Semantic Compositions

Hi-fi, hi-ho, to the Apple Store we go

Eric Bakovic wrote earlier today on the Apple iPod Hi-Fi, a gadget he might like to buy ([Between this and the title, that's two awful rhymes. Stop now. -- ed.]). In the course of his post, he mentioned some language in a review he read which left him bemused. SC was so determined to give a useful reply (and also had admittedly been curious about the product himself) that he ran out to an Apple store within hours of reading Prof. Bakovic's post, just so he could give a useful reply. But first, let's address the text in question:

So what do you need to spend all that money for? Sound is one thing: The IPod Hi-Fi does sound, well, like a high-fidelity unit. The sound can fill a room or even overfill it, if you crank up the volume sufficiently. It is rich, the bass is deep and the treble trills quite nicely. Some who've heard my evaluation unit complained about a lack of "midrange" sound; I didn't notice any.

Prof. B's follow-up:

Did the reviewer not notice any "midrange" sound, or did he not notice any lack of "midrange" sound? This is a critical matter about which we're left guessing, though presumably it's the latter, or else he would have presumably noted that he agreed with the "[s]ome who've heard [his] evaluation unit".

(But really, can we trust this reviewer to really hear anything? Why the scare-quotes around "midrange", but not around "bass" or "treble" -- are the latter somehow more discrete? Also, describing the bass as "deep" is hardly informative, and I don't know that I've ever thought of treble in terms of "trills". But anyway.)

While SC has lately been starving for material -- and more to the point, time -- this is a subject about which he's reasonably well-informed, and so while Mark Liberman ably handled his other linguistic concerns with the article, your host will tackle this one.

About Prof. Bakovic's first question, regarding the ambiguity over whether the reviewer was commenting on a "lack of midrange", or the lack of a lack of same, the answer is that the reviewer meant that he did not consider the device to suffer from a lack of midrange. A standard audio writing trope in praising a product is to offer some criticism alleged by anonymous friends or neighbors, and then to dismiss it, or at least use it as a foil.

Don't believe me? Here's the relevant passage from a Bob Reina review of a Sonic Frontiers power amplifier:

A reviewer from another magazine gave me an idea. Said journalist was visiting my summer place for the weekend, and we were hanging out listening to obscure alternative music. Of course, I am bound by the Stereophile Prime Directive not to reveal to anyone my views of equipment under review. (JA installs this little chip under the scalp when you join up. It doesn't really hurt, and, hell, Uncle Larry pays for the surgical procedure.) But my friend was free to say what he pleased; he wasn't reviewing the amplifier.

My friend was not impressed. "Yeah, it sounds like a Sonic Frontiers amp. It's boring!"

That's the word I was searching for. I love the Power 2 because it's...boring!

And a John Marks review of a Brinkmann amp:

A music-savvy friend who stopped by just after the Brinkmann arrived was taken aback at how different it sounded from the Unison Research, yet was reluctant to declare either "the better." "They're just different," he mused.

...

Back to our apples and oranges of low-powered tubes vs mid-powered solid-state. The Brinkmann has greater clarity and gives a more vibrant sense of immediacy than the Unison S2K—the music has more solidity—but also costs about twice as much. Sigh.

Occasionally, this works  in reverse, where the friend is the one whose big change of opinion is the news, as evidenced in the first few paragraphs of this recent Tom Gillette piece for Stereophile.  What, you notice that it's signed "Sam Tellig"? Mr. Gillette thinks his most-of-his-last-name-backwards pseudonym adds character to his writing; SC strenuously disagrees. He also digresses. ([And laughs his evil laugh, in homage to Mr. Gillette's favorite cliche? -- ed.])

Having established what the reviewer meant, let's decode the parts that led to Prof. B's parenthetical comment, starting with "can we trust this reviewer to really hear anything"? SC has never read a review by the author in question before, but the general answer is: yes, but not overly so. Audio writers are generally just well-heeled hobbyists; the reviewers linked above have made no secret of their personal wealth in past writings, and while many readers may recognize the name Anthony Cordesman from his national security commentary on CNN, they may not be aware that he also has passed as an expert reviewer for many years. (SC is not wholly unbiased here -- he blames Mr. Cordesman for an overly enthusiastic review that led him to the one truly regrettable audio purchase of his life.) There is rarely any special reason to believe the reviewer has better hearing than you do -- although this does not mean that they haven't acquired some meaningful experience from years of listening to better equipment. Certain comments correlate fairly well with measurements, and we'll discuss that in a minute.

