Note: The following two articles deal with an attempt to assess the validity of various words we use to describe audio. This is important to me because

1. I write for a living, and need words to have some sort of reasonably stable and general meaning,
2. the misapplication and appropriation of words for special use leads to exclusionary jargon and code words (which is broadly present in our industry and also an abuse of good language), and
3. we really NEED words that we can broadly and reliably use to describe sound. There aren’t very many.

Testing Microphones for Warmth, Brightness and Depth

Over the past two years I've gotten involved with an on-going study of microphones. This study began at Audio-Technica US about three years ago, where Jacqueline Green, VP of R&D, noticed that microphones with very similar measured specs often sounded quite different (sound familiar?). She initiated a study of microphone behaviors that attempted to correlate time domain measurements (waterfall plots of frequency response, in this case) with subjective listening impressions. The subjective tests were informal, and the results were interesting, but inconclusive.

A couple of bright young students at Worcester Polytechnic Institute, Eric Reuter and Ben Findlen, picked up the study at the suggestion of their mentor, acoustician Dick Campbell. First, they conducted some quite detailed lab measurements of a whole bunch of garden-variety microphones. Then they brought in Moulton Labs (that’s my son Mark and I) to help them develop some comparatively rigorous subjective test procedures to determine what correlations could be found between subjective impressions of these microphones and their physical behaviors, particularly in the time domain.

For these subjective tests, we made a set of recordings of a variety of musical instruments in both near and far fields with all of the microphones under test, and played back these recordings in AB pairs for a listening panel of approximately twenty reasonably expert audio listeners. We asked the listeners to compare the two microphones in each trial in terms of a given audio adjective - we chose the terms "warmth," "brightness" and "depth" for this test.

We turned the data over to my son the psychometrician (the real brains behind Moulton Labs) for analysis with the Rasch Model. After considerable data-munching and number-crunching, he came back with the insight that "warmth" and "brightness" really meant something to people, but that "depth" didn't. When Reuter and Findlen studied the relative scores of the mics for each of these terms and compared then with their physical measurements, they found some fairly significant correlations. All this was written up and presented to the Acoustical Society of America last October.

Do Audio Words Really MEAN Anything?

This has led me to think about a more basic question relating to this. When we talk about audio, we use a lot of really unrelated terms to describe what it is that we are hearing, like "warm," "bright," "edgy," "hard," "fat," "splashy," "brown," "fluffy" and so on. These are words that have no direct relationship to sound. So, I wonder, when we use these terms what do we really mean? Do any of us mean at all the same thing? Are these terms meaningful in terms of audio behavior, or are they simply code-words for "I like it," or "I think it sounds bad"? Are we just being linguistically sloppy, or can we really communicate something meaningful through the use of such terms? This turns out to be a fairly important question – surprisingly so, given how widespread and casual our usage of such words is.

Given also that we routinely use such subjective terms in the design process for new audio equipment, it is important to at least consider the question: what meanings do such terms have for both the professional and general populations? Are those meanings reliable and predictable? How do those meanings change as a function of audio experience, gender, age, economic status, etc.? Only after we know something about the general psychological meaning of these terms (or, in fact, if there are any such meanings), can we seriously begin to ask the question: "What are the correlations between such terms and physical and electronic behaviors of the audio devices?"

It's also worth considering what my son means when he says, "'Warmth' and 'brightness' have real meaning to people, but 'depth' doesn't seem to." How can he know this? What is he talking about?

When using the Rasch Model, testers need to be concerned about what they call the "unidimensionality" of the various items they are studying. Unidimensionality refers to the degree the test items refer to a single underlying construct (warmth, brightness, etc.) such that the objects measured (the microphones) unambiguously spread out on a scale without giving the statistical appearance of being in “two places at once.” The Rasch Model requires such unidimensionality in order to make sense of the data. It tests for it, first, by assessing the "unexpectedness" of each piece of data, second, by comparing the average error of each microphone's measure with the degree to which the microphones spread out on the scale, their standard deviation. A high standard deviation combined with a low average error for each microphone indicates that the test items and raters agree on the meaning of the underlying construct. Such was the case for "warmth" and "brightness," and that is why my son said that these constructs were more clearly defined while "depth" was more ambiguous.

Talking About Audio

What this means is that "warmth" and "brightness" had very consistent meanings for our hard-working listening panel. Assuming the panelists are representative of the audio community at large, we can comfortably assume that when we talk about "brightness," our colleagues are going to know what we are talking about. When we talk about "depth," they very well may not.

Words are tricky here. We don't have a large vocabulary to describe sound, nothing at all like the range of words we use to describe colors, for instance. So, we tend to make up words. What is fascinating to me is that some of those words turn out to have real universality among us, while others become jargon, more or less obscure and subject to private in-group meanings. Which is which, and how do we tell them apart?

It's funny I should ask. Following our original study, we undertook a second, larger one, using the same microphone recordings, but with a larger set of listeners and six subjective terms, not three (and, yes, we included "warmth" and "brightness" on our list). Next month, I'll tell you more about how all these terms came out, and a little bit about how the microphones fared.

Thanks for listening.


And so, the next month, I continued: