From the humblest of 4-track cassette recorders to the might of 48-track digital, multitrack is the technique used for 99.9% of all commercial recordings. David Mellor looks at the background to this powerful tool and wonders how we ever managed without it.
In days of old when knights were bold and multitrack not invented... basically, you had to get it right all in one take or go back and try again. "Take seventy-five, and go easy on the vibraslap this time," cries the producer.
The old technique, used in the Fifties because that's all there was, is often referred to as 'straight stereo', because the sound goes into the microphones and straight onto the stereo tape. Let's just forget about straight mono for the moment. Does anyone out there still record mono? I thought not. Although it seems just a little primitive now to we sophisticated recording engineers, both in the home and in the pro studio, remember that an awful lot of classic tracks were done that way. I'm currently working my way through the Atlantic Rhythm & Blues series of CDs which have many recordings going way back, mostly recorded 'live' with not an overdub in sight... What's an overdub? We'll soon see.
Let's look in more detail at the problems of recording a band all in one go. For a start, microphones have a habit of intimidating even the most professional of performers. When you know that every tiny mistake is going down onto tape, those tiny mistakes somehow seem to become rather less tiny and more frequent. Add that together with any technical problems in the recording process, extraneous noises etc, and it's a recipe for disaster. The only way to do it is to do it, and do it, and do it again until it's right.
There is one story which harks back to the days of 78 rpm records, where the recording was cut straight into a wax disc. Apparently, a certain band was having trouble with a brass player who kept fluffing a note. Over and over they tried it, and over and over he goofed. When at last the piece was performed perfectly, after the last chord had died away, there came a voice from the back, "Was that all right?"
It wasn't, the engineer hadn't lifted the cutting stylus from the disc. The music was perfect, and so was the musician's question - with no way of erasing it! Back to the top for another take, boys...
One step beyond straight stereo is sound on sound. Now where have I heard that expression before? You can read more about this technique in last month's 'Studio Magic' series. A common use of sound on sound, in the early days of recording, was to record a complete track, band and vocal, in stereo - or mono - and then mix that recording with a harmony vocal performed by the featured artist onto another tape recorder. I once saw a film which credited the invention of this to Buddy Holly. I'll suspend judgement and accept it as true because it's the right era and because I like his records!
Once people started doing this, it must have been a small step to think of having separate tracks on one tape, to take the various instruments. These tracks could be mixed down onto a stereo tape recorder later. Les Paul (the guy who the guitars are named after) was an early exponent of multitrack recording, and used an 8-track recorder to build up some terrific virtuoso song arrangements using little else but guitar and percussion to accompany his wife, Mary Ford. If you ever see a Les Paul record in a secondhand shop then buy it, if you don't learn anything about guitar playing then you'll certainly learn what can be achieved even with a primitive '50s multitrack tape machine.
To those who practice the art, multitrack must seem pretty obvious in its benefits, but there are some finer points worthy of note which do not seem to get enough attention in the technical press. Let's start at ground level to make sure we are talking the same language.
SUIVEZ LA PISTE
Funnily enough, the French word for ski-run is the same as the French word for tape track. Figure 1 shows a length of tape, recorded by a 4-track tape recorder. Figure 2 shows the arrangement of tape heads that recorded it. Most reputable tape recorders have three heads as shown. One notable exception has just two, but we shall ignore it for the moment. The function of the erase head is, of course, to clean the tape for a fresh recording and, yes, the record head is to record and the replay head is to play back. How can we use this to make a multitrack recording? The answer is we can't. There needs to be something more. The record head must be able to play back too.
On the face of it, it sounds like a contradiction in terms. On close inspection - microscopes out everybody - we find that the record head and replay head are pretty much alike. Figure 3 shows the general outline in exaggerated scale. The difference between the two is that the gap in the record head is wider than the gap in the replay head. The recording process occurs at the trailing edge of the gap, in the direction of tape travel, so the closer it is to the leading edge, the more the magnetism on the leading edge would interfere with the process.
The replay head needs a small gap because the magnetism imprinted onto the tape needs to form a magnetic circuit with the poles of the head. Since a frequency of 20kHz, on tape running at 15 inches per second (ips), has a wavelength of only 19 millionths of a metre then the head gap must be significantly smaller than this. I could explain in more detail - there is a lot of potential detail to go into - but the important thing for a studio engineer to realise is that there is a necessary difference between record and replay heads; small but not negligible.