But before we do, let's get to the last of Prof. B's questions, about the scare quotes around the word "midrange". The only reason they are there, but not around bass or treble is to again set off the anonymous foil's comments for disimissal. They are in no way "less discrete" (Prof. B's charitable hypothesis), and the stuff about "deep bass" and "treble trills" is just more stock jargon which means nothing more than "I didn't hear anything grossly unacceptable in those regions". Prof. Bakovic has probably never thought of midrange sounds as "lush",  especially when more or less flat frequency response is being described, but that's another example of the sort of descriptive language involved. SC has written about this sort of thing before, so go click on the link if you want more such verbiage.

With such contempt for the audiophile press, why the heck does SC subscribe to any of it? (Specifically: Stereophile, Home Theater, and Robb Report Home Entertainment.) The answer, alas, is almost as pathetic as the timeless excuse for embarrassed Playboy subscribers: I read Stereophile for the measurements. (And the others for the product announcements and gorgeous photography.)

Having said all that, your host also mentioned actually auditioning the thing. This is true, with the following caveats: 1) the store was not otherwise quiet (although nothing else was being played nearly as loudly in close proximity), 2) the recordings used for the demonstration were not familiar, and 3) while SC's test gear is reasonably portable, there is no way the store would have tolerated his setting up a calibrated microphone, stand, preamp, laptop with 1/24th-octave analyzer and other such equipment to do measurements, so all claims to measurements are based purely on your host's experience and knowledge of the field.

As your host commented on Prof. B's personal blog, he agrees with the "doesn't lack midrange" comment. Listening to both some generic pop and an unfamiliar "Concerto in D" boasting Yitzhak Perlman (according to the display of the iPod docked to the Hi-Fi unit), I didn't hear any obvious midrange anomalies. But that's where the good news ends.

Apple specs the device as having a frequency response of 53 Hz to 16 kHz, +/- 3 dB. That sounds pretty good, but there's a lot of room for mischief within that 6 dB variation (see this Floyd Toole presentation for details; as he puts it, that spec can mean anything "from junk to jewels"). It sounded to your host's ears like the speaker is tuned to have a fairly severe hump at its probable port-tuning frequency of about 60 Hz, a common trick to boost apparent bass response in small speakers. The wireframe drawing on Apple's site suggests that the two ports are actually part of a single folded duct, about the only way to get that low a tuning frequency with any kind of useful resonance in such a small box. Such systems tend to suffer from "organ-pipe resonance", a tendency to be loudest at or near their tuning frequency, and the Hi-Fi sounded to SC's ears like it is no more immune to the physics of these things than any other speaker. When a reviewer says something sounds "boomy", a sharp bass-region peak is almost always the culprit.

As Dr. Bose has known (and successfully traded on) for years, in addition to this pseudo-bass, peaky treble sounds exciting on first listen. While the company has prevented reviewers from printing response measurements for years, this graph reproduced from a rare exception in a 1999 Sound & Vision article illustrates the basic idea pretty well. A significant peak around 3-5 kHz generally results in people claiming that something sounds "brighter" or "sharper" compared to a flatter-measuring speaker at the same frequencies. This sells well, especially when the audition period isn't long enough for fatigue to set in, and the Apple Hi-Fi sounded to SC like it was tuned for similar response. This comment is restricted to the peaky treble issue, though -- the ugly midrange suckout in the graph, due to the poor matching between Bose woofers and cubes, is most definitely not present in the Hi-Fi, or in just about any other similar product (including the Bose SoundDock).

Regarding the purported "room-filling sound", SC estimates he listened to the device at about 90 dB from six feet away. It sounded like the amplifier was beginning to clip (generally defined as 1% total harmonic distortion), despite Apple's claims of a "maximum sound level" of 108 dB when running the device off of AC power. SC would guess that figure includes about 10% distortion for the amp -- which is less audible than it sounds like, albeit not good (but typical for the category). As for the speakers themselves, SC would guess that if you actually drove them to 108 dB, you'd be facing 20-30% distortion from them at the ~60 Hz bottom end, and about 5-10% wideband (also typical of the category), and would probably back off the volume  pretty quickly.  The claimed 108 dB figure is even more suspect in light of the fact that a good-quality pair of speakers producing 88 dB with one watt will hit 108 dB with a 100 watt input, assuming no amplifier clipping. SC doubts that more than 5 watts are going to those speakers before clipping sets in, and also doubts that Apple has produced breakthrough drivers with 100 dB sensitivity, which would have to be the case in order to manage that peak SPL with less than horrendous distortion. That said, the real application of devices like this -- or the Bose Wave radio, or the Boston Acoustics Recepter, or any other way-expensive table radio replacement -- is to provide background music at maybe 75 dB as much as 10-15 feet away, which any of them will do an adequate job of.

