Hard-Disk Recording

Ultimately, most electronic musicians want to record their music so it can be played and enjoyed (and hopefully bought!) by others. Most of us are familiar
Image placeholder title

Ultimately, most electronic musicians want to record their music so itcan be played and enjoyed (and hopefully bought!) by others. Most of usare familiar with analog tape, such as 1/4-inch reel-to-reel orcassette, which has been widely used for many years. These days,however, more and more musicians are recording their materialdigitally, which is fundamentally different from analog recording. (Forthe basics of digital audio, see "Square One: Digging into DigitalAudio" in the February 1996 EM.)

Image placeholder title

Digital audio signals can be recorded to tape, hard disk, MiniDisc,or computer memory (RAM or ROM). For long musical parts (e.g., vocals,acoustic guitar parts), tape, MiniDisc, and hard disk are the preferredmedia because they can hold much more data than RAM or ROM, and theyprovide permanent but changeable storage.


Once you decide to digitally record your tracks, you must thendecide whether to record to tape or hard disk. If you decide to use ahard-disk recorder (HDR), you must choose between a computer-based ormodular system. For now, we’ll focus on hard-disk recording.

Computer-based HDRs (which are also called digital audioworkstations or DAWs) require a PC, Macintosh, or other general-purposecomputer and include a software package and perhaps a hardwareinterface to get audio into and out of the computer. These systemsoffer a large graphic display (the computer monitor) and, in somecases, integration with MIDI sequencing. Effects are sometimesincorporated into the basic software and often can be added in the formof DSP plug-ins (more on this in a moment). Of course, computers arerelatively expensive, especially those with enough horsepower to handlehard-disk recording. Desktop computers are not easily portable, butmodern laptops can often perform hard-disk recording, as well.

Modular hard-disk recorders (M-HDRs) are self-contained units withbuilt-in hard disk, mixer, and, in many cases, effects. Their "brains"are nothing more than computers dedicated to hard-disk recording. Inmost cases, you can combine several M-HDRs as need requires and budgetallows, and you can sometimes control multiple M-HDRs from a singlecontrol panel. This is similar to modular digital multitrack (MDM) tapedecks.

M-HDRs are dedicated devices with familiar, tape deck—stylecontrols. In addition, they are quite portable. However, they offer farfewer editing features than are available in most computer-based DAWs.M-HDRs have a relatively small display, which makes even basic editingmore difficult than with a DAW. In addition, they require externalsynchronization to integrate with a MIDI sequencer. On the up side,M-HDRs are much more stable–crashes are far more common withcomputer-based systems–and because they have fewer in-depthediting features, they are generally easier to master.


Digital audio that is recorded at a sampling rate of 44.1 kHz with aresolution of sixteen bits requires about 88 KB per second per track or5.3 MB per minute per track. As a result, a 3-minute song with eighttracks would consume 127 MB of space. I recommend that you use a harddrive with at least 1 GB for digital audio data.

Some HDR systems use data compression to reduce the storagerequirements. Common forms of compression include µLaw (pronounced"myu-law"), Macintosh Audio Compression/Expansion (MACE), and AdaptiveDifferential Pulse Code Modulation (ADPCM). Some types of compression,including most of the types currently in use in audio systems, candegrade the sound quality. Lossless compression does nothing to thesound quality, but it yields the least amount of storage savings. Lossycompression actually removes some of the data, which can degrade thesound quality, but it yields greater savings, typically from 4:1 to 6:1or more.


HDRs can use several types of hard-disk media (which I’llcover shortly). However, they must all meet certain minimum criteria tobe practical for this application. The most important of these criteriaare average access time and throughput.

Access time (also called seek time) is the time it takes the driveto find a piece of data anywhere on the disk. Of course, if the diskdrive’s read/write head is near the location of the desired data,it takes less time than if the head is far away from the data. As aresult, the average access time is calculated and used as abenchmark.

