Wayward Down Deep

The audio diaspora is split on the subject of bass. Some audiophiles—surely the majority—consider the reproduction of low frequencies purely in terms of the weight and drama it adds to sounds with significant bass content. Others—the generalists—take a much wider view of the significance of extended bass response, noting that an audio system's ubiquitous high-pass filters are unusual in Nature and suggesting that this is one of the factors that separate, at the fundamental level, live sound from its poorer reproduced cousin. When John Atkinson wrote on this subject more than 10 years ago (Stereophile, November 1995, "As We See It"), he quoted a memorable line by Kal Rubinson that encapsulates this latter view: "Something in Nature abhors a capacitor."

For once, this is not an audiophile concern that audio academicians regard as delusory. In fact, there is a fair amount of material in the audio engineering literature on this topic, although that's not to say that an industry consensus exists on it, or that it has received more than cursory attention in the realms of recording practice and product design. That's a pity, because we are now in a position to address this issue at reasonable cost through the use of digital signal processing—a technology that facilitates solutions in this area that are impossible or impractical to achieve using analog filters.

One of the most persuasive academic voices speaking on the importance of audio systems' bass behavior was the late, great Michael Gerzon, who had a unique opportunity to investigate it as part of the B&W room-correction project he joined in the late 1980s, and that would have culminated in the launch of one of the very first DSP room-correction products but for the fact that it was canned at the eleventh hour because of internal disagreement as to its efficacy. Peter Fryer, B&W's head of research, who had initiated the project, couldn't hear any difference when the room corrector was switched in. Gerzon clearly did.

The relevance of a DSP room corrector to the issue of loudspeaker bass performance in particular, and to the high-pass nature of reproduced sound in general, resides in the fact that the B&W device Gerzon helped design wasn't just a room corrector (though that is how it is best remembered): it was also a loudspeaker corrector. In fact, when I heard the prototype demonstrated at B&W's research center in Steyning, England—yes, it was previewed to the press before the plug was pulled—I considered its loudspeaker correction to have a much more obvious effect than its room correction.

In defiance of so many before and since who have declared the human ear "phase deaf," Gerzon and his colleagues designed the B&W equalizer so that it could cancel speaker phase distortion, including the distortion that results from the speaker's inherent bass rolloff. Writing about the project in an article for Studio Sound, Gerzon observed: "The subjective effect of phase compensation of the bass from loudspeakers is very marked, giving a much tighter and more 'punchy' quality, with greater transparency, and interestingly a subjective extension of bass response of at least half an octave. The improvement is audible even on loudspeakers with a very high cut-off frequency, such as Quad electrostatic designs. . . . The benefits of bass phase equalisation are considered, by those who have heard it, to be a substantial improvement over what was hitherto possible with analog technology, and digital equalisation provides a way of improving bass performance without going to ridiculously large giant space-consuming power-hungry monster speakers, and is certainly a much cheaper route." (footnote 1)

Gerzon and his coworkers went even further, intending to equip the B&W device with a control that could be adjusted by the user to cancel the phase distortion resulting from concatenated LF rolloffs within the recording and playback electronics. This equalizer would have to be set by ear for each recording, but again, Gerzon was adamant about the benefit: "Although most of these filters (particularly AC coupling filters) have only a very small effect on amplitude response within the useful audio band, they introduce a marked and very audible degradation of phase response, which may extend up to middle audio frequencies." (footnote 2). Peter Craven, who worked with Gerzon on the B&W project, still has a prototype of the room corrector that I had hoped to borrow in order to experiment with this bass phase equalizer—until I discovered that had never been implemented. Nevertheless, the intention was there.

Elsewhere in the UK audio industry, the significance of loudspeaker bass phase distortion had already been recognized and addressed, although without resort to the power of DSP. Four years before Gerzon began his work on the B&W project, Laurie Fincham—then at KEF—had presented a paper at an AES Convention investigating the significance of loudspeaker phase distortion at low frequencies and concluding that, under the right conditions, it is audible (footnote 3). To prove his point, Fincham gave a demonstration before an AES audience that involved introducing phase distortion into a loudspeaker whose bass response had been extended to infrasonic frequencies. An eight-cycle 40Hz toneburst was used as the test signal, phase-distorted using two second-order all-pass filter networks, each approximately equivalent in phase response to a reflex-loaded bass system. The point of applying this unrepresentative amount of phase distortion was to exaggerate what Fincham admitted was a "quite subtle" but nevertheless audible effect.

This was around the time, you may recall, that KEF introduced its KEF Universal Bass Equalizer (K-UBE), an analog filter circuit for extending bass response to very low frequencies and thereby, it was hoped, removing the effects of phase distortion from the audible frequency band. In his paper, Fincham also referred to the possibility of using digital filtering techniques to correct phase distortion while allowing the speaker's amplitude to roll off normally. This approach, however, wasn't tried, "due to the instrumental complexity involved"—a reference to the DSP technology of the time, whose limitations made such filtering hardware-intensive and impractically costly. A few years later B&W managed to make the digital option work, but it was still at the ragged edge of what was technically and economically feasible. Another 15 years on, that situation has changed dramatically: DSP is now cheap, and its use in audio products is burgeoning.

Question
Bass phase distortion and its acknowledged but poorly defined effects on sound quality have intrigued me for some time. For this article, I set myself the task of answering one simple question: If you assemble a loudspeaker system that meets the conventional requirement of maintaining its amplitude response down to the lowest audible frequencies, by how much can its sound quality be improved, using suitable test material, by correcting the phase distortion due to its LF rolloff?

To perform this comparison, the first requirement is for a speaker whose bass response is as flat and as extended as possible—one that is easily met using a quality powered subwoofer. Beyond this we have also to consider the influence of the room, and here things get complicated. There are two distinct effects to consider, both of which arise because, at low audio frequencies, the wavelength is long (about 56' or 17m at 20Hz): longer than the distance between the speaker and the room boundaries, and of the same order as the major room dimensions.



Footnote 1: Michael Gerzon, "Measure for Measure," Studio Sound, July 1992 (downloadable free, as "Digital Room Equalisation," from the Gerzon Archive on my website).

Footnote 2: Peter Craven and Michael Gerzon, "Practical Adaptive Room and Loudspeaker Equaliser for Hi-Fi Use," AES UK DSP Conference, August 1992 (available from www.aes.org).

Footnote 3: Laurie Fincham, "The Subjective Importance of Uniform Group Delay at Low Frequencies," Journal of the Audio Engineering Society, Vol.33 No.6, June 1985 (available from www.aes.org).

ARTICLE CONTENTS

X