As for Apple's claims of an "enveloping soundstage", they're nonsense, but equally so far all speaker docks -- in order for the stereophonic imaging effect to work, there needs to be meaningful channel separation at your head, which simply cannot happen when said head's ears are spaced at just about the same distance as the left and right channels are from each other. Interference between the channels will occur starting at points in space roughly equal to the distance between the centers of the midrange drivers (assuming that each one is receiving just one channel, and not a summation of both), guaranteeing that at any likely real-world listening distance, you'll hear everything in the same glorious monaural fidelity that Thomas Edison achieved in the 1890s (albeit with better frequency response and lower noise). Physics is just disappointing like that.

Having said all that, SC's opinion is also "a pox on this product category", just as he is also dismissive of all but the very cheapest desktop PC speakers as a waste of money. While the integration features of speaker docks shouldn't be overlooked, as actual sound reproduction devices, the lot of them all suffer from the same basic acoustic flaws. For the same $350, a listener desiring to maximize sound quality could also buy a pair of Infinity Primus 160s, NHT SuperZeros, or Paradigm Atoms (any of which should run him about $200/pair), and pair them with a cheap receiver (with a little shopping, a bottom-end Sony, Sherwood or TEAC could be found for $100-120) and a $10 headphone-jack-to-RCA adapter cable, and get far better sound. The response would be much flatter (especially when not sitting in a straight line in front of the thing -- such large-diameter midrange cones will have awful off-axis behavior above about 4 kHz, a number calculable from their diameter), the sound would be far less distorted, portability isn't really an issue if you're considering a box of the Hi-Fi's size, and he'd get a tuner essentially for free, not to mention a remote (and also a huge upgrade for his TV sound). Get a long enough adapter cable for the iPod's headphone output and he could even have the iPod sit with him on his couch. That's SC's 2 cents, but ultimately, it's Prof. Bakovic's $350, and therefore his call.

What's a magazine worth?

Just a few days ago, your host discussed the business of selling very expensive magazines. Now we're going to talk about the other end of the business, where they'll practically give you the magazine for the cost of postage.

A number of years ago, SC suffered one of the great heartbreaking moments of his life when Audio magazine closed shop. The year was 1999, and in December, a postcard arrived indicating that the publisher, Hachette Fillipachi, had seen fit to discontinue it in favor of transitioning all subscribers to the wholly worthless advertising forum known as Stereo Review, later known as Sound and Vision.

Audio was a magazine full of serious (albeit popularizing) articles about the recording process, electronics design, and real musicology. Its reviews frequently included meaningful discussion of the circuit topologies or acoustic design principles behind the components being discussed, and to read the magazine was to get a real education in a wide variety of subjects relating to music reproduction. It covered the whole range of electronics as well, from the cheapest budget gear to the most exotic, built-only-by-commission equipment. Stereo Review was a magazine of little intellectual heft, which would accede to any demand from an advertiser, and every issue offered the exact same lineup of articles: Budget components! How to hook up color-coded wires! A comparison test of three components which barely perform the same functions! While the point about the price range of gear may seem like mere snobbery, this choice was not merely a matter of appealing to a wider demographic, but of restricting editorial content to the sort of promotional buyer's guide material that will offend no potential mass-market advertiser. Practically every review in Sound and Vision begins with the familiar litany of "Manufacturer X makes the interesting flagship product Y, which costs in the kilobucks. Instead, today we'll be talking about their entry-level product Z, which is essentially indistinguishable from every other product on the shelves at Circuit City." When it so happens that Z suffers from egregiously bad design or manufacturing, and will display measurements so grossly out of line with conventionally-accepted performance parameters as to be unrecommendable (read: Bose), Sound and Vision simply labels the review a "User Report" instead of an "Equipment Report", and then says whatever the advertiser has paid for. This doesn't tell the consumer anything about why they should or shouldn't buy something; it just puts a third-party veneer on the manufacturers' advertising.

Since purely promotional material is worth far more to advertisers than to readers, Sound and Vision is practically given away to subscribers, for the nominal charge of about $0.83/issue, despite a cover price of $3.95. Audio did not discount nearly as much, and cost something like $2/issue for subscribers, and around $5 for a newsstand copy. SC had a three-year subscription to Audio at the time of its cancellation, and so Hachette Filipacchi converted it into an eight-year subscription to Sound and Vision. Using the perverse logic that your host had received a special nonrefundable discount rate that he never asked for, HF then declined to refund SC's subscription payment, and so he has been a reluctant subscriber to Sound and Vision ever since. He is eagerly awaiting 2008, when the nightmare will finally end. Despite pledges from the editorial staff at the time to expand coverage in order to accomodate all the new Audio readers, the mix of reviews -- and the seriousness of the coverage -- has never elevated at all from the lowest-common-denominator, anti-intellectual editorial content of the old Stereo Review.

