Magazine Archive

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

How It Works - Hard Disk Recorders (Part 12)

An Alternative To Tape

From the analogue to the digital era, magnetic tape has been the prime production medium for recorded music for several decades. David Mellor examines new developments in recording and speculates where they might lead.

From the analogue to the digital era, magnetic tape has been the prime production medium for recorded music for several decades. David Mellor examines new developments in recording and speculates where they might lead.

New England Digital's 8-track Direct-To-Disk/Post-Pro digital recording and editing system, with Macintosh II front-end.

Music is a two-dimensional phenomenon. As natural sound travelling in air, it has the dimension of sound pressure - the actual pressure exerted by the air particles taking part in the sound wave, and the dimension of time the period of minutes and seconds over which the musical discourse unfolds.

To store or record music so that it may be retrieved at some later date, it is necessary to construct analogues of these dimensions which will have a lasting presence. Sound pressure is a transitory phenomenon, in free space, and time - as we all know — seems to vanish all too quickly.

There are several analogues for sound pressure that can be stored on a fairly permanent basis. Relative position from a reference point is one, as used in the good old gramophone record, where the position of the groove from its spiral course is a measure of how loud the sound is meant to be at any instant. Another analogue for sound pressure is the degree of transparency of a piece of photographic film, as in the optical soundtrack of a movie.

Magnetic flux density is another. Representation of sound pressure by magnetic flux has been possible since the turn of the century, and in the period between the late 1940s and the early 1980s this process has been the basis of nearly all professional sound recordings.

Representation of the time dimension of sound is nearly always in the form of length - the length of a piece of tape or the length of the groove in a gramophone record. For example, if a tape recorder turns each second of music into 15 (or 30) inches of tape, then a half-hour recording takes up 2250 feet of tape.

Digital recording also uses the length of a tape (reel-to-reel or DAT cassette) to represent the time dimension of music. But instead of using magnetic flux density directly as the analogue of sound pressure, it resolves all the possible sound levels into 65,536 discrete stages (in a 16-bit system) and allocates a number to each stage. These numbers are then stored on the tape in the form of changes in magnetic flux density.


In live performance a piece of music starts, then proceeds continuously throughout its length until it finishes. To record the music, the tape recorder is started just before the music, the tape spools through while the music is being played. Then when the music has ended, the tape is stopped. There is a direct and very obvious connection between the time characteristics of both the music and the tape — straight through from beginning to end, every second of music turning into 15 inches of tape (assuming the tape recorder speed to be 15ips).

But in rehearsal, things are quite different. Let's assume an orchestra is playing from a printed score. The conductor may run through the first page, then tell the musicians to go to bar 556. This takes a few seconds to turn the pages, then the music starts again. In effect, the conductor has taken a short cut through time. This process is mimicked on tape by the use of the Fast Forward button which, on a DAT machine, can go through a minute's worth of tape in one second.

There are more similarities: in the conductor's score, and the musicians' parts, there may be repeat marks, where the composer wanted a section to be played twice but didn't want to have to write it all out again. Or there may be a Da Capo, which is musical Italian for 'go back to the beginning'. The same processes can be performed with tape, by copying the relevant sections and splicing them back into the master.

The performance of a repeat in music is an instantaneous process. As soon as the musicians know that they have to make a repeat, and they know where to go to in the music (of course, they will have already mentally logged the start of a repeat section), they make the jump, and edit the time dimension of the music.

Editing the length dimension of tape (the analogue of time) is a more difficult process, requiring a copy to be made in this instance. The length of a piece of tape is not as manipulable as the time component of music.


Tape uses the dimension of length to represent time. I said earlier that a record groove is similar. True, it has a length. But also, because it is wound into a spiral, it has an extra dimension - width. This makes it easier to quickly traverse the time dimension, by taking a 'short cut' across the width of the record. It is very much quicker to find a particular passage of music on a record than it is on tape, due to the two-dimensional access we now have available.

So, with a record, we find it much easier to 'go to bar 556', because of the easy access. The problem remains that bar 556 is not identified on the record, as it is on the musical score — unless you count the record grooves as you traverse the disc. If we had a recording system that offered two-dimensional access together with proper identification of the timing and the different sections of the music, then we would have a system that offered as much scope for controlling and editing music as real musicians have - and, of course, it would play the right notes every time!


A recent arrival in the studio world is the hard disk recording system. It uses techniques developed by the computer industry to provide two-dimensional access of digital recordings, with identification of the timing of the programme content and musical sections. (Although I refer to music, it could just as well be segments of speech.) It sounds ideal as a storage medium for music, and in many ways it is ideal. But let's look first at exactly what a hard disk is.

Most of us are now familiar with floppy disks, used in computers, samplers, sequencers and synthesizers to store data. If you open the casing of a floppy disk (not one you intend to use again!), you will find a flexible plastic disc with a magnetic coating, very similar to the coating on magnetic tape. Indeed, you could consider it to be magnetic tape, but in a different shape and slightly more stiff.

