David Mellor takes a look at the components of a typical disk recording system, and passes on some hints on what to look for when deciding on a setup. This is the last article in a two‑part series.
In music and recording technology there is always a new development to keep things exciting. Some readers may remember when the solid‑body electric guitar was a new and exciting development; others will recall the Moog synthesizer as being the pinnacle of technological achievement. And then came polyphonic synthesizers, MIDI, sampling, sequencing and digital recording — all really tremendous developments when they first appeared and still vitally important to us now. So it shouldn't come as any surprise if I say I believe that the next big thing is going to be hard disk recording. Since hard disk recording is a bit of a mouthful I shall refer to it simply as disk recording from now on. This ought to become the popular expression, because it can encompass optical disks as well, something rather different to a hard disk, as I shall explain.
Tape has been great to us over the last 50 or so years as a storage medium, but its days are numbered — even if it isn't quite finished yet [see last month's instalment for some of the reasons why tape is still a valuable recording medium]. Tape has one intrinsic drawback that it will never overcome: it is a one‑dimensional medium. However many tracks you record across the width of the tape, the recording can only progress in a straight line from the start of the tape to the end. If you want to get from one point in your recording to any other, you have to fast wind through the tape. This takes time, and even at DAT's one‑minute‑per‑second fast wind speed there is still enough time to read a tabloid newspaper while it winds from one end of a 120‑minute cassette to the other. If you know your Star Trek, you'll understand how the Enterprise enters hyperspace to get from one point in the galaxy to another faster than the speed of light. It simply (!) moves from three‑dimensional space into another dimension where it can take a short cut to get from A to X, Y or Z. This is how a disk recorder 'fast winds': to get from one point on a disk recording to any other point, the pickup uses the third dimension of normal space to cut quickly across the spiral or circular pattern of the recording, traversing what might be two hours of recording in milliseconds. This is great for manual locate commands but there is another even more important advantage: to edit a tape recording, the tape must be cut and reassembled, or copied onto another tape, which takes time and, in the analogue domain, loses quality. To edit a disk recording, it is only necessary to give an instruction to the system to play from point A to point B, then from point Y to point Z, missing out points C to X in between. Even if this entails missing out many minutes of the original recording, it can be performed fast enough to make a completely seamless join.
Currently, editing of tape is usually regarded as something that occurs after the recording has been made, and is often thought of as a menial chore that merely cuts out unwanted material, or assembles the material in a different order. So far we are tending to fall into the trap of looking at disk recording systems in the same way — as glorified editing tools to be employed after the original recording has been finished. But in the future, as more and more people become aware of their potential, the way we record could change totally. We won't look on a disk recorder as a clever device for copying one good performance of the chorus of a song into the place of several bad ones. We will have the opportunity of assembling a recording from fragments as tiny as we wish — any note from any instrument at any time. This will change music itself, and I wouldn't be surprised to see a whole new type of music emerge as disk recording progresses.
Strangely enough, a disk recorder isn't something you buy in one box like a DAT or a tape recorder. It will be, but it isn't yet. Often you have to assemble the components yourself, or perhaps the system supplier will do the assembling for you. Either way, being the new owner of a disk recording system can be an educational experience.
Basically, a disk recording system breaks down into three parts:
- The disk itself, which is the storage device; think of it like the tape on your tape recorder, but one which is never removed and always re‑used.
- The heart of the system where all the processing happens, which could be housed in an unassuming black box, or perhaps on a computer card. This will appear to you like the engine in your car — something you can happily ignore until you hear a funny noise that you haven't heard before (I mean this literally!) or you don't hear anything at all.
- The third component, the system's interface. This may be a standard personal computer — Atari, Mac or PC — or it can be an interface which is purpose‑built by the manufacturer. These tend to be found in the more expensive systems but they have the advantage that they are function‑specific and have all the right controls in all the right places.
