What Has Dave Been Up To Now?

About 18 years ago I began fooling around with the high-frequency dispersion characteristics of loudspeakers. As I grew more familiar with these characteristics, and had a chance to listen to loudspeakers with improved high-frequency dispersion, I came to understand a great deal more about how humans “localize” sounds and perceive sonic environments. I also gained some insights into the behavior of the phantom images that are at the center, literally(!), of our stereophonic listening experience.

This has led me to some new ideas about the loudspeaker/room interface, control room design and loudspeaker performance criteria. I’m going to share some of these with you here. But first, some background.

As wide-dispersion loudspeakers become commercially available (this is finally in the process of happening), it is appropriate to consider the implications of this behavior for both stereophony (and surround sound) music production, as well as for control room design.

What’s The Problem?

The dispersion of sounds from loudspeakers varies dramatically as a function of wavelength (and therefore, frequency). As a general rule, low frequencies (long wavelengths) are emitted omnidirectionally and high frequencies (short wavelengths) are emitted in a very narrow beam. Around crossover points, these behaviors are often wildly erratic, due to the fact that the sound is being emitted by two drivers at different points in space with differing dispersion characteristics.

This means that any speaker that has flat frequency response on axis MUST have declining power response as the frequency increases and the wavelength shortens. Much less acoustic power is emitted in a narrow beam of energy at 15 kHz. than is emitted hemispherically at 100 Hz. , when both of these frequencies have the same acoustic sound pressure level on axis as measured by a test microphone in an anechoic space.

We make the perceived power response of the loudspeaker even worse through our practice of using high-frequency absorbent material (such as Sonex, etc.) in our control rooms. This practice is based on a time-honored, if mistaken, belief that high-frequency room reflections obscure the details in recordings. Such treatments usually do little or nothing for low frequency reflections and the result is that we often work in playback environments whose reflected sound fields, by design, really degrade the already poor power response of the typical modern studio loudspeaker monitor.

We have come to accept this state of affairs as “the way it is,” and to accept this loudspeaker behavior as “the way loudspeakers sound.” However, in physical reality, other possibilities exist and loudspeakers can be made to have different behaviors that sound different, and interact with rooms differently.

Meanwhile, we face a difficult conundrum. Our recordings contain information about recorded sound sources and the space (real or artificial) that they were recorded in. At the same time, the room in which we play back the recording has its own ambient character (which, as we’ve noted, is often degraded). We would prefer to not have the details of the recording obscured or degraded by the ambience of the playback space. How can we do this?

There is a school of thought that says we would do best by doing all of our production work in an anechoic environment (i.e. one without surface reflections). In such conditions, power response becomes irrelevant (it is never even detected as such) and room reflections of the playback space have been eliminated.

Sounds ideal, right? Not so, in my experience. I’m one of a very few recording engineers who have actually done some serious critical listening to stereo in an anechoic chamber, and I am convinced that such environments are not appropriate for music production work, for a broad range of reasons. These range from cost to a sound quality that is incompatible with reverberant playback by end-users. See my book, Total Recording, for a more detailed discussion of this.

Over the years, I’ve come to the realization that there is a better way to do this, a way that makes productive use of the playback room and its early reflections. While this new way works best, of course, with both room and speakers optimized, it also works well with any reasonably-behaved speaker, and better and better as the power response of the speaker improves.

What happened is that I realized that humans have no problem at all integrating early reflections of a sound with the direct sound. Further, I found that the really important thing to do to enhance the stereo illusion is to enable us to connect the early room reflections in the playback room to the sounds of the recording, not with “the sound of” the playback room. This calls for two things: loudspeakers that emit “recorded information” consistently in all directions (or, at least, horizontally), and playback room surfaces that don’t alter that information upon reflection.

There are several ways to create loudspeakers that have suitable high-frequency horizontal dispersion – dispersion that yields approximately the same frequency response as on axis, and no more than 6 dB change in total level all the way to 90° off-axis, for 180° total horizontal dispersion. In my experience, ribbon tweeters (such as the Genelec S30 employs) or acoustic lenses such as I have been developing yield such dispersion.