Inside the disk drive of the computer, or sampler or whatever, there is an arm roughly similar to the pickup arm of a record player, which is able to traverse the rotating disk, under the direction of the disk controller circuitry and software. On the end of the arm is a magnetic head, similar to the heads of a tape recorder. The disk's magnetic head is optimised for the conditions under which it works, and can play back (read) as well as record (write).

A hard disk is very similar to a floppy disk, but with some important differences. The hard disk drive contains several disks mounted on a common spindle. These disks are rigid and are never removed (except in the case of certain types used solely in mainframe computer systems). The unit is completely sealed from the outside world (except perhaps for a pressure relief vent) to prevent contamination of the disk surfaces.

Figure 1. Hard Disk internals.

Figure 1 shows the internal structure of a typical hard disk unit. In this instance, there are six active disk surfaces, each with its own read/write head and head positioning arm. The heads do not touch the surface of the disk but ride above it on a lubricating film of air.

The advantage of the hard disk over the floppy disk is that the data storage can be much more dense. Because the disks are rigid and protected from contamination, the heads can be allowed to ride much closer to the disk's surface. Magnetic theory states that the closer the head is to the recorded data, the more tightly packed the data can be. Even so, the data storage density is not as high as digital tape. With tape, the heads can get even closer to the recorded data.

Figure 2. Hard Disk sector, cylinder, track, and block arrangements.

Let's assume for the moment that one channel of digital audio is to be recorded onto the hard disk. Figure 2 shows the organisation of data storage into blocks, tracks, cylinders and sectors.

Each of the circular tracks on each disk surface is divided into a number of blocks. The block is the smallest unit of storage and will contain a number of milliseconds of one channel of digital audio information. As well as recording the audio data, additional control data is recorded to identify the timing of the audio in the block - ie. its relationship to the rest of the programme. Incoming audio data is split up into blocks which may be recorded at any position on any disk surfaces. Continuous music does not have to be recorded in a continuous path across the disk, the controlling software will place the data where it thinks fit, and keep tabs on which blocks should be replayed in which order.


Let's now jump to the user's viewpoint of a hard disk recording system — probably in front of a QWERTY (typewriter) keyboard and VDU screen. Imagine you have two 'takes' (versions) of a piece of music, already mixed, finished, and available on a conventional tape storage medium. One take has a wrong note a quarter of the way through, the other has a similar fault three quarters of the way through. The hard disk system is to be used to edit these two takes into a single perfect performance.

The first stage of the process is upload. Starting with the first take, the music is digitally copied onto the hard disk. The two takes will be stored as two sound files, analogous to the text or data files stored on a computer's hard or floppy disk. As Take 1 is copied into File 1. the hard disk recorder will split up the music into blocks of data, and record these blocks separately and probably discontinuously onto the surfaces of the disks. Similarly, Take 2 is copied into File 2.

Once copied, the hard disk recorder now has virtually instant access to any part of either take. Ask the unit to play Take 1 and Take 1 will start instantly. Ask it to play Take 2 and Take 2 will start as soon as the instruction has been entered. If you want, you can skip about from section to section of the music, in either take, and the hard disk recorder will respond as quickly as you can type - and, depending on the software, once you have identified the sections you are interested in, this may be done as simply as pressing one key. If you could look inside the hard disk unit, you would see the positioning arms hopping backwards and forwards, extracting the data from the blocks and loading it into a buffer memory, from which it would be read out as a continuous stream of data to the digital-to-analogue convertor (DAC).

Now, to edit the two takes together, you would find the in and out points just as you would using ordinary tape, but controlled by the keyboard. The essential difference is that with tape you have to physically cut the tape and join it together. Either that or copy the two takes, played on separate machines and switching at the right moment, onto a third recorder. But with the hard disk recorder, all that is needed is the edit decision data. You tell it where to make the join and the hard disk recorder will perform the edit in real time. It doesn't have to copy the result onto a separate area of the disk. All it does is read the data from File 1 up to the edit point, and from File 2 thereafter. This happens in real time every time you replay the edited version of the music. No data is changed and no copy is made. Figure 3 shows the operation with a real-time crossfade at the edit point. All this is done using the speed of access of the hard disk, and software processing of the data.

Figure 3. Crossfade between two sound files.

When you are happy with the result, the finished edited performance is downloaded back onto digital tape.


So far, all we have discussed is single channel audio recording. What more can hard disk recorders offer?

The answer is that they can function as digital multitrack recorders, too. Several systems are currently on offer that can provide up to 8-track operation. Eight tracks doesn't sound a lot when most of the studio industry runs on pairs of synchronised 24-track machines, but hard disk recorders have other advantages to compensate.

The first point to emphasise is that it doesn't matter to the hard disk itself whether it is being used to record single channel or multiple channel audio. All that is important is that there is enough time for the system to get all the data onto the disk and back off again. A typical hard disk can store perhaps two hours of single channel audio. With the correct number of inputs, outputs, D/A and A/D convertors and the appropriate control software, the same hard disk could record 15 minutes of eight simultaneous channels. The capacity of a hard disk is specified in terms of the number of track-minutes it can accommodate. The recorder itself will be capable of handling several hard disk drives - how many you have depends on what you intend to use the system for, and how much you can afford to pay.

