So why do we do it like that? Science explains audio practices.
By Jay Kadis
Practitioners of sound recording and mixing are always looking for new “tricks” that will make their recordings sound better. Many engineers describe their techniques in articles about how they work and accumulating ideas from such presentations is certainly helpful. But behind each technique, there is a scientific principle that explains why it works. The Science of Sound Recording gathers these basic scientific principles and explains how many common techniques work.
Let’s start with dynamic range compression. Compression is easily described, it is equivalent to an automated hand on the fader turning up soft sounds and turning down loud ones. Compression works because of a characteristic of our auditory system, whereby sounds of similar frequencies stimulate a single area of the cochlea that resonates at the particular frequency range. Multiple sounds that fall in the same frequency range stimulate the same region of the cochlea, which can send only a single output that integrates all the input in that range. Loud stimuli cause a rise in the threshold of audibility for the area in question, making lower amplitude sounds in the same frequency range less audible. This is known as the masking effect. By using dynamic range compression, the lower amplitude components are boosted relative to the louder ones and some then exceed the threshold of perception. Compression works to overcome the masking effect, whereby louder sounds hide softer ones. Psychoacoustics explains why compressing signals makes them sound fuller.
There are frequently questions about the relative merits of analog and digital recording technologies. Theory predicts that digital sampling is able to fully capture and reproduce any input signal. Experience tells us that this is not always achieved. To understand why, we can look at the physics of analog recording systems and contrast them with digital systems. We find that both approaches have imperfections, some due to theoretical limitations and some due to imperfect implementations. Both are complex physical systems with often-competing strengths and weaknesses.
Analog recording converts sound pressures to voltages and then into magnetic polarity changes on a moving magnetic medium. While the original signal is converted to proportional voltages and magnetizations, each step slightly alters the information content of the original signal. Overall, analog recording produces an output different from the input in characteristic ways that reflect the physics of converting between electrical and magnetic representations. Noise is added and the spectral content altered. The sound of analog recording has been accepted as the sound of rock music since so much of the recorded music we hear was recorded that way. We have grown to accept the sound of magnetic recording as part of the sound of popular music.
Those who work in recording studios know that what they hear in the studio is rarely exactly what they hear coming back from a tape recorder. Theoretically, digital recording should exhibit none of these faults since the signal is not converted to a magnetic form. The conversion from the input signal to a digitized representation involves a different set of compromises, though. Imperfect analog-to-digital conversion results in a different type of noise, one due to inaccurate measurement of very small signals. This noise can be worse-sounding than tape noise because it is correlated with the changing signal. To help reduce the noise from errors in conversion, wideband noise is added. Thus, digital recording also adds noise to the signal. This noise is far lower in amplitude than the noise resulting from magnetic tape and can be controlled by selecting special noise spectra, a technique known as dithering.
Overall, properly designed digital systems are capable of less signal alteration than analog magnetic recording systems. What we may be missing is some of the alteration magnetic tape imparts to recorded signals that sounds desirable. We may also hear shortcomings in analog-to-digital converters that contribute harshness to the signal while analog shortcomings have a more natural and therefore familiar sound. Good digital systems can be more nearly transparent but the best converters are quite expensive. Unfortunately, the cost and scarcity of analog tape may soon reduce our choices about what type of recording system we are able to use. Taking advantage of improving digital systems will make the transition somewhat less painful. Understanding the fundamental techniques employed in digital audio will also help make the transition to digital audio seem less mysterious.
Jay Kadis is the author of the just published book The Science of Sound Recording. Jay has played guitar since his school days, written and recorded original music, built studios and done research in psychoacoustics and music technology. He teaches sound recording at Stanford University’s Center for Computer Research in Music and Acoustics.