This is the seventh edition of what has long been a standard work on the subject—the first edition appeared in 1977! The current edition has been brought thoroughly up to date. It includes, for example, such recent developments as the Devialet Speaker Active Matching (SAM) system. Over the years, the book has steadily expanded in size and scope. And the present edition includes not only the information on specifics of speaker design of the early editions but also surveys the whole subject of assessment of loudspeakers, both technically and in listening terms. The book has now grown to considerable size—more than six hundred pages—and it represents a summation of Martin Colloms’ long career as a speaker reviewer and design consultant. The fruits of extended experience and impressive expertise are very much on display here. This is an indispensable work for anyone who has a serious interest in speakers, who wants in particular to go (far) beyond the informal scuttlebutt of on-line forums and Internet reviewing.
A book of this nature is necessarily somewhat technical, and the general reader might wonder whether it would be readable to the point of being interesting and useful for a person without engineering training. The answer to that is definitely yes: The book will be valuable and interesting to anyone with real interest in loudspeakers, regardless of technical background. The mathematical level almost never goes beyond high school algebra, and one can in fact read the book with interest and comprehension of a high percentage of the content without taking in the mathematically expressed content at all. I would caution the reader, though: If you want to take in as much of what the book has to offer as possible, you should read it carefully.
You can get some idea of the startling scope of the book by looking at the table of contents (you can do this free of charge online in the “look inside” button on Amazon for example). Practically everything one might want to know about speakers, almost every topic one can imagine, is listed, everything from the history of the subject with which the book begins (truly fascinating it is, too), to the physical design of speaker enclosures and drivers, to crossover design and behavior, to every aspect of speaker testing and evaluation, It should be obvious just from the table of contents that you need and want this book if you are seriously interested in music heard through speakers and how it all works. And a look at the bibliography will show you how comprehensive is the literature that stands behind the book. One of the keys to learning a field in depth is to have a solid and extended list of references, and the book is invaluable for this, quite in addition to its own value.
Few people can compare with Colloms in the extent of his experience with loudspeakers. He has, he says, reviewed more than four hundred speakers over the years, and readers of his reviews for HiFi News and other publications will know that his reviews are detailed both in listening terms and in technical analysis and measurement. This book is the fruit of accumulated experience. And such a distillation of a lifetime of work in the field can be expected to be of great value—and is in fact just that. But valuable though it is, one cannot fail to notice certain limitations.
Long But Not Long Enough
The most fundamental limitation is that the book is too short, even though it is over 600 pages. So vast and at the same time so subtle is the subject that the book would have benefited from being a two volume work, or even being two books, In particular, the latter part of the book treats many of the topics concerning performance evaluation in too summary a way to do them justice. And some hugely important topics are hardly treated in detail at all. This difficulty is compounded by the fact that measurements are not provided to illustrate many topics that could be illuminated by measurement illustrations.
For all the large number of subjects covered, a number of really important experiments and studies are omitted. There is a reference to the work of Jorma Salmi and Anders Weckstrom on models of sound perception in-room and their startling acoustic-bypass experiment. But there is no extended discussion, even though this is surely one of the most surprising and significant of all audio experiments. (In this experiment it was observed that a flat-response loudspeaker had an output almost audibly indistinguishable from its input, if the acoustic output were recorded anechoically. Also almost ignored is the work of Soren Beck, O. Juhl Pedersen, K. Rasmussen et.al., in the Archimedes Eureka project. The name Beck does not even appear in the index! (The indexing is quite inadequate. Consequently it is really hard to find things. For example, the name Linn appears only once in the index referring to page 22, but an important if brief discussion of the Linn Exact speakers occurs on page 537—with no index reference. It is really hard to be sure exactly what is in the book without looking at every paragraph, and it is equally hard to find an item one knows is there if one does not recall exactly where it is.)
Also omitted are Siegfried Linkwitz’s observations about diffuse field response versus frontal arrival response and the meaning of this for speakers (though there are other references to Linkwitz’s work); Gunther Theile’s work on how the ear/brain processes speaker sound in stereo (the name Theile does not appear in the index); the exact relationship in perceived frequency response between a phantom centered image and the perceived response from an actual centered source (the details of this are known but not presented here); the work from Lyngdorf Audio on corner woofers combined with free-space mounted satellites using digital delays; the multiple subwoofer idea promulgated by Earl Geddes and realized in concrete form by D. LeJeune’s Audio Kinesis Swarm; Roy Allison’s fundamental results on radiation loading by room boundaries, which is referred to briefly but not explained in detail; and many other things of vital significance. The list could go on and on. Much is presented but far too much vital information is missing.
Not Enough Measurements
An example is the discussion of digital signal processing (DSP) for correction of speakers. This is a truly major development and gets fairly extended treatment in the book—in words. But after noting that the possibility exists of canceling narrow, high-Q resonances to an extent completely beyond analog devices, as well as adjusting broad-band balance, the book provides no measurements of the results of this procedure. If this correction process works well, as in my experience it does, it is revolutionary indeed. If it does not, one should say so and offer evidence. Measurements should be offered, one way or the other. Such measurements are available elsewhere. The Essex device for this purpose appeared more than twenty years ago and has been thoroughly documented. The omission altogether of the measured results of DSP correction of various kinds is particularly odd and regrettable in that elsewhere in the book importance is attached, for example, to phase behavior, which can be and is being corrected by DSP, in the Essex device in particular.
A similar lack of measurement illustrations applies to DSP room correction, another truly major development. Again, this is revolutionary. But while it is discussed in a somewhat noncommittal way, no measurements are provided of the results—in spite of the book’s (correct) concern elsewhere with the room/speaker interaction.
Similarly, horn speakers are described anecdotally as having superior perceived dynamics, at least to many listeners. But of all the things that are easily measurably, dynamic correctness surely ranks near the top. Measuring levels is basic. But no measurements of dynamic behavior of horns versus other types of speakers are offered. This is a question that could be answered, but it is not.
Much of the book, especially in the latter parts, reads as if it were descriptions of what would be provided in the way of concrete information if in fact concrete information had been provided. This is seriously unsatisfying.
Omissions of this type go on and on. Assertions are made, sometimes quite surprising ones, but are not supported nor illustrated by measurements or even technical explanations in words. (Example: On page 603, there is a mention of the “lack of clarity of many amplifiers.” Oh really? How about a specific instance or two and some technical demonstration that this lack of clarity actually exists?)
Lack of Coherence and Connectivity.
There is also a strangely granularity to the book, which makes it convenient to dip into (except for the inadequate indexing, which cuts into that), but robs it of coherence. The material is subdivided but often not really connected together later. For example, Colloms rightly gives considerable attention to the Quad ESL-63 and provides measurement data on its performance. And elsewhere, he talks about L. Fiedler’s investigation of the use of 1/3 octave (analog) equalization to correct speaker errors, and he also discusses, as noted, DSP correction. (Fiedler makes it into the index—but without his first name or initial. Didn’t anyone copy edit this book?) But there is no discussion at all of how the response irregularities of the Quads—and there are some of audible significance—might respond to such corrections. In practice, even 1/3 octave EQ can improve the sound of the Quads a lot, as I know from long experience, and DSP correction even more so. Measurements of the corrected speaker compared to the uncorrected would be interesting, but no such measurements are provided for the Quads nor, comes to that, for any other speaker before and after correction, whether analog or digital. This is typical: Little or no connection is made among topics even when they are closely related, at least potentially. There are many grains of sand, but there is little beach.
Lack of Decisiveness
Much of the latter part of the book on speaker assessment and speaker performance in audible terms, which will be for many readers the most interesting part, seems to be characterized by a curious indecisiveness, in good part because all the correction possibilities fail to be investigated. The possibilities of varying things about a given speaker are mentioned without showing what the audible significance of these various things would be. The reader is left wondering what really counts.
