of time, but I prefer the one drum at a time approach because I like to add a bit of ambience to each drum. Not a lot, but just enough to make me think that I recorded a real drum in a real room with a real microphone. I may, or may not, add more reverb in the mix later.

"...what we require of our recording system is that it should be as versatile as possible, allowing creativity free rein with very few restrictions. The combination of MIDI and multitrack can do this."

After the drums, I lay down my guide keyboard track onto tape, throw the lever switch in my brain into multitrack mode, and get on with some more creativity.

For subsequent tracks, instead of working purely with the sequencer I have to work with the tape, which means using the tape transport controls to find my way around the music. This is never as convenient as doing it from the sequencer, using bar numbers. As we shall see shortly, things won't always be done this way. But for the moment, I have to locate the tape to a point just before the start of the track. Then when I start it, the sequencer will also start at the correct timecode value and I can record a bass line with a new synth or sampler voice onto a fresh sequencer track. When the bass line is finished, I dump it onto tape and start with the next layer. So throughout the course of recording, a new track is recorded into the sequencer — which is always synchronised to tape — then transferred onto a clear tape track on the multitrack. Eventually, the tape fills up and we have to proceed to phase three of the MIDI/multitrack process...

When the tape has no free tracks left, it is still possible to record extra parts into the sequencer — in fact, as many as you have MIDI instruments and free channels on your mixer. These will be played 'live' by the sequencer during mixdown to stereo. If I'm using a piano sound on my sampler, then I generally like to play this live from the sequencer into the mix because, even with a well looked after tape recorder the real thing — well, the real sample — always sounds just a touch cleaner. Other candidates for live playback include anything with a very trebly sound, which may lose some of its sparkle if recorded on tape, or sounds that need heavy EQ adjustment, which can boost tape noise to an unacceptable level.

The real problem with working like this is that it is so much more difficult to find your way around, with the tape recorder leading and the sequencer following, than it is using the sequencer by itself. This makes it a bit of a nuisance to work out a new part while listening to what you have already recorded on the tape. One way around this is to use a spare synth, set to a fairly neutral sound, to play all the tracks you have so far recorded into the sequencer (or whatever tracks you think necessary), by switching them all to the same MIDI channel on the sequencer and setting the synth to that channel. If you can spare a drum machine as well, then this is pretty good for overdubbing to without the hassle of shuttling tape back and forth.

One day the real answer will be commonly available, which is to have the sequencer control the tape recorder. This is possible now using the Fostex R8 with its accessory MTC1 MIDI timecode processor and Steinberg's Cubase sequencing software. When all multitracks and sequencers can work in this manner, we will indeed be able to consider ourselves fortunate.

An alternative to working on a 'sketch' version of the music to map out the general song structure is possible when your MIDI system has reached a level of sophistication where it can produce almost the finished article all by itself, and multitrack is used to add the finishing touches. In this case, a good way of working is to arrange and mix the track, purely in the MIDI domain, into stereo and record it onto two tracks of the multitrack. Extra layers of interest can be added onto the remaining tape tracks and then finally mixed onto stereo tape, or even DAT.

There are at least two advantages in working this way: firstly, it is very easy to keep your eye — or rather your ear — on how the track is going to sound when it is finished, without taking the music apart and putting it on tape strand by strand in the conventional multitrack way. The other advantage is that it's cheap! An 8-track recorder is quite sufficient, which can cost several thousand pounds less than the 16-track Tascam MSR16 or Fostex G16 that would otherwise be the minimum necessary for top level work using the one track per instrument method.


Yes it is, to combine the flexibility of MIDI with the raw power of multitrack tape. When you consider that a machine like the Fostex R8 (which may be small but sounds damn good) is available at a price not much higher than that of one good synth, and it expands the capabilities of a MIDI system enormously and lets you record guitars and vocals, buying a multitrack tape recorder is a step that just has to be taken sooner or later.

Looking upmarket at a 16-track recorder expands the horizons still further, and if you already have a good sampler and a couple of synths, then it may be time to start asking whether progressing further in that direction is going to invoke the law of diminishing returns. With a MIDI/multitrack system, you will not only get the best from your existing equipment, you will also get the best from yourself and you'll find that, more than any system which relies solely on the 5-pin DIN, the possibilities are indeed limitless.