The practice of offering such heavy discounts to subscribers makes it clear that subscription revenue is not at all the primary source of profitability for magazines of this ilk. Much the same is true of other hobbyist magazines; Car and Driver asks only $1/issue, Motor Trend also wants only $0.83, Popular Photography wants $1, and Home Theater has the nerve to actually ask for $1.08/issue. At these rates, it's painfully obvious that cover prices have little or nothing to do with an accurate representation of what it costs to run magazines.

Of course, it's not necessarily the case that a magazine is hopelessly editorially compromised just because it makes more from advertising than subscriptions. Car and Driver in particular is unafraid to write harsh reviews like this panning of the Ford Crown Victoria and Pontiac Bonneville by comparison to the new Chrysler 300. But such reviews just don't come along all that often in the audio press. More to the point, when you can't charge for a magazine, you can't devote many pages to actually interesting material, like interviews with musicians other than whoever the biggest labels are currently pushing, or discussions of the science involved. Unpaid articles don't pay the bills, after all.

It's not as though some of these laments haven't appeared here before. What is new, though, is the first sign of brutal honesty about the for-sale nature of editorial content in audio hobbyist magazines. Last week, Stereophile Ultimate AV (formerly Guide to Home Theater; henceforth UAV) -- by no means the worst offender in the consumer AV press, and one of the few to still publish more than the very most perfunctory measurements -- announced that it would be discontinuing its print operations, and becoming entirely a web-based operation. Primedia, the publisher, is doing no such thing with their other magazines in the same general topic area, like Home Theater, Stereophile, or the preposterously-named "Connected Guide to the Digital Home". Of the lot, Home Theater and UAV cover nearly identical beats, with much of the same range of gear (if not the same specific pieces), but Home Theater's measurements are always less comprehensive (this is not open to debate; compare a typical HT speaker measurement section with one from UAV), and their writers much less willing to make negative comments (this is admittedly a more subjective judgment). Both attract the same advertisers, in roughly the same proportions, as can be seen by flipping through issues of both. It would appear to make sense for a publisher that controls both to rationalize their operations by only producing one -- but the one they chose to save is the one which is far less editorially valuable to its readers. What can we conclude from this?

It isn't quite right to say that magazine content is valueless just because it is very cheap. After all, no listener pays for radio traffic reports, which many people find useful. And it's not as though that information is then biased to suit the agendas of advertisers who pay for it -- it would be absolutely nonsensical to contemplate the idea of advertisers paying traffic reporters to lie and suggest that only the roads leading to their businesses were relatively traffic-free. Nevertheless, it's clear from the practice of essentially giving away the magazines that both the editorial staff and the magazines' real clientele -- the advertisers -- believe the editorial content to be have little value apart from being a delivery mechanism for the ads. Looked at in that light, the highest return on investment is going to come from the magazine least likely to produce negative PR, and so it's no great shock that UAV is going while Home Theater stays. It is, however, a depressing loss as the popular audio press races to the bottom. SC would be greatly surprised to hear that the situation is much different in any other enthusiast media.

Semantic Wars

There's still more coming on George Lakoff, folks, but it's taking longer than expected to write ([especially when SC has the nerve to take a vacation over a three-day weekend! -- ed.]). In order to avoid coming off like an obssessed lunatic, your host was going to make you wait until this afternoon, or even Wednesday morning, for significant new linguistic content. Anything to avoid even more ranting about Emperor Darth George Lucas. But Radagast suggested the post title, in honor of what are apparently the two tracks making up the SC mind, and so I can't resist. Without further ado:

In the just-arrived November edition of Home Theater, there's an interview with Mark Hamill, known primarily to the rest of the world as Luke Skywalker. If you know him by anything else, SC pities you for the part of your lifetime which has been irretrievably wasted. Mr. Hamill seems to have a healthy attitude about this fact; he's quoted as saying "I'm not someone who really enjoys going back and watching my own stuff". In other words, he's made peace with what Star Wars was. And like SC, he wishes that someone else would, too. Reproduced without further comment, SC's very most favorite (read: damning) parts of the Q&A:

Q: The power of home video! Has [The Empire Strikes Back] director Irvin Kershner been a good sport about the changes?

A: Remember the old saying, "It's good to be the king!"? I guess with George, "It's good to be the Emperor!" If he wants to make them into musical comedies, that's his choice.

Q: And Return of the Jedi?

A: With Jedi, I was a bit disappointed because I said "Gee, it's all so pat and tied up neatly in a bunch." I voiced this opinion to George and was hoping that we'd be able to even top Empire. George explained to me, "Remember, this is meant to be a film for children."

An extended version of the interview can be found here.

Your host recognizes that it's quite easy these days to mistake this blog for being an Orwellian Hate Week directed at people named George. It's not true, it really isn't. But oh, SC will never convince anyone of it.

I want my movies back

About 10 days ago, a rant on the new DVD version of the original Star Wars trilogy appeared in this space. However, willpower not being one of SC's strengths, your host gave in to the Dark Side and purchased the widescreen version of the discs (there is never any reason to buy the "fullscreen" version, a rant for another time). Having watched all of the movies at this point, several comments are in order:

1) SC's recollection of the 1997 editions is highly suspect, and he doesn't want to claim that he knows if it's gotten worse in all such respects, but the heavy use of CGI does not make the films look good. The technology available to George Lucas during the original filming involved a heavy reliance on puppets, costumes and models, and admittedly, it's easier to add all sorts of fine articulations to computer-generated simulations -- but the net result is to make the movies look more fake than they did. Especially out in space (which has been appreciably darkened, a good thing), the CGI spaceships tend to glow unnaturally, as do the internals of the Death Star during the attacks in both the first and third movies. It's not often that CGI is a step back in special effects, but in this case it is.

2) The inclusion of this character was the occasion for a wholesale revision of the iconic Max Rebo Band scene in Return of the Jedi. It still ends as it has to -- with a slave dancer dropped into the Rancor pit -- but the same caveats about the awkward fit of CGI animation to the rest of the movie apply here as well. And anything that takes away time from the visual of a keyboard-playing blue elephant is a major minus in SC's book. Note that the "Behind the Scenes" information says he showed up in the 1997 release, which only goes to prove that SC's memory of the originals (which he has seen multiple times in other formats) is better than his recollection of the seen-once Special Edition films.

3) In fairness to Lucas, dubbing Ian MacDiarmid in as the Emperor's voice in the pre-Return movies works out well. It's the right kind of fix for continuity purposes (unlike sticking Hayden Christensen in), and it's carried out subtly.

4) Anyone buying the set for the Episode III trailer is going to be sorely disappointed, unless by "trailer" they understand a 9 minute film about the machining of a single part of one costume. That part happens to be Darth Vader's helmet, so it's a reasonably big revelation (SC always thought that III would end with Vader donning the mask for the first time; it looks like it will happen considerably earlier than that). But finding out that The Original Bad Dude (tm) is going to take up more than a minute or two of the film isn't worth $45. The Ken Burns documentary merits at least part of that price, though.

For more optimistic views than SC is prepared to give, but from credible sources with better playback equipment, have a look here or here. Despite the raves, these are not reference-quality discs like Episodes I and II (judged strictly on technical merit), and SC will be very surprised if they're on display much at the show he'll be at in November.

UPDATE: Courtesy of Arts & Letters Daily, a rather more cynical take on DVD extras.

(Edited on 10/2/04 at 5:35 p.m.)

A real fraud

^{Geoff Pullum and Mark Liberman have been having a lot of fun with the story about Dan Ra}th^er. Your host having nothing unique to contribute to that discussion, we'll talk about that greatest doctoring of culturally significant texts to happen this year, the release today of DVDs purporting to contain the original Star Wars trilogy.

SC has taken shots at Episodes I and II before, both here and elsewhere. Let's take a few more before going on. How are Episodes I and II different from the collected oeuvre of Ed Wood? I count at least three ways. First, one body of work represents a filmmaker's serious attempts to muster all the creative and financial resources at his disposal to produce genuinely entertaining films, if not great art. The other group is the latest set of entries in the Star Wars universe. Second, even if Wood's movies were terrible, at least his life story resulted in a terrific film, something that will never be the case for George Lucas (unless the film pretends he died in 1985). Third, no matter how bad an actor Tor Johnson was, he showed a wider range of sincere emotions than Hayden Christensen. Vampira is similarly better than the utterly talentless Natalie Portman. It's enough to make SC want to throw in with the Empire.

That said, your host will probably see Episode III in its first week of release, same as the last two.

But as was hinted at the beginning of this post, SC considers Lucas' greatest artistic crime to be the new DVDs; the latest movies are dismissable as merely an exercise in toy sales, fast food marketing, etc. Of course, the reasonable reader might object that, since the movies weren't released until today, there's no way that SC has had a chance to see them, and this rant must be based on hearsay. That's not entirely off the mark, although the "Special Edition" films were a preview of the problems to come. But for a lot of other information, SC will be working off of this review, which agrees almost perfectly with his editorial stance.

