The View from 2005: It’s even worse now, with our high-rez digital madness. Reading specs and making sense of them is an essential sport if you wanna be good at audio. Read the first page of this carefully, even if you don’t read the rest. There are some important truths here. I stuck a little of this in Total Recording.

The Handbook for Sound Engineers lists no less than 36 fundamental quantities, from length, mass and time to magnetic flux and phase shift, that can be measured in an attempt to understand audio performance. Audio magazine tracks more than 85,000 specifications for some 5,000+ different products in its Annual Directory each October. The Cold War may be over, but here in audioland the numbers race is still on.

Readers of H&SR are all too familiar with the world of specifications and know well the bittersweet nature of those numbers as they both inform and confuse. Part of what makes specifications so confusing and so often misleading is that they are created and used to serve the interests of three very different groups of people: engineers, salespeople, and (lastly!) us consumer and end-user types.

Engineers (i.e. the people directly engaged with the design and manufacture of the product) need specifications as a tool to check the validity and quality of their design concepts and manufacturing methods. They are generally trying to measure very specific elements of the product's performance, as opposed to the overall 'awesomeness' of the product. So, engineering specifications can be difficult for us to use. First, we may need an engineering degree just to understand the meaning of the number. Second, and more importantly, it may take an engineer's knowledge of the product and the way the measurements were made to fully understand the context in which that measurement is meaningful. Too often we read a particular specification (for example: “Equivalent Input Noise [EIN]: -121 dBM”) and allow our imagination to create what we hope it means, a better product than a competing product with a higher (or is it lower?) value for the same specification. The engineer, without meaning to, has helped to mislead us when we misunderstand or mis-apply the specification.

Those in marketing and sales need specifications (extremely high or extremely low numbers please) to differentiate their product in the marketplace. "Buy this, it goes to eleven," they say. The audio electronics industry is fiercely competitive, technological progress keeps making things better and cheaper, and manufacturers are fighting for their market share and our attention. Promoting good specifications is an effective way to sell gear. An unfortunate fact of life is that we are swayed in our purchasing decisions by many peculiar, non-audio factors when we purchase our recording gear: lots of knobs and meters, piano-like action, cool retro look, and so on. Who can blame the marketing people for providing all those specifications, some of which may be utterly meaningless, when our $500 purchase might turn on something as irrelevant (to audio) as the color of the faceplate?

Finally, those of us who purchase and use any given audio product use specifications in a couple of ways. First, we use them as part of our purchase decision. Generally, we are looking for a sound value (nyuk, nyuk!). We want the best sound for the lowest price. Keep in mind that it's 1993, 200 years after the discovery of the electron, 40 years after the invention of the solid state transistor, and 10 years after the invention of the digital compact disc. These days, most audio products sound very good. So, to compare the sound quality of two products, we turn to the specifications (an exercise which generally involves the eyes, not the ears. Interesting . . . ). Specifications lead us to make very fine levels of distinction, such as the difference between .01% Transient Intermodulation Distortion (TIM, to you) and .001% TIM. Which would you prefer? Or does it really matter?

We then use the specifications in another way once we own the product: we embrace those numbers as we grow fond of our acquisition. "Man, you really learn to appreciate 16-bit sound,” we tell our friends. “True 16-bit just blows away 14-bit!" Later on, with the next piece of cutting edge audio gear we buy, we’ll say, “You know, once you’ve worked with 20-bit audio, 16-bit just doesn’t cut it.” Our enjoyment of the sounds we are creating is enhanced by the presence of these numbers. In the end, we use the numbers right along with the engineers and salespeople to make judgments about the capability and quality of the product and its implied effect on the resulting quality of our work.

An interesting dilemma arises here. There is always a difference between our subjective feeling about an audio characteristic and any measurable, quantifiable characteristic. Table 1 tries to compare some important subjective audio attributes to some quantifiable values - some or all of which may be familiar.

Table 1

Subjective Quality	Measurable Quantity
Loudness	Sound Pressure Level dB VU
Pitch	Fundamental Frequency
Timbre	Spectrum Frequency Response Overtones
Distortion	Total Harmonic Distortion Intermodulation Distortion
Hum and Hiss	Signal to Noise Ratio Noise Floor

Each subjective quality does not have a clear, directly comparable measurable quantity. This is a fundamental characteristic of audio specifications: any sensory quality that is to be measured can only be estimated through the quantification of different, somewhat related observed values, usually as reported by a range of test subjects (in response to directions such as “Mark which of the following sounds is exactly twice as loud as the reference sound.”).

See the Sidebar on Lord Kelvin