If you take suitable steps, then sync problems shouldn't cause too much trouble. The very worst thing that can happen is that for some horrendous reason you lose the sequence entirely. There are a million ways this can happen and it stands to reason (and to Murphy's Law) that one of them will get you in the end. For example, try the following with your computer, sampler or sequencer, or anything that stores data on a floppy disk:

Starting off with two blank disks, create a few dummy files and save them onto one of the disks, keeping a count of how many files there are. Insert the other disk, create another file, and initiate the process of saving up to the point where you have only to press one button to activate it. By this point, the computer will have looked at the disk and found that it is blank. Now swap the blank disk for the one containing the test files. If your computer saves the file onto this disk without losing, destroying, or damaging any of the existing files, then it passes the test. If some of the files do disappear, then be warned — this is a bug that will get you sooner or later.

Of course, the answer is to always keep a backup. But actually, if the sequencer tracks you have dumped to multitrack are all OK, then you don't really need that data on disk any more [unless you wish to redo the song at a later date - Ed], All you need is the timecode start point and the tempo, plus any tempo changes, and you can record new tracks without difficulty. After some disagreeable experiences with more than one sequencer, I resolved always to write down the start point, tempo, and tempo changes, which takes about 10 seconds. I haven't lost a night's sleep since. If you use an external sync box, rather than a sequencer with an integrated SMPTE/EBU capability, you'll have to keep a written record of such things anyway.

The other problem that you will typically have to face is loss of timecode. Unfortunately, the timecode readers built into sequencers are sometimes not the best. If they lose code for an instant, they will probably grind to a halt. If the code on Track 16 becomes faulty, then the answer (at least one answer which doesn't require extra equipment) is to record fresh timecode onto another track. In doing so the start time will change, so you'll have to determine this by trial and error, but it is surprisingly easy to do. End of problem.

Of course, the number of things that could potentially go wrong is limitless, but the two mentioned above are the worst. Here are a few more minor difficulties and possible solutions:

The sequencer will not sync to the tape.

The recording level of the timecode may be too high or too low. Set it to around -7dB to -5dB, or to whatever level the sequencer manufacturer suggests. Timecode should not be recorded with dbx noise reduction on, nor with equalisation. And make sure that the tape heads are clean!

The playback level of the timecode coming into the sequencer may be incorrect. Experiment with the level until you find one that works reliably.

Sync is intermittent.

Any bass instrument recorded on Track 15 will leak into Track 16, affecting the timecode. From time to time, or indeed perhaps all the time, this will make the code impossible to read. Erasing the track and recording at a lower level, or on another track, will usually cure this.

The sequencer will not sync during recording.

Once again this is probably due to crosstalk/leakage from Track 15 or even Track 14. During recording, crosstalk in a combined record/playback head is very high and may leak into the timecode. Use a lower recording level.

The sequence plays back at the wrong speed.

Some sequencers will automatically recognise the frame rate of the timecode and play at the correct speed. Others have to be manually set to the correct rate otherwise the tempo will be incorrect.

Series - "Recording Techniques"

Read the next part in this series:

All parts in this series:

Part 1 | Part 2 | Part 3 | Part 4 | Part 5 | Part 6 | Part 7 | Part 8 | Part 9 | Part 10 (Viewing) | Part 11 | Part 12 | Part 13 | Part 14 | Part 15 | Part 16 | Part 17 | Part 18

More from these topics

Browse by Topic:




Previous Article in this issue

Masterbits Vocal Sample CD

Next article in this issue

Fostex G16

Sound On Sound - Copyright: SOS Publications Ltd.
The contents of this magazine are re-published here with the kind permission of SOS Publications Ltd.


Sound On Sound - Sep 1990

Donated by: Bert Jansch / Adam Jansch






Recording Techniques

Part 1 | Part 2 | Part 3 | Part 4 | Part 5 | Part 6 | Part 7 | Part 8 | Part 9 | Part 10 (Viewing) | Part 11 | Part 12 | Part 13 | Part 14 | Part 15 | Part 16 | Part 17 | Part 18

Feature by David Mellor

Previous article in this issue:

> Masterbits Vocal Sample CD

Next article in this issue:

> Fostex G16

Help Support The Things You Love

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!

Donations for December 2021
Issues donated this month: 0

New issues that have been donated or scanned for us this month.

Funds donated this month: £4.00

All donations and support are gratefully appreciated - thank you.

Please Contribute to mu:zines by supplying magazines, scanning or donating funds. Thanks!

Monetary donations go towards site running costs, and the occasional coffee for me if there's anything left over!

Small Print

Terms of usePrivacy