As far as editing the eight tracks goes, the procedure is the same as in the single channel example given above: lots of versatility and real-time playback of edit decision lists.


The first advantage an 8-track hard disk recorder has over an 8-track tape recorder is that the tracks can be slipped backwards and forwards against each other in time. For music, this means that a lagging synth part (say) can be replayed a few milliseconds earlier to compensate. In film dubbing (currently one of the major applications of hard disk recorders), layers of sound effects can be easily synchronised to the action. To do this with tape means copying onto a separate recorder, adjusting the timecode offset, and re-recording back onto the multitrack - a time-consuming, and thus costly, procedure.

A second advantage is that although only eight tracks can be replayed at any one time, there could be any number of tracks contained on the hard disk, up to the limit of its capacity. There could be a dozen takes of a guitar solo, or a vocal, but only the best would be chosen and allocated to an output. More than this, the 12 takes could be edited - using the edit decision list technique once again - to combine the best parts of each.

Producing an extended mix, or perhaps a two-and-a-half minute radio version, of a song would also be child's play compared to the endless copying and editing required to do the same thing with tape.

Example display from DAR's Soundstation II, which utilises a novel 'touch screen' for control.


The current generation of hard disk recorders do have their limitations, the major one being the access time necessary to retrieve a piece of audio information from the disk. To move the head to the correct disk location takes a few milliseconds, and the music can't wait.

The time that is lost moving the head is made up by reading out the data more quickly into the buffer memory, which can then read it out at the correct speed, but there is a limit. There is also a limitation on how quickly the data can be processed. Real time crossfading takes a certain amount of computing power; crossfading on eight channels simultaneously takes more computing. An eight channel crossfade combined with varispeed may just be enough for the system to say 'Hang on a minute'. But as with most areas of hardware and software development, things are getting better all the time.

There are also other problems peculiar to hard disk recorders. You may have noticed one already: "These disks are rigid and are never removed." A quote from an earlier paragraph. A typical hard disk editing session would start with uploading the disk from tape, a procedure that takes as long as the recording itself, and end with downloading the result back to tape. Time is money, and some hard disk recorder manufacturers rationalise this by saying that you will probably want to listen to the music and make some editing decisions anyway, before the session starts properly. And when you have finished, you can audition the result and make a cassette copy during the download. If you just wanted to store the products of a session overnight before continuing the next day, then a special tape storage system is sometimes available. This can store the contents of the hard disk on tape cartridges, freeing the disk itself for other use. This kind of tape storage works more quickly than real time, and subsequent upload is faster too.

Of course, the major disadvantage of the hard disk recorder is the cost. We are talking about tens of thousands of pounds. But it is early days yet. When the power of the hard disk recorder is more widely recognised, manufacturers will start to find ways of bringing the costs to more reasonable levels. Remember that the key to the versatility of the system is its two-dimensional access to the data. It doesn't necessarily have to be a rotating rigid disk and it certainly doesn't have to be magnetic. There may well be other methods to achieve the same result.

Looking into my crystal ball, I can see the demise of the tape recorder as we know it, and its replacement with a much more versatile recording system. And it might happen much more quickly than we imagine.


Many samplers are now incorporating hard disks, either as standard equipment or optional extras. It is important to remember that these hard disks are fairly small in comparison to the hard disks necessary for recording digital audio in bulk.

A rough guide is that 1 Megabyte of storage can cope with 10 seconds of single channel audio (16-bit, 44.1kHz). Therefore, to record a 30 minute stereo programme, a 360 Megabyte disk drive would be necessary - some 10 times larger in capacity than those commonly used in samplers and most personal computers.


Developments in hard disk recording and editing systems are rapid, so it is worthwhile contacting the manufacturer or distributor listed below for current information on the capabilities of their system.

AMS/Calrec AudioFile (Contact Details).
Audio+Design Sound Maestro (Contact Details).
Digital Audio Research Soundstation II (Contact Details).
Lexicon Opus (Contact Details).
New England Digital Synclavier and Direct-to-Disk system (Contact Details).
WaveFrame AudioFrame (Contact Details).

Previous Article in this issue

Elka CR99 MIDI Disk Recorder

Next article in this issue

Mister MIDI

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 - Apr 1989


Digital Audio



How It Works

Part 1 | Part 2 | Part 3 | Part 4 | Part 5 | Part 6 | Part 7 | Part 8 | Part 9 | Part 10 | Part 11 | Part 12 (Viewing)

Feature by David Mellor

Previous article in this issue:

> Elka CR99 MIDI Disk Recorder...

Next article in this issue:

> Mister MIDI

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 October 2021
Issues donated this month: 8

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

Funds donated this month: £52.00

All donations and support are gratefully appreciated - thank you.

If you're enjoying the site, please consider supporting me to help build this archive...

...with a one time Donation, or a recurring Donation of just £2 a month. It really helps - thank you!

Small Print

Terms of usePrivacy