In the last few months I've covered various aspects of basic tape technique. I've talked a little about recording concrete sounds, about ways of manipulating sounds on tape, and about editing. So far we've generally thought about one sound at a time and how each of the processes discussed can be applied to a particular sound. But, if you want to get beyond simple experiments and start making pieces, sooner or later you will want to superimpose one sound (or more), or a group of sounds, on top of another. Perhaps you want two lines (melodies if you like) running in counterpoint; perhaps you want to play one sound progression over the rhythmic background of a loop; or maybe you want to superimpose several sounds to make one composite sound or a chord.

In order to do this it is necessary to have a way of playing from two tape sources at once and mixing the two sources into one composite signal to be recorded. Having more than two sources is an advantage but it is possible to work with only two. If you go into a "classical" tape studio you will notice that there are usually a large number of tape recorders. For example, in the studio that I run, fairly typical, but by no means the largest, there are (at present): one 8-track machine, two 4-track machines, three 2-track machines and two full-track mono machines.

This may sound like a lot, but I've often seen them all in use at once. While a set-up like this makes many things a lot easier, it is possible to achieve a great deal with just two machines; and it is even possible to work with one reel-to-reel and one cassette machine if you use a bit of ingenuity. If you have access to two reel-to-reel machines you will be able to save yourself some copying stages, but the principle is the same, so, for the purposes of this article, I'll assume the second machine is a cassette recorder. A few years ago I would have thrown up my hands in horror at the very idea, but the quality of cassette recording has improved so much recently, and the prices, compared with domestic reel-to-reel machines, are so low that a cassette recorder (particularly a portable one that can be used for location recording) can be very worthwhile in a low-cost set-up. But do get hold of one with noise reduction.

The other thing you will need is a way of mixing two signals into one. Some tape recorders are designed so that, when only one channel is selected for recording, both inputs are mixed to the selected channel. If this facility is available on your machine then you'll be all right. Otherwise you can build a small mixer very cheaply (the one in the May issue of E&MM should do the job nicely), and you'll need some sort of mixer sooner or later anyway.

The basic procedure is as follows: Say you have eight sounds (individual sounds or edited tracks) each of which is on a separate tape. I will call these "layers" 1 to 8. I will call the master recording machine A and the playback machine (or cassette recorder) B. Take layer 1 and put it on machine B, (if machine B is a cassette machine you must put the original tape on A and make a copy of it on one track of the cassette - if B is reel-to-reel you have only to put your original tape on B and record direct).

Now put some clean tape on machine A and record layer 1 on track 1. In the same way put layer 2 on machine B and record on track 2 of the same tape on machine A. You should now have a new tape, which I will call copy 1 with layer 1 on track 1 and layer 2 and track 2. Now repeat this process putting layers 3 and 4 onto tracks 1 and 2 of a new section of clean tape (probably further on the same reel) which I will call copy 2. Note that, at this stage, no actual mixing has yet occurred; all we have done is arrange two separate layers onto one tape so that they can be played simultaneously on one machine.

There are two important points that must be made at this stage. First, always make copies at the highest level possible without distortion, even if the sound will end up very quiet. In this way noise levels can be kept to a minimum. Remember that noise is cumulative - it will always be increased at each stage of the process, and once there it cannot be reduced. So the trick is to keep it to a minimum at every stage. If a sound is going to end up quiet try to reduce its level at the latest possible stage, as the accumulated noise will then be reduced with it. If you have to increase a low level later, the noise will be increased too, and you may have the accumulated noise of many layers to worry about.

Secondly, it is often very tricky to get good synchronisation between layers using this method. Sometimes this may not matter, but often it is essential. If you mark the start of the sound on each (reel-to-reel) tape very carefully and make a start mark on the new tape you are recording on, then always line up these marks in exactly the same place you should, with practise, be able to start both machines consistently together. If it doesn't line up exactly the first time, slightly shift one of the tapes forward or backward, as appropriate, and try again. If you are careful and patient you should be able to get it right. Remember, be sure to get the first ones really accurate or you'll have more and more trouble later on.

Now for the second stage, which is almost a repeat of the first. Put copy 1 onto machine B and mix both tracks of machine B into track 1 of machine A. At this stage it will be necessary to get the relative levels of layers 1 and 2 correct, but still keep the overall level as high as possible. Now repeat the process, mixing layers 3 and 4 onto track 2 of the same section of tape. You should now have a tape (I'll call it Mix 1) with layers 1 and 2 mixed onto track 1 and layers 3 and 4 mixed onto track 2 - note that we still have control over the relative levels of these two pairs of layers. Now repeat the whole process so far with layers 5 to 8 replacing layers 1 to 4 to make a new tape (Mix 2) with layers 5 and 6 on track 1 and layers 7 and 8 on track 2 (the intermediate stages will be copy 3 and copy 4). Now repeat from the second stage using Mix 1 and Mix 2 in place of Copy 1 and Copy 2, so that now you have all eight layers on one tape (Mix 1) divided into two groups of four on tracks 1 and 2. If you want to keep it as a stereo mix, with different material on left and right tracks, then that is your finished product; otherwise you can mix the two tracks as necessary onto another tape. If you have more than eight layers the whole process can be repeated using the other layers so that you end up with two tapes equivalent to Mix 1 which can then be combined in exactly the same way as Mix 1 and Mix 2 were; and this can go on indefinitely, with the proviso that noise will increase relative to the number of stages involved. I've successfully done "montages" with as many as sixty layers this way - it can be done if you're careful.

Essentially the process consists of one procedure that can be repeated again and again like a sub-routine in a computer program. I have given a diagram of the basic procedure in Figure 1. Just select the appropriate inputs each time and note that in the first stage, combining two "layers" to make a "copy", the instruction "mix onto track n of machine A" is only a copy as there is only one signal to mix each time.

You will have noticed that I have been very careful to give each stage a clear identifying label. As I've said before it is very important to be meticulous about all this. Always label the tapes and keep notes on what each one contains, otherwise confusion is inevitable. The use of different coloured leader may also help to distinguish the different stages. The best thing is to evolve a consistent system which you always use.

Finally, a few hints about specific points. First, try to work out the best order of combining sounds before you start. For example, if you want to end up with a background texture plus a couple of more prominent "voices", do the background as one part of your mix and get that right on its own, then combine your main voices in the correct balance, and, finally, put the two parts together, when all you need to do is get the right relationship between background and foreground. You'll find this sort of thing, both the planning and successful execution, a lot easier when you've had some experience. Don't get discouraged if it all comes out wrong the first time - it can be done if you persevere. Another point, try to work with the shortest possible sections and edit them together after they're mixed. It will be much easier to get it right if you do a short section each time, and there won't be so much work wasted if it all goes wrong. Exactly how to divide it up will depend on what you're doing, but some specific examples will be discussed over the next few months, when I plan to go through the whole process of making a short piece of "Musique concrete" from recording the original material through to the finished master. A lot of the things I've talked about so far will be included.

I'll leave you with a little problem which will crop up in the piece. A number of layers start at different times but all finish together with an exactly synchronised chord - what do you think is the best way to do it?