Earl looks at measurements from the point of view of the loudspeaker designer, rather than that of the marketing department. Most real-world measurements that give genuinely useful information will probably never be seen by us because they look awful compared to the oversmoothed on-axis curves that have been our steady diet for years now. Earl examines many different measurement techniques and topics.

Of enormous interest to me was the information on polar response, directivity index, and power response. As you will see, Wayne has been barking up the right tree all along with his emphasis on matching up the directivity indices. He's taking an acoustic solution, which is worlds better than relying on equalization alone because equalization cannot fix problems that are fundamentally acoustic.

In particular, take a look at Figure 4-6 and Figure 4-10. The first shows two very similar-looking on-axis response curves. The second is the power response curves for the same two speakers. As you will see, the problems introduced by inattention to directivity issues are huge, way beyond what a bit of digital EQ can begin to address.

Enjoy!

Duke

• Online chapters of Earl's book

Earl focuses more on measurements as a means to get enough useful information about what a loudspeaker is really doing, but he does give some interesting psychoacoustic information along the way regarding audibilty of THD and group delay. Apparently loudness level plays a significant role in the ear's detection of distortion. But Earl doesn't talk about things like how the skewing the frequency response or power response this way or that is perceived.

One of the concepts that was new to me is "Transfer Characteristic", or TC, which (if I understand) is basically all of the changes that a loudspeaker does to the input signal. It's very complicated and varies with both frequency and loudness level. Earl does talk about the relative audibility of distortions as they relate to the transfer characteristic. Apparently looking at (and understanding the psycohacoustic implications of) the transfer characteristic predicts that soft clipping is subjectively benign while hard clipping is quite objectionable, and that the very small glitch in Class A/B amplifier operation as the voltage swings through the zero point ("crossover distorton") is especially irritating.

Duke

Duke

Like don't waste your time with THD measurements - things like that.

Peter Aczel attempted such articles many times, and I think came quite close, though he pretty mcuh only cited his own & his associate's experiences, and his data was never published anywhere, except the audio critic of course. To hi credit, he is the only person I know who actually introduced scientists like Floyd Toole to the general audience.

thanks
-akhilesh

But to me, harmonic distortion is one of the most offensive commonly occuring anomalies created by audio reproduction equipment. The most offensive is response anomalies. Things like a complete loss of treble or a large peak in the midrange, that sort of thing. That bothers me the most. But clipping is definitely high on the "ugly" list, and it is characterized by exaggerated amounts of odd harmonics.

Certainly, low levels are imperceptible. I doubt anyone can hear a response anomaly of 0.1dB and few can hear 1dB, but most everyone can hear a 10dB swell or dip in a particular band. Midrange is easier to notice, just like the phon curve shows. Same must be true of distortion. Probably nobody can hear harmonics that are -50dB, but -10dB is another matter. And same as the phon curve, the band which harmonics fall in is important, as is their relation to the fundamental.

So what I'm saying is that I think reductions in harmonic distortion are some of the best improvements one can make in an audio system. The figure of "total harmonic distortion" might not be particularly relevant, but I am certain that moderate to high distortion levels are very offensive, and probably one of the worst things a component in an audio system can cause.