Throughput (also called data transfer rate) is the amount of datathat can be sent to and from the disk per second. You can record manytracks on most hard-disk systems, up to the capacity of the disk.However, you can only play a limited number of tracks simultaneously,depending in part on the processing speed of the computer and theaverage access time and throughput of the disk.

If you’re shopping for a hard-disk drive to use for digitalaudio recording, you might think it’s as easy as finding a drivethat meets certain seek time and throughput specs. As a general rule,any hard disk used for digital audio recording must have an averageseek time of 12 milliseconds (ms) or less and a sustained throughput of3 to 4 MB per second (MB/s) or more. However, it’s actually a bitmore complicated than that. For one thing, the minimum acceptable specsdepend on the number of tracks the system must deal withsimultaneously; an 8-track HDR requires a faster hard disk than a2-track system.

In addition, many hard drives perform a routine called thermalrecalibration, which compensates for slight changes in the size of thedisk platter due to temperature variations. If this occurs duringrecording, you might miss several milliseconds of data. You cansometimes turn this function off or tell the drive to perform it onlywhen it is not writing or reading data. Fortunately, many modern harddrives now use other means of compensation for temperature changes.

Finally, the type of connection between the computer’s centralprocessing unit (CPU) and the hard disk affects how much data can betransferred to and from the disk in a given amount of time. Some of themore common types include SCSI, IDE, and ATA. In addition, each type ofconnection includes at least two variations, each with its own maximumthroughput. For example, the original version of SCSI can deliver atheoretical maximum throughput of 5 MB/s, while the newer Fast SCSI-2can sustain 10 MB/s. Then there’s Fast and Wide SCSI-2, which cansustain up to 20 MB/s. (For more on these hard-disk connections, see"The Windows Studio" in the July 1996 EM.)

Fortunately, you don’t typically need to worry about theseissues. All you need to look for is an "A/V capable" drive, whichshould meet all minimum requirements for recording digital audio. Infact, most modern hard drives are A/V capable. Some computer-basedDAWs, such as SADiE Inc.’s SADiE 3, supply a turnkey system,complete with suitable hard drive, while others, such as the varioussystems from Digidesign, provide a list of drives that have been testedfor compatibility.


There are several types of hard-disk media, and most are availablein external boxes or as internal units for computers or M-HDRs. Thetraditional type of hard disk is called a fixed disk, which ispermanently sealed within an enclosure.

Removable cartridges behave much like floppy disks, but they holdmuch more data. Most of the older removable-cartridge drives are notfast enough for hard-disk recording, but recent advances in removabletechnology have enabled a few such products to be used for thispurpose. Examples include the Iomega Jaz (1 GB) and SyQuest SyJet (1.4GB) and SY270 (270 MB). Akai uses magneto-optical (MO) removable mediain their 8-track DD1500 M-HDR. This requires a custom controller chipand some sophisticated buffering of the data because MO technology isotherwise too slow for hard-disk recording.

One of the most important advantages of removable media is the factthat you can easily store the audio for each project on a separatecartridge. This makes it easy to keep track of your data and lets youtake the cartridge to other studios. Removable media are also great forbacking up your data (more on this in a moment).

MiniDisc is a relatively new type of removable MO cartridge that isbeing used in low-cost M-HDRs from Yamaha, TASCAM, and Sony. Theseunits resemble the ministudios of the past, except that they useMiniDisc cartridges instead of cassette tapes. They can record and playup to four tracks of audio (no virtual tracks are possible), and thestorage capacity of a cartridge is 140 MB. These units use lossy 5:1compression to record up to 37 minutes of audio per track. (You canalso record in stereo for a total time of 74 minutes per track or inmono for a total time of 148 minutes.) According to the manufacturers,if you want to record four tracks, you must use MD Data cartridges, butfor stereo recording, standard MD Audio cartridges work fine.