Another category, which to me seems very promising, is also developing. This type of disk recording system uses a personal computer for the display part of the interface, but the operator uses a dedicated controller instead of the normal computer keyboard. This could prove to be the best of all possible worlds in the future.
If you buy a very high‑cost system, it will be supplied with a disk drive, or drives, which have been specially tweaked by the manufacturer. Hard disk and optical disk drives are made in their millions by the computer industry for purposes other than audio, and the continuous flow of data that characterises a digital audio signal is not seen as anything of importance by the majority of drive manufacturers. For almost all other computer applications, the precise timing of when a particular byte of data appears matters hardly at all, as long as the average flow of data is fast enough. This means that there can be great holes in the data stream, for reasons I have outlined in a separate panel. If you have a high‑cost system where the disk has been modified, you should have no problems. But the type of system a typical Sound On Sound reader — or writer — will be interested in buying will probably not fall into this category (the tens of thousands of pounds bracket!). We have to be content with standard computer industry drives, so bear in mind that this is a potential problem area.
A hard drive consists of a number of platters which are coated in a magnetic material, as in Figure 1. Each platter has two surfaces, with a pickup for each surface to write and retrieve the data. All the pickups are on a common spindle so they can't roam independently over the disk, but they can get to any point pretty quickly. You'll never see these components of your hard disk, because once you disassemble it in anything than a 'clean‑room' atmosphere it will never work again, but it's good to know what's going on in here, because then you'll have a better understanding of how the whole system works and what its limitations are.
The disk drive has a certain amount of its own intelligence, and it can choose where to store data. This is one significant advantage of the hard disk — you don't have to find a blank space to record; just give the file a name and let the disk do the rest.
The most significant parameters relating to a hard drive are its access time and its data transfer rate. The former is the time it takes to find and begin to output a particular piece of data; the latter is the rate at which it can pump the data out. One problem with these parameters is that they can be measured in a number of ways, which produce different results. Access time is also frequently confused with seek time. Seek time is the time the drive takes to position the head over the required sector, while access time can be defined as the time it takes to position the head and transfer one block of data. I would say that it is not a good idea to compare drives by specification — at least, not until there is some agreement between drive manufacturers and disk recording system manufacturers on how these measurements should be taken. And don't forget that the drive manufacturers are not really interested in the requirements of the audio industry, so reliable specifications that we can place our confidence in are still some way off.
There are three ways of selecting a drive.
1: Ask the manufacturer of the disk recording system, who will give you a model and a firmware number of a suitable drive. The firmware is the drive's internal software, which must also be compatible with the disk recorder you intend to use. This will be modified from time to time and it may be impossible to obtain a drive with the version you need.
2: Buy the system as a package from a dealer and complain bitterly if the drive doesn't work as well as you were told it would.
3: Buy a drive on a sale or return agreement, like I did. Make sure that the dealer you buy from understands that it is to be your decision whether the disk is suitable or not.
Other factors that will influence your choice of drive include whether the drive should be SCSI or IDE. SCSI stands for Small Computer Systems Interface and IDE stands for Integrated Drive Electronics. The disk recording system you choose will govern which type you need. When you have all these details sorted out, you'll need to work out what size the disk should be. The standard rule of thumb is that stereo audio running at 44.1kHz requires 10Mb of disk space per minute. Obviously a bigger disk will be more expensive, but not as expensive as buying a disk that turns out to be too small, meaning that you have to upgrade.
When disk recorders have grown up, they should make life easier in the studio. As they are at the moment, they have a tendency to make things possible that were impossible before, but they don't make normal recording operations any simpler
One footnote about disk drives before I move on: remember that it is the make, model and firmware number of the disk drive that is important. Many suppliers buy bare drives (as they are known) from different manufacturers, put them in cases and stick their own badge on the front. Unless the supplier claims to be a supplier of drives to the audio industry, this badge means absolutely nothing.
When disk recorders have grown up, they should make life easier in the studio. As they are at the moment, they have a tendency to make things possible that were impossible before, but they don't make normal recording operations any simpler — in fact, they can introduce new difficulties that never existed before!
