The Science Of Sound Recording: Digital Recording

August 14, 2012, by Jay Kadis


Here we present a portion of a chapter in the book “The Science Of Sound Recording” by Jay Kadis, published by Focal Press.


Whatever system is used to acquire digital data, we are faced with the same dilemma we encounter in the analog recorder when it comes to storing the information.

Although computer memories may store the data temporarily, most computers use dynamic random access memory (DRAM) chips that lose the data when the power is removed.

For long-term storage, magnetic media are used for digital recording as well as for analog.

Because digital data require only two states and not the complete linearity demanded by analog recordings, there is a difference in the way the process is applied.

Digital magnetic recorders use saturation recording, leaving all magnetic domains polarized in one direction or the other with no intermediate levels required, so the bias current used in analog recording is unnecessary in digital recorders. The density is quite high in digital recorders, which introduces some problems not encountered in analog recorders.

The number of bits per unit area of medium is limited in the longitudinal recording method used for analog record, although it is sufficient for the demands of analog recording.

Digital magnetic media may benefit from closer packing of domains, which is achieved by using perpendicular recording, in which the domain magnetic fields are magnetized perpendicular to the medium surface instead of along the surface as in analog recorders (Figure 1).

Digital data stored magnetically requires only two discernable states for binary information. This requirement is easily achieved by magnetizing domains fully in one or the other polarity.

Figure 1: Comparison of longitudinal recording used in analog magnetic recording and perpendicular recording used in many disc drives. (click to enlarge)

Though this approach avoids the nonlinear region of the M-H curve, it introduces another problem: the interference between closely occurring bits. If the write head and the medium are not capable of altering the magnetic polarities as rapidly as the bits are changing, the magnetization from the previous bit will affect the next bit, which causes the data to be altered because the overlap makes discriminating between ones and zeros unclear. This intersymbol interference limits the data density that can be stored.

We have several options for storing digital data, including dedicated devices using tape or discs as media and general-purpose personal computers with added interfaces to acquire and convert analog audio. The high data density required for storing digital data made early digital recorders quite complicated, requiring rotating magnetic head recorders designed for video recording or using high tape speeds with stationary heads requiring more than one data track for each audio channel to provide enough bandwidth.

The personal computer has largely replaced the mechanically complicated digital recorders as the preferred storage device for digital audio recordings. The low price and wide availability of large, fast disc drives has spurred a move to the computer as the digital audio recorder of choice, especially as the computer can take on the functions of editing, mixing, processing, and storing the entire project in a single device.

A recent development is the flash RAM chip, popular in USB memory sticks for example. This magnetic nonvolatile RAM – though slower to read and write than a hard disk – is becoming popular for non-time-critical recording such as backup of sound files and stereo sound file distribution.

The evolution of digital recorders has been rapid. Rotary-head modular digital multitrack recorders and stereo DAT recorders enjoyed only a few years of widespread use before the move to the general-purpose computer as the preferred platform for digital recording. These machines temporarily bridged the gap between high-cost stationary head professional digital recorders like the Sony DASH and Mitsubishi Pro-Digi systems and analog multitrack.

The Alesis ADAT and TASCAM DTRS machines used videotape, which was cheap and readily available, to provide inexpensive access to digital recording for a wide range of users. These machines, though initially inexpensive, suffered from their complexity when head wear and transport malfunctions required difficult repair and diagnosis procedures.

Yamaha produced the DMR/DRU series of recorders, which used stationary heads and proprietary tape cassettes, to deliver 20-bit 8-channel digital recording in the early 1990s, but they were expensive relative to the ADAT and DTRS machines and never caught on.

None of these tape-based systems survived the move to computer-based systems, and all have been phased out or will soon be retired. Although tape provides the advantage of removable media, the large size of hard drives and the availability of plug-and-play computer interfaces for storage media has diminished the attractiveness of tape-based digital recorders.

The ability to use inexpensive, mass-produced personal computers for digital audio recording and mixing has greatly expanded the accessibility of these tools.

The addition of a FireWire, USB, or Thunderbolt audio interface and some software is all that is required to create a digital studio entirely within the computer.

This change has had a dramatic effect on the recording studio and the music business in general.

Essentially, the entire recording studio can now be contained in a single piece of equipment, with the ability to recall the entire project and studio configuration in a few seconds.

The advantages of digital audio are hard to ignore, even for those dedicated to the analog studio paradigm.

Use of personal computers for audio recording has introduced a new set of difficulties.

Each operating system and hardware platform requires different software, and there are differences in the bus structures and interface ports available that complicate the choice of peripheral audio interfaces (Figure 2).

Figure 2: Computer interface speeds. (click to enlarge)

Input/output buses include FireWire and USB high-speed serial interfaces, both of which are possible choices for connecting multichannel A/D and D/A modules to the computer to provide audio access. The Thunderbolt interface protocol promises even faster device interconnection.

The software for recording interacts with the operating system to access these audio inputs and may do so with differing speed capabilities on different computers.With the main choices for personal computer operating systems – Macintosh OS X,Windows, and Linux – several types of interface are supported, but different recording programs are required and the performance of the audio interfaces may differ due to differences in the hardware and device drivers employed in the particular computer used.

Recording engineers must now have some knowledge of the internal workings of their computer. If something goes wrong with the recording system, it becomes necessary to isolate the problem by troubleshooting a complicated series of interactions between software, computer, and peripheral hardware that may not be well documented. Each manufacturer provides information about their part of the system, but no one company is responsible for the entire system, leaving the user to deal with the problem.

An issue we encounter with digital audio that is not found in analog systems is related to the time it takes to execute instructions. Even in complicated analog systems, computations occur in real time or instantaneously to human observers. Digital processes take varying amounts of time to complete, making parallel processes no longer synchronous.

When monitoring inputs through computer-based audio systems (software monitoring), there is a time lag between the sound input and the sound played back by the program. The delay is a function of the sample rate and buffer sizes chosen.

An alternative to this software monitoring is to monitor the analog inputs through a mixer at the input rather than through the software. Many systems include automatic delay compensation to resynchronize internal processes, but this function does not eliminate the delay we encounter from the A/D and D/A conversion processes.

When digital audio devices are connected, their clocking must be identical to maintain synchrony. Thus digital audio introduces a need for clock distribution that is not found in analog systems.

One of the advantages of software is the ability to refine and upgrade its performance over time. This ability can also be a drawback if the compatibility issues we encounter with continual updates continue to render code obsolete at a rapid rate. Not only the inherent performance of the software itself must be considered but also the interaction of the applications with the operating system used by the computer.

The operating system (OS), the code that determines the operation of the CPU and peripherals, is developed either by the company that makes the computer or by an outside company that provides the OS software. The recording application software may be written by programmers without advanced knowledge of the new developments in the OS. Keeping the OS and application software synchronized can therefore become a major issue.

A full-time studio technical staff often provided such maintenance in the past, but the personal computer–based studio is frequently the responsibility of a much smaller staff or simply the engineer alone. The engineer must therefore become a knowledgeable computer technician in order to keep computer-based recording systems working smoothly.

“The Science Of Sound Recording” by Jay Kadis, published by Focal Press (ISBN: 9780240821542), is available here. Purchase the book from and use discount code FOCAL30 at check out to get 30 percent off and free shipping in the U.S.

Return to articleReturn to article
The Science Of Sound Recording: Digital Recording