Ear Training

Back when I was in college, part of completing the music major degree was to log several hours of "ear training." This consisted of sitting in front of a computer while it played chords, melodies, and scales at you while you attempted to identify the notes you heard. Most people hated it. First of all, it was kind of boring just sitting there with some archaic Mac software listening to "beep-beep-bloop..." Mostly people didn't like ear training because it was really hard. It's one thing to listen to a chord or melody and comment, "That was pleasing," or, "That was dissonant," but it's another thing altogether to be able to critically discern, memorize, identify, and transcribe what you hear.

The things that make ear training so difficult and the progress so slow is that the human brain is so automatic in interpreting what we hear, and that sound is so transient. But what's important to realize is that it is possible to sharpen your listening skills, and that doing so can have benefits. As a composer, doing all those hours of ear training really helped me to become a better composer, allowing me to mentally "hear" music on a page before I actually played it, and also giving me the ability to know how a certain chord or melody should sound, in the event that someone plays it wrong. Unfortunately, all this "musical" ear training did very little to help me when I started designing speakers. Just because I could identify a wrong note in a chord did not mean that I could identify a problem with the sound of a speaker. I mean, sure I could tell you something wasn't right, but finding out specifically what the problem was was akin to bobbing for apples.

I wasn't alone in my inability to critically listen to sound, either. In fact, in my years involved in speaker building and various audiophilia, I was noticing that the vast majority of people don't really know how to identify when something sounds wrong with a system--other than "it doesn't sound right." But even that seems to be overlooked sometimes (ie: I've read many accounts of people who didn't realize they wired a driver in reverse polarity) The problem with this ignorance is that it leaves us as complete suckers to the power of suggestion--sometimes taking us way off-course, and costing us large sums of money ("This speaker sounds too hissy; maybe I need to buy more expensive capacitors." "The midrange sounds muffled; maybe I should buy expensive speaker wire.")

Recently, I was made aware that Sean Olive of Harman has developed a software program called "How to Listen" to help testers of audio equipment sharpen their hearing so that they can identify problems in a speaker or piece of audio gear. This was developed internally for Harman, but he claims it will be released to the publicly at some point. If this software were released, it would be a huge benefit to every audiophile. In the meantime, until Sean publishes his software, I will give you my [in progress] treatise on how to critically listen to sound.

Learn your ears

How well do you know your ears? Open a textbook and you're likely to read statements about the average frequency range of human hearing, or see a table of decibel levels. Open an audiophile magazine, and you're likely to read claims of audio fidelity way outside these ranges. How can we know which one to trust? Should we trust A) science or B) music lovers? How about C) find out for yourself what you can hear!

Frequency

The common rule of thumb is that humans can hear frequencies between 20 Hz and 20,000 Hz. How high can you hear?

http://audiocheck.net/audiotests_frequencycheckhigh.php

Kind of makes you wonder why marketers of audio equipment brag that their equipment has high frequency extension past 22,000 Hz? It should be noted that this "upper limit" is precisely why redbook CD audio is recorded at 44,100 Hz (see Nyquist Theorem). Of course, there are those who claim you can "feel" sound up in the ultrasonic range. Of this, I have no opinion. How did 21,000 Hz make you feel?

Amplitude

Perhaps you've seen a chart of decibel levels. What does this mean in terms of how you listen, and how much detail you can hear?

http://audiocheck.net/testtones_dynamic.php

For this test, its best to use an SPL meter. Use it to set the volume of the full-output noise to 85-90 dB. However, if you don't have one, then just set the volume to a volume that's "bordering on uncomfortable." Start with the -6 dB and work your way down. You may be surprised to find out that there is a limit to how quiet you can go until the sound is eclipsed by the ambient noise in your room. This is the "noise floor" of a room. I have found that most rooms can't get below -60 dB, even with all HVAC an appliances turned off. (If you really want to be shocked, try this test in moving car--let someone else do the driving, though!)

What's also worth mentioning about this test is how good the dynamic range is of 16-bit audio, which is capable of -96 dB. Can you hear -96 dB? If not, then why bother with 20- or 24-bit audio, which would be -120 and -144 dB, respectively?

Equal Loudness Contour

Within the range of our hearing, our ears are very sensitive in certain frequency spots, and almost deaf in others.

http://www.phys.unsw.edu.au/jw/hearing.html

After following the directions on this page, take a look at your answers. It will probably have a trend that looks something like this

What this means is that if someone played you a 1 KHz tone at 60 dB (which is mildly annoying, by the way), in order to hear a 20 Hz tone that sounded the same volume to your ears, it would have to be 40 dB (10,000 times) louder! Likewise, to hear a 10 KHz tone that sounded the same volume to your ears, it would need to be 10 dB louder. Why are we so sensitive in the "middle" and so deaf at the extremes? I dunno. My personal theory is that it's evolutionary; sounds between 400 Hz - 4,000 Hz include things like speech and rustling leaves, which are vitally important to survival. The telephone companies figured this out a long time ago, and completely eliminated all audio information outside of this range to save on bandwidth. Of course, it's not the last word in high fidelity, but it's still intelligible.