In the early days of disk recorders at the high‑end professional level, there was a good deal of discussion between manufacturers and users about how they should be operated. Since the terminology varied from one manufacturer to another, things became a little confusing. But now there is a reasonable consensus, and most systems have a lot in common, internally and operationally.
Recording audio onto a disk is a lot like recording onto tape or cassette. Figure 2 shows the record window of E‑Magic's Logic Audio system, which integrates disk recording into a MIDI sequencing environment. The meters look, on the screen of the computer, very much like the meters of any other recorder. The only difference you will find with disk recording systems is that there is a tendency for the meters to be very slow to respond. Any decent system will have clip lights which show that the maximum level was exceeded at some time during the recording. In Logic Audio the peak indicator always shows peaks accurately, even if it is sometimes a little behind the audio.
Audio is recorded onto disk as digital data, which you can edit directly with some systems. This type of editing changes the data on the disk, so it is sometimes referred to as 'destructive' editing — make a mistake here and you may be in trouble. Destructive editing is useful when you're certain that you want to make a change of some sort, and that you will always want the audio to retain the change you have made. Once your audio is on the disk, and topped and tailed to taste, you'll probably want to enter a non‑destructive mode of editing. This is the disk recorder's natural mode, where you simply create a list of instructions describing which parts of the data should be replayed at which instant. Some systems use a simple list structure (often called a 'playlist'); others — like Logic Audio — use a graphic display of audio segments or regions.
Figure 3 is the region editing display in Logic Audio, which shows the non‑destructive topping and tailing of a snippet of audio on the disk. The start and end points of the desired sound are shown by the S and E markers. Many systems also allow you to define a 'sync point' within a segment or region other than the start point.
Other things to notice in Figure 3 include the 'overview' display, which shows the audio available on the disk. This is handled in different ways on different systems, but the object is to see the relationship between the segment you are working on and other audio data on the disk. Once you're used to working in this way, you get to recognise the features of the waveform and you can find your way around lengthy recordings very quickly. Remember that even when the start and end points of a region have been defined, there is nothing to prevent you using the same audio data in another region, perhaps with earlier or later start and end points. And any segments or regions you create you can use over and over in your arrangement or playlist.
Some disk recording systems encourage you to create segments at an early stage in the proceedings. Other disk recorders steer you towards recording long segments, and then deciding what chopping and changing needs to be done. If you're coming to disk recording from a background in sampling, you'll probably have a good deal of skill in creating music from short segments; on the other hand, if you come from a tape‑based recording background, you'll tend to work on a longer timescale. Remember that disk recording is all about putting an extra dimension into your music and that you should be moving and copying segments around to get the benefits it can offer.
Figure 4 shows the Tracks screen of Deck, a system which uses Digidesign hardware, like Logic Audio. Here I have created segments from music, sound effects and dialogue takes and used Deck to synchronise them to a movie running simultaneously in a window on the computer's screen. Here we see a number of segments which are arranged on tracks. In Deck, the first four tracks correspond to logical outputs which, with a single Digidesign card installed in your Mac, will be mixed down to a stereo physical output. Subsequent tracks are called 'work tracks' in Deck ('virtual tracks' in other systems) and are there for temporary storage only. You can move tracks up and down on the screen but only the top four will sound. The Soundscape system is similar, but each of the outputs is given a colour and segments are highlighted with these colours to allocate them to outputs. Once again, there is a restriction on the number of segments that can sound at one time, but you don't have to move segments up and down the screen, as with Deck.
Bearing the Tracks window of Deck in mind, I'd like to point out some of the things you will probably want to do with your audio on almost any disk recording system.
- You'll nearly always record more audio onto the disk than you need, and trim it later. To trim it you may have to cut larger segments into smaller ones. Most systems allow you to place a vertical marker, perhaps crossing several tracks, and then cut the segments at that point in time.