What I am leading up to is that the record head can, if required, be used as a replay head. It will not perform to the same standard, but this fact has an important spin-off. Take as an example, our 4-track tape recorder. Suppose a stereo signal is recorded onto tracks 1 and 2. If tracks 1 and 2 of the record head are then used to play back this recording, a third track can be recorded while listening to the first two. As you can see, part of the record head is recording while another part is playing back. I think nature is a jolly useful thing to allow this to be possible. Imagine what would happen if the replay head had to be used to monitor the first two tracks. If the record and replay heads were physically spaced one inch apart, then the third recorded track would be spaced one inch away from the first two on the tape. Not just an inch in length but 66 milliseconds in time. Although you might think you were recording the third track 'in sync', ie. synchronised with the first two tracks, when replayed it would be wildly out of time. 66 milliseconds can be a lot in musical terms.
Although I haven't stated it so far, the name for the playback output of the record head is the sync output. Hope you can see why. A professional multitrack tape recorder will have, for each track, an Input, a sync output and a replay output. The sync output is used for laying extra tracks in time with the first. The replay output is used for mixdown when maximum sound quality is wanted.
As you probably know, multitrack starts with 4-track and goes all the way up to 64-track in pro studios. What benefit do the extra 60 tracks bring? After all, they keep on saying that the Beatles' Sergeant Pepper was recorded on 4-track.
I'm sure that Sergeant Pepper is still earning a crust for its creators, but times have changed. I'm not saying that 64 tracks are necessary to make music, just that the currently popular musical idiom is highly multitrack based. See what I mean? Up to a point, the more tracks you have available, the greater the chances of success, given a sufficient input of musical ability. 64 tracks may be overkill, but so many top producers seem to need 48 that we all have to take notice of what is going on in these higher echelons.
To realise why a large number of tracks is beneficial, let's backtrack a little. In the days of 4- and 8-track recording, when these were pro studio standards, it was common to record several instruments on one or two tracks. Bass guitar and drums on one track was not unheard of. (You wouldn't dream of doing it now!) The extra tracks were used for adding extra gloss to the production. With 16-track recording the situation became slightly different. Now there were enough tracks to consider recording each instrument on a separate track. It can be done with 8-track, but you are never going to build up a really 'big' production this way.
Having the instruments on separate tracks makes it possible to obtain an ideal sound for each, using equalisation on the mixing desk. Of course, once you get to sweet 16, you yearn to be grown-up 24 and so on. Above 24 tracks, another consideration comes into play: there is room on the tape to experiment. Let's face it, if you can't get a big sound on 24-track then you would be better off working a pneumatic drill. With 32 or 48 tracks available there is room to have an idea, put it to one side, then bring it back into play if you feel like it. Alternative takes is the name of the game. Let the guitarist have ten attempts at his solo, we can decide which one we like later - either that or combine the best fragments from each one.
INTO THE GROOVE
Looking in more detail at the recording process, I'd like to explain a little about how bands are generally recorded, as it throws light on the whys and the ways in which multitrack is used. For the purposes of this example, 16 tracks should be sufficient. (8-track makes things a little tight, but not impossible.) The first stage in the recording process is the basic track. This is where the guys and girls in the group who play the most structurally important elements of the arrangement record together in the studio, their performance going down onto tape in one go. Imagine the microphone plan is as follows:
That's enough for starters, and will take up eight tracks on the tape. I have simplified the drum miking a little, but it's not impossible to get a good sound with this technique providing the studio acoustics are okay. Typically, the musicians will all wear headphones so that they can hear each other clearly. Not many bands can get from start to finish of a song without hearing what the singer is meant to be doing, so he gets a microphone as well - and since he has a mic, you might as well put his track down on tape too, it may come in useful later on even if it is a bit rough in places.
