You are here

Hard Disk Recording Basics: Part 1

Paul White looks at both sides of hard disk recording, and concludes that although it's generally a good thing, it isn't destined to take over the world just yet. This is the first article in a two‑part series.

One of the questions we're constantly being asked here at SOS is whether hard disk multitrack represents a viable alternative to conventional tape multitrack. Certainly, it is now technically viable to make multitrack recordings direct to hard disk, either via a suitable computer‑based system such as Digidesign's Session‑8, or using a dedicated hardware solution such as Akai's DR4D or Roland's DM80, but before you can decide whether or not this way of working is right for you, you need to be aware of both the strengths and weaknesses of tapeless recording. That hard‑disk recording can provide the same audio quality as CD is something that can now almost be taken for granted, though computers with their own in‑built analogue to digital converters may fall a little short of this ideal.

A major advantage of recording to disk becomes apparent when you realise that it doesn't matter where something is recorded on the disk's surface — it can always be accessed almost immediately. At the simplest level, this means that winding and rewinding tape is abolished at a stroke; just a couple of keystrokes gets you directly from the end of a song back to the start. This is the wonder of so‑called random access, but the advantages go much further than simply ridding you of the tedium of rewinding tape.

Because any part of the disk can be accessed in an instant, you aren't forced to play material back in the same order as you record it, as is the case with a linear medium such as tape. You can record your choruses, verses and solos in whatever order you like, and if the end result doesn't suit you, then it's a simple matter of moving them around until you're happy. If you can only sing a chorus properly once, no problem — just copy it and use it for all the choruses, and if the bass line for every verse is the same, why bother playing it more than once? In fact, when you look at the amount of repetition in a typical pop song, you'll find that the amount you need to record is nowhere near the actual playing time of the track.

This ability to re‑order, copy or remove material without affecting the original recording in any way is known as non‑destructive editing, and a great many of the processes employed by hard‑disk systems are non‑destructive. However, there are times when you want a change to become permanent, in which case the processed data is written back to disk to replace the original data. This might be necessary when you've decided to silence small sections of noise, or where more advanced editing has been used to change the signal at waveform level, such as in the removing of a click or glitch.

Just how much editing you can do on the data, and at what level, depends largely on the system you choose. For example, on a computer‑based system with a waveform display, you may be able to zoom in on the audio waveform and eliminate all those annoying little guitar squeaks or vocal grunts before you get to the mixing stage. With tape, you have to spot erase each one manually, which runs the risk of wiping something you might want to keep.

With digital tape, you lose the ability to thread the tape backwards for the creation of special effects, but with hard‑disk systems, you can usually reverse individual sections of specific tracks to create special effects, or add fade‑ins and fade‑outs (again to individual sections of tracks), change the levels of various sections, and many other things impossible with tape. It all sounds so wonderful that it makes you wonder why tape sticks around at all, but the truth of the matter is that hard disk recording has a size 14 Achilles heel.

Tape may be low‑tech in many ways, but it is reliable, doesn't suffer computer crashes, is easy to use and incredibly cheap. For example, if you're running an ADAT, you pay about 24p per minute for all eight tracks, which works out at just 3p per track minute. A stereo DAT tape works out at around £5 per stereo hour, which, on a track minute basis, puts it into a similar 'pence per minute' ballpark.

The story changes when you look at hard disk, however. A single track‑minute of audio at 44.1kHz uses up around 5Mb of disk space, and with hard drives working out at a little under £1 per megabyte, that makes it more or less £5 per track‑minute, and even being generous, that puts it at 100 times the cost of tape. The cost of disk drives is continuing to fall, with new removable disk systems appearing all the time, but nobody expects disk space to fall to anything like the cost of tape space in the foreseeable future.

The way professional hard disk users get around all this is to back up their work to something like DAT, using dedicated backup software that stores all the edit information along with the audio. The penalty here is that, although DAT is cheap, time is money, and backing up even an 8‑track single could take over half an hour. And, when you want to resume work on the project, you have to wait another half an hour to get your work back into the machine, always assuming that you don't have to back up its present contents first! If you now imagine the backup time when working on an 8, 16, or even 24‑track album project on hard disk, the backup and restore time could run into days!


Having painted first a rosy, and then a very dismal picture of hard disk recording, it's time to put the whole thing into perspective. It's true that hard disk recording uses a lot of expensive disk space, and for serious multitrack applications, a Gigabyte drive is probably the practical minimum size. It's also true that backing up data takes time, but you don't necessarily have to do all your recording on hard disk. Most SOS readers have first‑hand experience with MIDI sequencers, and when working with MIDI, no audio really needs to be recorded at all until you put together the final mix.

The major sequencer manufacturers have now integrated hard‑disk recording into their sequencers, using either the computer's own hardware (in the case of the Atari Falcon) or third‑party hardware, such as the Digidesign cards and audio converter boxes designed to be used with Apple Macs. Even with a relatively small hard drive, you have the capacity to add vocals, guitar and perhaps a solo or two to your latest composition, and because the MIDI data is handled on the same screen as the audio, everything works together quite comfortably. If you do most of your work with MIDI already, then adding a few tracks of hard disk audio is a very practical and flexible way to upgrade.

If you normally use tape but feel hard disk might offer some benefits, devices such as the new ADAT interface might appeal. Again, these rely on Digidesign's hardware, and allow multiple tracks to be cloned from ADAT to hard disk and then manipulated. Once the manipulation is complete, the tracks can be cloned back to ADAT with no timing errors or glitches — you can even, apparently, take something from tape, process it and then put it back exactly where it came from with no evidence of a join.

For the person who likes to record and keep endless numbers of takes or to capture live performances, tape still takes a lot of beating, and I don't think the medium will disappear until archivable, removable disks are available at a minute‑for‑minute cost that's about the same as tape. The truth of the matter is that both hard disk and tape have their own advantages, and as long as that remains to be the case, I'll continue to rely on both.

Random Access

Having extolled the virtues of Random Access digital recording, how does it work? After all, the hard disk itself is a mechanical device, and it takes a small but finite time for the head to move from one place on the disk to another. So why aren't there audible gaps as the head changes position? The secret is the use of buffer memory, which holds a few seconds of audio data. This allows the hard disk's head to take occasional time off from reading or writing when it needs to move to a different part of the disk. It's rather like continually topping up a bucket with holes in the bottom to produce a shower — as long as there is some water in the bucket, the shower keeps running.

Hard Disk Editors

Unlike disk‑based multitrack systems, hard disk editors exist specifically to manipulate and rearrange stereo material. There are two main areas in which hard disk editors tend to be employed. The first of these is for compiling individual tracks to produce a Production Master recording, which will then be used to produce an album.

The second main application is in music editing, where an individual song may be broken down into blocks or sections and then reassembled with a different arrangement. This last application is often used to create extended remixes for 12‑inch release, though it may also be used to shorten songs, to experiment with alternative arrangements, or to assemble a song using the best parts of several takes. Many multitrack hard disk systems may also be used to perform editing, but these may lack the specialist features of a dedicated editing package or system.

A further application for the more comprehensive editing system is for salvaging or improving damaged or imperfect recordings. For example, using Digidesign's Sound Tools, it is possible to redraw small sections of the waveform to eliminate clicks, or use the DINR noise removal software to reduce the level of background hum or hiss in a recording. Other post‑production processes such as EQ or compression may also also be applied to the stored data.


Because hard disk recording systems already require DSP power to process data, it only requires more processing power to add digital mixing, EQ and maybe even a few effects. This is exactly the route taken by Digidesign with their Session‑8 and Pro Tools product lines, and the introduction of TDM, their new multiway digital buss system, means that third‑party hardware (mainly plug‑in cards) and software can be combined to produce a highly sophisticated audio recording and processing system, all based around an Apple Mac and Digidesign's core hardware.

Professional audio workstations already take the integrated approach, while audio with MIDI programs, such as Cubase Audio, Logic Audio and Studio Vision include basic mixing facilities with level, EQ and pan that may be automated via MIDI. The current state of the art sees many of these systems now providing video support, and this enables soundtrack work to be undertaken sync'ed to picture. This works by using random‑access, low resolution, on‑screen video from disk, so complex productions can be completed without the expense of traditional audio/video edit‑suite equipment.

The future will almost certainly see even more computer‑based, direct‑to‑disk systems emerging, but with the addition of dedicated hardware control surfaces, which will make them feel more like a conventional mixer and recorder. As the number of features crammed into these systems continues to increase, it will be no longer be practical to control them all using just a mouse, and more ergonomically appropriate control surfaces will become essentials, rather than the luxuries they are currently considered to be.

Hard Disk Hardware

Audio data is recorded directly onto a hard disk unit via interface hardware. This comes either as an add‑on for a personal computer, or as part of a dedicated hardware disk recording system. When you take the computer add‑on route, the hardware comprises a set of analogue‑to‑digital and digital‑to‑analogue converters to handle the audio, and usually some form of DSP card; most computers can't cope with high‑quality audio as they stand. The obvious exceptions to this rule are the Atari Falcon and the AV Macs.

The hard disk used for audio is exactly the same as used for more normal computer work, though multitrack recording demands fast access times and data transfer rates, so don't rush off and buy the first bargain drive you see until you've checked that it'll do the job. Also, avoid recording audio on the same drive you use for running your day‑to‑day software. A separate drive is more convenient — it allows you to empty the drive ready for a new job without having to run defragmentation software every session or two.

As a rule of thumb, stereo material sampled at 44.1kHz (the same as CD) uses around 10Mb of disk space per minute, so if you're putting together a stereo system to edit albums, you'll need around 600Mb of disk space to hold an hour's worth of material. Of course the usable time is halved every time you double the number of tracks, so using the same 600Mb drive, you're looking at a quarter of an hour of continuous 8‑track or less than 8 minutes of 16‑track.

Audio can be recorded to disk either from a digital source such as DAT, or as a conventional analogue signal. Analogue recordings are generally used for multitrack work simply because most of the source material is analogue, but when editing stereo material, it's best to come in digitally, using the SPDIF connection from a DAT machine so as not to sacrifice any signal quality. It is important that good quality cable and connectors are used for digital data transfer, and that the cable is kept as short as possible. Audio cable is not really satisfactory for the job — 70Ω digital cable should always be used to minimise errors, which may cause clicks or glitches.