Now, were the changes restricted to the original Special Edition cleanup of artifacts induced by '70s technology -- and the already-disturbing historical rewrites of that time -- SC would not be complaining. The original Star Wars was famous for introducing Dolby Surround as a viable format after quadraphonic sound had died; remixing the soundtrack to showcase Dolby Digital and THX is true to that spirit of innovation. Unfortunately, the Ministry of Truth Lucasfilm has chosen to engage in a wholesale erasure and rewriting of history that will be intolerable until everyone who knows better is dead (just 70 more years, George!).

One early -- and annoying -- revision of the plot of the original Star Wars had the simultaneous effects of both eliminating the maturation of Han Solo and making the movie even more implausible than it already was. This hasn't changed. What has changed is the relentless editing out of actors and voices from the originals to make the people from the new films appear to have been part of the entire history all along. Thus, Temuera Morrison's voice has been dubbed in as Boba Fett's, even though he wasn't cast as Fett's father until some 30 years after the first films were made. Ian MacDiarmid's face has been digitally substituted into places where it never was before (although, in fairnes, he actually was the Emperor in part of the first go-round). Gungans -- read: Jar Jar Binks -- have been inserted into the already overdone celebration at the end of Return of the Jedi. And most despicably, the insufferable Hayden Christensen's face has been placed where Sebastian Shaw's was as the ghost of Anakin Skywalker at the end of Return of the Jedi. This is wrong. Darth Vader is the scariest villain in the history of the universe. Hayden Christensen is a whiny little punk. SC knows Darth Vader, and Hayden Christensen is no Darth Vader.

This behavior can be defended as merely editing the films to conform to Lucas' original artistic vision. Darth George has made exactly this argument himself. But aside from updating the special effects, much of the tinkering is merely done for the sake of being different, not better, as the list of audio tweaks mentioned in Alexandra DuPont's review suggests. Worse, though, is the idea that continuity is represented by simply pasting in the last set of faces. It would be like going back and editing We Don't Want a Doctor (if you don't get the joke, click here and read the 7th bullet point) to show Pierce Brosnan everywhere that Sean Connery appeared. Bond fans have no difficulty seeing them both as James Bond; why, oh why, do we need to see Hayden's face everywhere to think Anakin Skywalker? (Then again, maybe SC should be thankful they didn't try to fit Jake Lloyd in instead.)

Mmm...silver...

Languagehat takes exception to recent skepticism by Mark Liberman and your host on the subject of winetasting jargon. It occurs to your host that he could be more precise about what bothers him with winetalk.

First off, though, your host wishes to be clear about his intentions in the last post. Languagehat is of the opinion, as is Mrs. SC, that statements like "the nose starts off with hints of leather and caramel" are very much reasonable in the context of wine-tasting. SC doesn't think this is a priori wrong, but that it requires a certain amount of justification. Within the context of audio electronics, where this sort of language is often applied, the justification is largely lacking (with the usual qualifier about loudspeakers).

Consider, for example, the products of a company called Kimber Kable. As a disclaimer, SC uses some of their products, but only the lowest-end ones. This is justified by the quality of their connectors, and the fact that they publish electrical specifications, which few competitors do. It is not justified by the claims made by either the company or reviewers regarding the effects which their cables may have on sound. They offer a particular cable design in several "flavors"; with all copper conductors, it's PBJ; half-silver and half-copper, it's Silver Streak; all silver is called KCAG. In order to justify purchasing Silver Streak or KCAG, one has to believe that different sounds are associated with the different metals used. Fortunately for Kimber, they found a reviewer to make exactly that claim. In the linked article, Corey Greenberg actually writes that the silver cable sounds "brighter" than competing copper cables. Needless to say, there is not a controlled listening test anywhere in the world that lends support to this statement.

Languagehat makes another comment that would seem to make that last point moot:

I've taken courses, and I assure you there is a great deal of chemical information about the components of taste that can be learned by sniffing tubes with essence of chocolate, herbs, &c, and learning to discern them as components of the sensory impact of wine.

In short, subjective experience seems to correlate well with the claims being made. As your host remarked in an initial comment back, it's plausible that the situation in wine tasting is considerably different than that in audiophilia; the number and proportion of chemicals within wine presents a much greater number of variables than are available to be examined in audio. The argument that special qualities are subjectively apparent even in the absence of obvious explanations is frequently made within audio to justify claims of superiority for high-end equipment. Unfortunately, when asked to attempt to isolate these qualities under controlled conditions, practitioners of subjective reviewing often either refuse to participate, or denounce the reliability of the results. Characteristic of these arguments is this infamous exchange between the late subjective enthusiast Steve Zipser and skeptical reviewer Tom Nousaine (although the particular retelling is Nousaine's, and is perhaps somewhat biased).

It's hard to ever conclusively demonstrate that subjective audio reviewers aren't hearing something different, because reports of experience involve attempting to communicate phenomena which other people may simply not be sensitive to. Vacuum tubes are often considered to sound "sweet" relative to transistors; SC has never been able to hear audible differences between the two. Paper loudspeaker cones are often considered to sound "warm" relative to plastic ones, and this is held to be different than the "warmth" that describes exaggerated response in the 80-200 Hz region. Tubes are also claimed to be more "musical", a term which is never even remotely close to being defined, although sometimes it's contrasted with "mechanical" or "harsh", which are usually pejoratives for digital electronics. These effects are often held to be quite subtle, and subjectivists will frequently fall back on the claim that being forced to concentrate on listening for them creates an observer paradox, where the very act of paying attention makes them unobservable.

Similarly, the fact that SC is never reminded of strawberries, blackberries, or cherries when drinking wines does not mean that they don't stimulate other people to experience such sensations. However, to the extent that these claims are meaningful, it ought to be possible to identify chemicals within the wine which are identical to those found in the berries. Someone who doesn't detect these flavors is presumably not sensitive to those chemicals at the concentrations they occur at within wine instead of berries. If, however, practitioners of winetalk claim to be able to distinguish these flavors, and it is objectively demonstrable that none of the molecules in the wine even bind to the same receptors on the tongue as those in the berries do, then their claims are considerably weaker.

Finally, Languagehat says:

Really, you'd think linguists would know better than to make fun of peculiar subcultures!

This is a fine line to walk on. It's definitely the case that linguists were able to respect the grammatical subtleties of "he eating" and "he be eating" long before anyone else recognized that Black English wasn't merely a matter of English illiteracy. As a profession, we're temperamentally disposed to avoid rushing to judgment, or at least most of us like to think so. However, when an electronics reviewer or a wine taster makes claims to special expertise, the rules are different -- claims to truth need to be backed up with evidence. When a reviewer insists that the evidence can only be observed in sighted tests, and occurs in spite of a lack of obvious correlates, there are only two possibilities for interpretation. One is that the reviewer is acting in bad faith, and the prospect of being paid either in cash or free wine is leading them to make statements they know to be false. The other is that they're acting in good faith, and that the phenomena is so observer-dependent that no amount of effort on the part of those who are not sensitized appropriately can ever lead them to see the truth. In this case, the statements might be true, but they're certainly not helpful. If it's just a matter of culture, though, and we're not meant to take these statements as anything more than participation in a ritual of some sort, then it's not fair to ask non-participants to accept them as truth-conditional propositions. Personally, SC suspects that wine tasting is actually on firmer ground than that -- but if so, that makes it fairly unique among the topics which are discussed in this style.

Fruity noses and sweet tubes

Mark Liberman has an excellent post today about winetalk, the habit that food writers have of attributing a variety of tastes and scents to wine, which are at best dubiously provable by a decent food chemist.

SC has always suspected this sort of talk of being an out-and-out fraud, even though he will happily agree that the products in question often do have fairly unique properties, and are not perfect substitutes. But he'll also concede that in the area of wine tasting, he is a dilletante at best.

However, there's another area where he considers himself expert, and that's audio electronics. In this area, your host is a lot more confident that the terminology is, to put it mildly, bunk.

A long time ago -- right after World War II, really -- there was an explosion in popularity of do-it-yourself kits for building amplifiers and loudspeakers. Not phonographs, which require more sophisticated machine tools to build (making a flat platter to hold records is more difficult than you think, unless you've used a lathe before). The necessary science being well understood by engineers, but not by casual consumers, there was a need for an educated press to explain this stuff and provide credible reviews. Into this void stepped a gentleman named Julian Hirsch, the first serious reviewer of audio equipment. Being a trained engineer, Mr. Hirsch's reviews tended to be long on measurements and short on descriptive language. It might be reasonably contended that he had something of a tin ear; certainly, his loudspeaker reviews of the '80s, '90s and '00s suggested that he wasn't overly bothered by the sound of loudspeakers whose measurements suggested aural disaster (conspiracy theorists would argue that his employer, Stereo Review, simply wouldn't allow anything negative to be said about advertisers).

In reaction to the Hirsch style of reviewing, another gentleman named J. Gordon Holt founded a magazine called Stereophile, dedicated to what he called "subjectivist" reviewing. Mr. Holt's innovation was to bring a limited version of "winetalk" into discussions of the sound of electronic equipment. Holt was generally an honest man, writing at a time when the differences among electronic devices (as opposed to electroacoustic ones, like loudspeakers) were clearly audible. Thus, equipment perceived to emphasize treble sounds might be described by Mr. Holt as "bright", a loudspeaker with heavily distorted bass as "muddy", and an amplifier with diminshed midrange response might be "lean".
This sort of subjective discussion didn't preclude recognition of objective measurements; Holt's negative review of the original Bose 901 loudspeaker contained both flowery winetalk-like language and serious engineering objections.