Basically, recording on an HDR is similar to using a traditionalanalog tape deck. Most HDRs provide tape-style transport controls, suchas Play, Record, Rewind, Fast Forward, and Pause. Unlike tape decks,however, HDRs take virtually no time at all to jump from any point inthe music to any other point thanks to random access. This means thatthe hard disk can find any piece of data in roughly the same time asany other data. By contrast, tape is linear; it must be shuttled tofind a particular spot in the music. You can also punch in and out, andthis does not necessarily replace the material in the punchsection.

Most HDRs let you record lots of tracks (though rarely more thaneight at once), and you can typically play between two and eight trackssimultaneously from a single unit. Some systems (such asDigidesign’s Pro Tools III) allow expansion to sixteen tracks andmore. Even though an HDR might be called an 8-track device, it cantypically hold many more tracks of data. These are sometimes calledvirtual tracks, which let you record many takes of each part and selectthe best one for playback.

In most cases, random access, nondestructive punches, and virtualtracks are possible thanks to the use of pointers, which are internalindicators the computer uses to identify and manipulate differentsections of the audio data. For example, let’s say you’verecorded a guitar solo, and you punch into the middle of the solo tocorrect some mistakes. On tape, this would destroy the originalmaterial in the punch section.

With nondestructive editing on an HDR, however, the new material isstored on a different part of the disk, leaving the old materialuntouched. When the solo is played back, the computer uses pointers tojump to the new material and back to the old material at the correctmoments. This lets you use either version of the punched section.

Pointers are also used to select the virtual tracks you want toplay; you can even assemble material from several tracks into onecomposite track without destroying the original data. The userdoesn’t work with pointers directly. Instead, you tell the HDRwhich parts of which tracks you want to use by creating a playlist(sometimes called an edit decision list, or EDL), which is a list ofthe audio sections you want to play back in a certain order.

Many studio operators need to sync the HDR with a MIDI sequencerand/ or tape deck. This is not a big issue if you are using a digitalaudio sequencer that integrates hard-disk recording and sequencing. Ifyou are using separate programs to sequence and record to hard disk,internal synchronization usually works well but not always. There aredifferent degrees of sync: some systems just trigger the audio filesand MIDI files and hope they stay together, and other systemsrepeatedly check the synchronization and adjust the playback asrequired. Obviously, the more often the system checks itssynchronization, the tighter the sync is likely to be.

Most M-HDRs include some form of synchronization capability, such asMIDI Time Code (MTC) or SMPTE. In some cases, these devices can only bethe master time-code source, which means that the other devices in yoursystem must sync to the HDR. In other cases, the HDR can be master orslave. If you are using a tape deck in conjunction with an HDR, thetape deck must be the master, so your HDR should be able to sync toit.


One of the biggest advantages of HDRs is their editing capability,which is not available on tape decks for the most part (unless you likeediting with a razor blade). Typical editing functions let you cut,copy, and paste sections of digital audio. When you copy and paste asection of digital audio in a nondestructive system, the data is notactually copied and pasted. Instead, pointers are used to play the dataat any moment you want during the song. This lets you record a shortriff and repeat it as many times as you want without using up valuablestorage space. Similarly, when you cut a piece of data, the data is notactually erased from the disk, only the pointer to that data. You canalso merge the data in several tracks to one track so they all playback together.

Other common editing functions include fade ins, fade outs, andcrossfades between different sections of data. Normalization adjuststhe amplitude of the material so the highest peaks correspond to thesystem’s highest allowable level to maximize the signal-to-noiseratio. You can also reverse a section so it plays from back tofront.

In most cases, these operations are nondestructive. If you edit apiece of data and you don’t like what you did, you can undo it,and the original material returns unchanged. A few operations might bedestructive, but the HDR will typically warn you of this before itproceeds with the operation.

Most HDRs include various effects-processing operations that can beapplied to any section of the data. These include reverb, delay,chorusing, flanging, compression/limiting, and EQ. In many cases, theeffects can be applied in real time as the data is playing back. Thisis much like sending the audio through an outboard effectsprocessor.