What is really audibly important? For a specific example, Colloms talks about phase linearity, both in the bass and in the higher frequencies. How important is it? One can actually find out in listening terms because one can linearize phase via DSP without changing frequency response. But no report on what happens when you do this is offered, and the reader is left in doubt about exactly how important this matter of phase linearity really is in the higher frequencies. And the importance in the bass, while discussed, is not very well illustrated by measurements, It is claimed to influence—and in fact does influence—the perception of bass transients and what has come to be called “timing,” but sufficient concrete measurement data is not provided, even though alteration of bass timing is on the face of it something that would be easily observed in measurement terms. Phase linearity overall is praised in one place in a discussion of the Linn Exact speakers, but in section 10.5.13 it is downgraded in importance (“assuming that the result is worth the effort”), and incidentally the main difficulty of obtaining it in analog speakers is identified with the need for time alignment of the drivers, whereas in fact the major difficulty is the need to use first-order crossovers, which puts extreme demands on the drivers. (The Dunlavy and Thiel speakers elegantly solved the driver alignment problems. This is not really the main issue here. Curiously, neither Dunlavy nor Thiel, whose designs were surely the major exemplars of phase linearity in multi-driver speakers in modern times, is anywhere mentioned.)
Colloms again quite rightly objects strongly to manufacturers not supplying consumers with frequency response measurements. This is indeed something of a disgrace and embarrassment to the industry. Consumers should object. And he also talks about how important subtle changes in response can be and often are. But he gives few concrete examples, and when he is treating, for example, the Wilson WAMM Master Chronosonic, he gives no measurements, neither his own nor anyone else’s. And his discussion of the importance of time of arrival and time synchronization of drivers is not supported by any measurements here, even though it is quite easy to measure. The right hand seems not to care what the left hand is doing. (Many of the discussions of specific speakers do include measurements but not in this case).
Lack of Consistency
Then there is the question of consistency. Colloms emphasizes the importance in principle of small changes in frequency response, and quite rightly so in my view. But when he gets to specific speakers, he seems unconcerned about such errors. As mentioned, in his discussion of the Quad ESL 63, he rides right over the fact that the measurements shown indicate that 9kHz is 4dB elevated over the entire 1–2kHz octave in on-axis response, for instance. And when he gets to the Wilson Chronosonic he does not discuss response at all. These are just two examples of the lack of agreement among the sections of the book and indeed the lack of any apparent attempt to reconcile the sections.
After a while, one can get very frustrated by the absence of concrete and detailed data and discussion of the meaning of the data when data are supplied, and even more frustrated by the lack of reconciliation of different views. General assertions about “sound quality” are no substitute for precise criteria and the reader is constantly left wondering about which things really count. Among many audible things, which are important? This fluctuates from place to place in the book.
There are also moments at which the book degenerates completely into what can only be described as improvised nonsense. Consider the discussion of speaker cables. Absolutely no concrete data is presented. In particular, there is no discussion at all of the changes in frequency response that cables can introduce. (Readers with long memory will recall the denunciation by Peter Aczel in The Audio Critic of Colloms’ failure to determine the frequency response behavior of the cables tested in a survey in HiFi News, where, according to Aczel, the leading cable in fact induced large response errors.) Nor is there any solid discussion of what various aspects, other than resistance, inductance, and capacitance, of cable design can be shown to do to the sound in technical terms.
Instead, we have the assertion that in high performance systems, spending 20% of the cost of the speakers on cables is a sensible goal. This is in the complete absence of any indication at all of exactly what aspects of cable design would justify or require high cost. The reader is left wondering what, if one bought speakers costing, say, $80,000 a pair, exactly a cable manufacturer could offer for the $16,000 the estimate would generate for cable purchase. Apparently the reader is expected to believe that “you get what you pay for,” and that high-end manufacturers have checked on the value and validity of their products in private, even in the absence of any concrete public technical evidence at all. This 20% assertion is a very far cry from being logically established, and the automatic identification of perceived sonic quality with price is not given any justification. Whatever one thinks about the importance of cables, it is illogical to express the whole issue in terms of price percentages.