- You can never be sure of cutting the audio in exactly the right place, so it's usual to be able to adjust the start and end points, after you have made a cut, to 'reveal' or 'conceal' audio data on the disk. With Deck, if you select a segment, 'handles' will appear, which you can drag to adjust start or end points.
- If a segment doesn't play at the right time, you can simply grab it with the mouse and move it to a different point on the screen. Some systems will allow you to nudge the audio forwards or backwards in time by a pre‑determined amount.
- If you want the audio to play via a different output, you can simply drag the segment up or down the screen. In many instances it will be important that the audio doesn't slip sideways while you do this and lose its original start point. In Deck, you can hold the shift key to ensure that the start point is maintained.
- As well as moving segments, you will want to move and copy segments. To do this with Deck, you hold down the Option key as you move.
- There are many occasions where you will want one segment to start exactly when the previous segment ends. With Deck, you just hold down the command key and this will happen automatically. With at least one other system you have to magnify the display and judge this by eye — tricky!
- Setting the start and end points of regions involves creating a discontinuity in the audio waveform. Unless this is a very smooth discontinuity you will hear a click — sometimes a loud one. You will frequently find yourself wanting to fade segments in and out smoothly, and join one segment to another unobtrusively, so whichever system you choose you should make sure this is easily possible.
It's easy to expect a disk recording system to be able to do everything a tape recorder can do, and more. But disk recorders have problems all their own, and you need to be aware of what can go wrong. One very significant difference between multitrack tape recorders and disk recorders is that if a tape recorder claims to have eight tracks, it will always give you eight tracks — at least while it's still working properly. But the number of segments that can be replayed from a hard disk is not a constant. There is a divergence of opinion among the high‑end manufacturers. Some will allow their systems to reproduce as many simultaneous segments as it can until the limit is reached, when alarm bells will ring. Others advertise their systems as being capable of a certain number of simultaneous segments, knowing that a user will always get that number, except perhaps in extreme circumstances. As a rule of thumb, you can say that a single hard disk is capable of producing eight simultaneous single‑channel segments most of the time. Bear in mind that the rest of the system has to be as fast as the disk to achieve this, and if the audio has to go through the processor of a standard personal computer this may not always be the case. If the audio was originally recorded on the disk in long continuous segments, and you haven't done much editing, then playing back this audio will be easy. But if you have edited the audio heavily into short segments that occur frequently, and rarely in the order of the original recording, then the mechanism of the disk may not be fast enough to maintain the required data flow.
This is the big question, and my answer at this stage and in the more affordable price bracket is no. If I had been able to answer yes to this question, I wouldn't have invested in a Fostex RD8 digital multitrack recorder, which I hope will see me through to the time when a hard or optical disk recorder is a total replacement for tape. The one thing that a tape recorder has, digital or analogue, which disk recording systems do not, is simplicity. I can connect a simple remote control to the Fostex, as I did with my previous analogue multitrack, and produce and engineer a vocal session with my fingers on just five transport keys. Operating the machinery takes approximately 0.1% of my total brain power. I have tried to do the same using a disk recording system, and although it's difficult to put a figure on the amount of brain capacity it uses, there certainly isn't enough left over to get a decent performance out of the singer. Tape recording can be done by touch, and by almost subconsciously hearing the whirring of the machine as it changes between transport modes. Drop‑ins are instant and intuitive. Disk recorders usually only offer programmed drop‑ins — which are great for some purposes, but you can't talk to your artist while you are programming a drop‑in! Look out for systems which claim drop‑in on‑the‑fly capability, but apply a delay between hitting the button and actually starting to record. If this delay varies, you're in big trouble. Continuing the theme of drop‑ins, disk recorders will give you the advantage of being able to keep all your attempts and choose among them later, but most systems haven't reached the stage where this can be said to be easy enough to use in the heat of a session, unless you have a producer to soothe the anxious artist. The other great feature of tape is that if you use DAT or 15ips quarter‑inch analogue tape for your stereo masters, you're compatible with anyone in the world who does sound recording. There are moves towards compatibility between disk recording systems, but it's not commonplace yet.