In other cases, the computer must take some time to process the datawith the desired effect. In this case, you must wait for the computerto finish its processing before you can play the material. This issometimes called offline processing, which is often destructive, butthe computer typically warns you before it proceeds.

Computer-based HDR software often accepts ancillary programs calledplug-ins, which let you add various forms of signal processing. For acomplete rundown on plug-ins, see "The Budget Desktop Studio" in theSeptember 1996 EM.

In the end, you will probably want to mix your tracks and recordthem onto a stereo master tape. Most HDRs include their own internalmixer. In computer-based HDRs, the mixer appears on the screen. Thefaders and other controls can be manipulated with the mouse, but thisis inefficient and sometimes impractical. It’s much better to usea MIDI fader box or other dedicated mixing surface, such asJLCooper’s FaderMaster or CS-10 or Peavey’s PC 1600. Mixermanipulations can usually be stored and recalled during playback, whichprovides automated mixing.

Some M-HDRs, such as the Roland VS-880, include a physical mixingsurface, which makes it easy to mix. Other M-HDRs, such as the E-muDarwin, have an internal mixer with no physical controls. In this case,the mixer is controlled from a MIDI fader box or computer software.Because the internal mixer and effects are usually digital (i.e., thedigital signals never leave the digital domain), there is none of thedegradation of signal quality that so often accompanies conversionsfrom analog to digital and back again.


Despite their popularity, hard disks are finicky creatures; theycrash and corrupt data. As a result, it’s critical to back upyour data regularly (preferably after each and every session).Removable cartridges provide the easiest solution; if your primarystorage crashes, the cartridge can be used immediately. You can alsouse a data DAT deck, but in that case you have to restore the data to ahard disk before you can use it. Many modern computers and some M-HDRsinclude a SCSI port, which lets you connect an external device, such asa removable-media drive or data DAT deck. (Some PCs require an add-onSCSI card.)

A few systems let you back up to an audio DAT deck or MDM, such asan Alesis ADAT or TASCAM DA-88, but this also requires that you restorethe data, and it’s a bit of a kludge. For example, the FostexDMT-8 has no SCSI port, but it backs up two tracks at a time to audioDAT via the S/PDIF digital audio outputs. The Digidesign HDR systemscan also back up to audio DAT using a program called DATa. This programbacks up the playlist information followed by the actual audio data,all of which is sent to the DAT deck from the digital audio outputs onthe interface hardware.

As you record and edit material, it is stored in different places onthe disk. If data already exists on the disk and the system is unableto write an entire file in one contiguous area, the data must be storedin bits and pieces that are reassembled by the computer upon demand.Eventually, the data is so fragmented–spread out–that thecomputer can no longer find it efficiently. This also occurs when youdo a lot of destructive edits, resulting in data being added anddeleted. If the data becomes fragmented enough–and this canhappen surprisingly quickly–disk access is slowed, and eventuallycrashes can occur.

As a result, the disk should be defragmented or optimized every sooften, depending on how much you use the system. (Defragmenting putseach file in a contiguous space; optimizing not only defragments eachfile but reorganizes the entire disk so that data files are storedcontiguously, applications are stored contiguously, and so on.) This iseasy on a computer-based system; simply run a disk-maintenance programand defragment the disk. (Of course, make sure to back up the diskbefore performing the defragmentation; see "Desktop Musician:Don’t be a Crash Dummy" in the September 1996 EM.) Most M-HDRsdefragment automatically or provide a method to initiate the processmanually.

If you opt for a computer-based HDR, I recommend that you dedicate aseparate hard disk to your digital audio data. Of course, you canrecord onto your primary hard disk, which also holds your operatingsystem, applications, and other data files. But dedicating a hard diskto digital audio makes it easier and safer to defragment and back upthe disk on a regular basis.

Affordable hard-disk recording is a great boon for musicianseverywhere. This technology lets you record and edit audio withunprecedented ease and flexibility. All that remains is to try it outin your system.

Scott Wilkinson backs up his hard disk after every sessionto keep the crash gremlins away.