This book contains much valuable material. And anyone serious about the subject needs to have it on hand and to read it carefully. But it falls short of being the great work it could have been. It could have been the book that audio really needs, the book that synthesized the vastly extended research that has been done on speakers and how we hear them into a coherent whole that would connect our listening experiences with technical behavior in a way that would have been invaluable for the future of speaker design and for purchasers of speakers as well. Instead, as it goes along it becomes more and more scatter-shot, and the reader becomes less and less sure of what he ought to believe. And there are more and more appeals to anecdotal descriptions of what people claim to hear instead of things that are concretely checked—and this is true even when those things could in fact be concretely checked.
Anyone writing seriously about speakers has to face the fact that standards of correct behavior are not really complete. While low distortion and reasonable flatness in some sense are agreed upon generally, the questions of radiation patterns and power response (total energy radiated into the room as a function of frequency) are not agreed upon. Every possibility from hyper-directionality to omni radiation has its advocates. All one can really expect from a theoretical analysis is information on how the different choices sound in practice, Similar ambiguities arise in terms of the desired properties of listening rooms. But information on what happens audibly for different technical choices is the heart of the matter.
For a book on speakers to be really useful, such information on how the various possibilities will sound and how they will turn out in measurement terms is absolutely needed. And the more specific this information is and the more it is backed up by measurements, the better.
For all the information provided, this book falls short of the goal. As one approaches the end of the book, hoping for some synthesis of all that has gone before, there is a discontinuity. The author apparently suddenly realizes that the previous six hundred pages have failed to treat in anything like enough detail the questions of power response and radiation patterns, which are in fact enormous determining factors of speaker sound, Indeed, in a world where on-axis response is easily altered by penalty-free DSP EQ, the response into the room is arguably the most crucial matter altogether. But time seems to have run out.
A whole chapter on radiation into the room ought to have been provided. But instead, starting in Section 10.6, one is given a largely anecdotal and often misleadingly quick summary. Take for instance the question of wide baffle versus narrow front speakers. The possible advantages of wider fronts are simply dismissed without discussion. These do exist: The baffle step can be moved down far enough to be below the midrange, so that possible coloration of the mids from this source is prevented, avoiding the discontinuities of narrow fronts which have baffle steps around 500Hz generally.
It just will not do to dismiss this sort of thing on no apparent grounds other than the popularity of narrow front floorstanders, a popularity that is based in good part on visual fashion. The book’s treatment here has degenerated into being superficial as well as being severely misleading, There are also strange and ambiguous claims that the ear hears the on-axis response of the speaker itself in the higher frequencies. This is to some extent true, but this important point is not supported by evidence nor made quantitative. In fact, the part of the book from 10.6 onward, rather than being a grand synthesis, collapses by the end in a final despairing statement in which all systematic technical information is subsumed by the listening impressions of some group of qualified reviewers, though qualified on what basis is not specified, listening to source material of high “sound quality,” though how that is to be determined is unclear, in rooms with correct acoustics, though that too is unspecified in any meaningful and definitive way.
By the end, for all the specifics that have gone before, we are left in the position of Omar Khayyam, who “heard great argument about it…but evermore came out by the same door as I came in”
Perhaps the eighth edition will be the book that straightens out the tangle that contemporary speaker design and evaluation has become and establishes solid and precise connections between technical behavior and what is heard, and what is audibly correct, in some precisely specified way, or perhaps in several different but precisely specified ways, among which one might choose on the basis of personal preference. The present seventh edition is a start, but there is a very long way to go.
High Performance Loudspeakers: Optimizing High Performance Loudspeaker Systems
Authors: Martin Colloms. Paul Darlington (Chapter 3)
Publisher: John Wiley and Sons
Publication date, this edition: 2018