In conclusion, I would say that the time is ripe to move into disk recording, as an addition to your existing equipment but not as a replacement. Don't hang back waiting for the perfect system to arrive because great benefits are available now, and the more of us who join the party, the more fun and excitement there's going to be!
Most hard disk drives were never intended for audio or video use, but rather for routine data processing where short, random bursts of data are the norm. These drives use gaps in the data stream to carry out calibration and error correction tasks that can take up to half a second to perform. This is obviously a serious drawback for audio and video, where a continuous flow of data must be maintained. The problem can be alleviated by storing data in a buffer memory so that there is always data available when the disk chooses to recalibrate itself to compensate for thermal expansion, as it will without reference to the host computer. Even so, the risk of missing an audio segment or playback stopping is increased, because the disk is not inherently capable of delivering data in the continuous way required.
Fortunately, one disk manufacturer has wised up to audio and produces a range of disks specifically for audio and video requirements. The Micropolis AV range is available in sizes from 1Gb all the way up to a massive 9Gb. These disks have been optimised to produce a high sustained data rate, and although it is still a good idea to consult the manufacturer of the disk recording system, they are likely to produce more reliable results than disks which were never intended for audio.
I recently installed a Micropolis 2217AV 1.7Gb drive in my Macintosh Quadra 650 and I can report that 'Disk too slow' messages are now a thing of the past for me, although I do only have 4‑track software, so I haven't given it the ultimate test yet. The only drawback of this disk is that its mechanical noise is higher pitched than my other drive and rather annoying, but if your disk recording system uses SCSI drives and the manufacturer gives you the OK, a Micropolis AV drive is recommended at the moment.
Something you need to consider when choosing a disk recorder is the number of physical outputs it has. Analogue and digital tape recorders always have one output per track, but if a disk recorder is capable of playing eight simultaneous segments, that doesn't mean it has eight outputs. It may only have four, and you will have to allocate more than one sound to each output. This imposes limitations on any EQ and processing you may want to provide during mixdown. Even if the system offers some internal EQ or effects, I would say that the optimum provision for a system with a single hard disk drive would be 8‑track capability with eight physical outputs.
At the time when I was looking at suitable drives, I only intended to use a disk system for stereo editing and for assembling masters. The maximum running time of a CD is about 80 minutes, so my first thought was that I needed an 800Mb drive. Then I realised that I would need to record extra material to edit down to this length, so a safety margin would be necessary — 1Gb seemed about right. Then I thought about my other recording activities, which involve recording classical concerts. It occurred to me that if I could copy a whole two‑hour DAT onto the disk in one go, then any timings I took during the recording would apply to the disk recording too and editing would be made much easier. But since the DAT I use for location recording only samples at 48kHz from the analogue inputs, this takes more disk space — so now I'm up to 1.35Gb. I nearly made the mistake of buying a disk this size, but I then realised that formatting and partitioning reduces the capacity, so I had to buy a disk which was nominally 1.4Gb, and now I can record just over two hours and two minutes of stereo at 48kHz All this points out that you'll have to think very carefully about the size of disk you need.
Optical disks are very similar to hard disks and can be used for disk recording in many cases. Their great advantage is that they are removable, so it is possible to afford many more megabytes than you can using fixed hard drives. The disadvantage is that the optical mechanism is heavier than the pickup of a magnetic disk and therefore access time and data rate are not as good. I use a Sony optical disk which can store 300Mb per side and is just capable of stereo recording with my system on a good day. It can play back a stereo recording on any day, so I usually record onto my hard disk and copy onto the optical for storage. But no matter how big I set the buffers in Deck, I can't get 4‑track playback. It just isn't fast enough. Having said this, there are new, larger optical drives coming onto the market which may be viable for multitrack disk recording. This is something I intend to check out in the near future.