It's amazing how long it can take to get this bit right. If one musician makes a mistake, then it is possible to correct his part individually, but it is usually better to get the all-important backing track done in one go. If the worst comes to the worst and the players just can't get it right (if you're thinking of starting a small studio, be warned that many amateur bands are like this!), then one possible answer is to edit the multitrack tape. Most people are fairly familiar with the concept of editing stereo ¼" tape, but are reluctant to cut the multitrack. Do it now! It will give you such a feeling of power, especially if you use 2" tape. I can't say that it always works musically, because tempi can vary so much, but technically it's hardly any different to joining ¼" tape - you just use a bigger splicing block and wider tape. Subsequent recording over the join should not cause any problems that can't be covered up with a bit of reverb, and usually no problems at all.
When the basic tracks are captured with sufficient feel or groove - as trendy recording engineers say - it's time to move on to the overdubs and vocals. Overdubs can take a lot of time, but at least it's a straightforward matter to go back and correct mistakes. This is the technique known as 'drop-in', or to the more Americanised of us, 'punch-in'.
When the guitarist (as they always do) gets his finger trapped under the string in the middle of an otherwise perfect solo, it seems a pity to go back and re-do the whole track. Why not just patch up the bad part? It's time to get back to the technicalities of multitrack.
I said earlier that it is possible to monitor the sync output of the record head so that you can hear previously recorded tracks when you overdub. Let me describe what happens when you play one track in sync mode, with record ready switched on for that track, then press the Record button. As you play, the sound of what is already on the tape is heard through the speakers. When the Record function is selected, the tape recorder switches an internal relay so that it now directs what is coming in straight to the sync output. This means that when a musician is doing a drop-in, he can hear his previous performance in the headphones up to the point where Record is pressed, then he can hear the sound of what he is currently playing. This is a very handy facility because the player always knows where he is in the track, and what he is supposed to be doing.
That explains the basic automatic switching operation. Now let me tell you how this can differ. Firstly, older multitrack recorders, such as the Studer A80 Mk 1, do not have this automatic switching facility. It's not an insuperable problem, but a point to be borne in mind when you shop on the secondhand market. Secondly, some multitrack machines, such as the Fostex range, have extra switching facilities which let you monitor the input signal even when you are not recording. It's not always a case of the more facilities the better, just that if you are into adventurous equipment purchase procedures, ie. secondhand, then make sure that your mixing desk can cope with your multitrack.
I could at this point mention a certain other multitrack which, in some versions, does not allow sync monitoring when the Record Ready button is pressed - only input-monitoring. This means that you have to use the Record Ready button on the machine to perform drop-ins, rather than the Record button on the remote control unit. Also, it means that after recording a track, you have to go back to the machine and take it out of Record Ready mode to listen to what you have just laid down - you can't just press Play! How silly, and if anyone else has the same ½" 8-track that I have, then accept my commiserations. I hope you can figure out how to modify it (it can be done!).
Stereo tape recorders have a lot of little nasties which try to make your recording worse than it should be. Multitrack machines have all these and more...
Of prime concern, especially with narrow-gauge multitrack, is 'head crosstalk'. This becomes troublesome when you try to play one track and bounce it onto the adjacent track (as you might do if you were trying to create some empty tracks on the tape). In this situation, one layer of the record head is trying to play back, the next is trying to record. That high pitched whistling sound in your headphones is feedback. Crosstalk between adjacent tracks of the tape head forms a loop similar to the loop between microphone and loudspeaker in a PA system. It's sometimes possible to get around this by lowering the recording level, but the best way to avoid it is to plan more carefully, so that you never have to bounce adjacent tracks. On an 8-track recorder this may prove impossible.
The second problem is 'sync response'. When I said earlier that the record head could be used for playback, I indicated that the sound quality wasn't quite as good as from the replay head proper. Since the head gap is wider, high frequencies are curtailed. This isn't a problem for monitoring purposes but if you want to bounce tracks around on the multitrack tape - you have to use the sync output from the record head to do this - then some quality is lost. A typical situation is where you have several tracks, perhaps percussion, which could be mixed into stereo onto just two tracks of the multitrack tape, leaving tracks free for further work. In doing this the brightness of the metallic sounds would be lost to a certain extent. Having said that, head technology improves year by year and with the latest multitrack recorders, the sync response is hardly distinguishable from replay. That has not always been the case.
'Gap scatter' is an interesting phenomenon which explains how difficult it is to make a multitrack tape head. If you read 'How It Works - the Tape Recorder' [SOS Feb'87] then you will understand that the azimuth of a tape head - ie. the vertical alignment - has to be correct to preserve a good top end frequency response. Unfortunately, each track of the multitrack head tends to have its own slight variation from true. It will not be a problem in most studio work, but it's useful to be aware of what can go wrong, especially when tapes are transferred from studio to studio.
Multitrack is not an easy subject to put into a convenient nutshell. It's a situation where you have to have hands-on experience to know the delights and disasters that can occur. Multitrack recorders are becoming cheaper and cheaper and I am sure that soon they will be as commonplace as the home video - almost! Perhaps the current formats will be outdated by newer developments. Interesting possibilities include the Akai 12-track cassette-based (not compact cassette) format, and direct to computer-type hard disk recorders. Akai have also recently announced a Video 8 based digital multitrack. Perhaps R-DAT might have something to offer too. One last alternative is that there may be a complete reversal in trends and we shall all return to recording music as it is performed, straight onto stereo tape, with all its pitfalls and perils. Do you think that might happen? I don't.
To my knowledge, there are no 48-track tape recorders in current production, although Sony promise that their digital multitrack is upward-compatible to this standard. To carry out 48-track recording, two 24-track tape machines are required, together with a synchroniser to lock them together. 48-track is actually not quite a precise term because timecode has to be recorded on one track of each tape for the synchroniser to know where each should be, therefore only 46 audio tracks are left.
48-track procedure normally involves the concept of 'master' and 'slave' reels. When the basic tracks of a song are built up, they go onto one reel of 24-track tape. When this reel is full, the whole of the music so far is mixed down onto two tracks of a second reel on another 24-track machine - the slave reel. The original (master) reel is put away and all further overdubs are done using the temporary stereo mix of the basic tracks as a guide. This saves having to wait for the synchroniser to lock up and also saves the valuable master reel from shuttling back and forth against the heads, wearing down the high frequencies.
When the recording is complete, the master reel is brought back out and master and slave reels are synchronised together on two 24-track machines ready for final mixdown. There is no reason - apart from the number of mixer channels available - why this cannot be extended to 72-track and beyond.
Older engineers - not me! - call this process 'remix' because they thought of multitrack as 'taking apart' the performance before putting it back together again. Most music nowadays is put together for the first time on multitrack, so the mixdown is the first time it all comes together as it should.
It goes almost without saying that mixdown involves mixing the 16 or 24 tracks of the multitrack onto a ¼' stereo tape - or perhaps mixing a portastudio 4-track cassette onto an ordinary stereo cassette. Any multitrack recording, however, will be noisier than a stereo recording of the same combination of instruments. This is because noise adds by 3dB (decibels) for each doubling of the number of tracks: ie. 8-track is twice as noisy as 4-track; 16-track is twice as noisy as 8-track. If each track could be recorded to its maximum level all the time, then this would not occur as signal and noise would add in the same proportion. Unfortunately, this is impossible in practice.
The best way to avoid multitrack-generated noise is to mute any track that has no signal on it. For example, if the vocalist is taking a break, then his track should be switched off for that length of time in the mix, thus cutting out the noise contribution of the track. When automated muting is a feature of all mixing consoles - as it surely will be, in time - then this will be dead easy to implement. Until then we'll do it with difficulty because it makes our recordings cleaner and quieter.
THE 6dB PROBLEM
If you use multitrack, then this paragraph on the subject of drop-ins may be useful.
Have you noticed that when you drop in there is an increase in level? It depends on your monitoring arrangements, but the traditional - and most useful - method is this: Before the drop-in point, the previously recorded track and the signal to be recorded are monitored. This is so that an instrumentalist can play along with the track, so that the music will flow smoothly over the drop-in. After the drop-in point, obviously, only the signal being recorded will be heard.
If your multitrack recorder has automatic switching from sync to input when you press Record, then the level will increase at the drop-in point, which can be annoying. This is because when the musician is playing along with the previously recorded track, the two signals are uncorrelated - even though the same notes are being played. After the drop-in the signal being returned from the tape recorder is exactly the same as the input signal, so the two are correlated. Uncorrelated signals of equal level will give a 3dB rise when added together. Correlated signals will give a 6dB rise, however, so the level goes up.
Some clever mixing consoles can compensate for this. Most do not. It's annoying, but we have to live with it.