Over time, though, the differences in sound among amplifiers largely vanished. While vacuum tube-based amplifiers were often prone to complex interactions with loudspeakers, producing recognizable "sounds", solid-state equipment tends to be considerably more predictable. This greatly reduced the credibility of discussion of winetalk-type language in the consumer audio press, except as applied to loudspeakers, which still exhibit considerably more variation in their sounds and interactions with listening rooms than any other piece of equipment.

Since the product changed so drastically, so did the market for audio journalism. There is now a split in the press between Sound and Vision, heir to Stereo Review, where such language never appears (but neither does serious criticism of flaws obvious from the data), and magazines like Stereophile, The Absolute Sound, and The Audiophile Voice strive to pile ever thicker layers of descriptive language onto equipment which is more and more narrowly differentiated in measurable performance. Consider this line from a recent Stereophile review of a vacuum tube-based preamplifier:

In the process, the surrounding instruments became somewhat less mechanical and clattery, and the edge (that harmonica!) got smoothed off some: The Cary, though never dull, was consistently free from treble nasties.

This statement is very hard to reconcile with the published measurements, which indicate that the signal passes through almost perfectly unaltered. In fact, despite the vacuum tube design, its measured performance is indistinguishable from a good-quality solid-state device. (In fairness, tubes can be very linear, low-distortion devices; the difference in performance is almost as much about improved manufacturing tolerances as anything intrinsic to the devices.)

Just as Prof. Liberman notes that winetalk deals with "luxury food products", this sort of talk largely occurs with luxury electronic goods. However, unlike wine, much of it really cannot be "finely differentiated...in a plausibly reproducible way". It is no overstatement to say that if the current purveyors of "audio winetalk" decided to stop speaking as they do, the high-end electronics market would collapse overnight. Certainly, there would be no more market for wires claimed to be "fast and airy", neither of which has ever been demonstrated to correlate with anything reproducible.

Another thing I'm not sure Prof. Liberman is quite right about in his assertion that "other food products like beef and sweet corn have been exempt so far". Certainly, it's not true to the same extent, but some people are trying; witness Omaha Steaks' attempt to label some of their meat "Private Reserve", a name which would hardly look out of place in the wine aisle of your local grocery store. Ditto for some of the claims made about Japanese Kobe beef (described in the link as "sweeter and lighter" than "ordinary" beef), which interestingly is already subject to a certain amount of scientific skepticism. As for corn, well, in San Diego County, at least, the name Chino really does have cachet among the gourmet set as branding for produce -- although SC hasn't yet found anyone saying that the corn displays "a nose of soil and sugar, with just a slight hint of Roundup, and a firm, leather-like finish".

Despite being wildly skeptical about winetalk, SC doesn't let it get in the way of enjoying his electronics hobby. Even when it's very, very hard to be sure that the terms describe anything that really stands out as much as is claimed, there are often real differences in things you can still enjoy: craftsmanship, appearance, ease of use. And the language isn't always inept; sometime in the not-too-distant future, SC will publish a few audio measurements taken around the house to explain why he likes his headphones better than his computer speakers, his home theater setup better than his headphones, and why his current speakers are far from likely to be his last.

Information theory and movie reviewing

Not long ago, SC had occasion to write about the free speech issues involved in distributing censored versions of movies. Today, he learned about another alternative, which he's also not likely to make use of, but which certainly sidesteps the free speech issues.

A husband-and-wife team have started a website called "ScreenIt", dedicated to reviewing movies, not on aesthetic grounds, but simply to catalogue their content in a variety of potentially offensive categories. Their self-description:

Screen It! was created to give parents a way to access the content of popular entertainment their kids are exposed to. It is not intended as censorship. Rather, it is designed to allow Hollywood and Record Labels to continue to produce movies, videos and music while informing parents of the content in them. Some people argue over the moral quality of films while others want to ban certain albums that contain material that particularly offends them.

That is censorship, and it's not right for others to decide what you or your children can see. That decision lies with you. Until now, however, there was no way for parents to find out about the content of movies, videos, or music. For movies, the MPAA rating (G, PG, etc...) is a start, but offers just a one line, generalized description of the "offending" material. A few newsletters here and there offer a little more information, but are usually biased or are lacking in detailed content listings.

They list 15 types of potential offense for each film, and have a couple of possible values for each of them. No key is provided, but each one seems to range from "mild" through "moderate" to "extreme". A sampling of their reviews indicates that for each movie they cover, they've got a fairly exhaustive catalogue of the various scenes that lead them to assign each value.

Although SC is pretty much happy with the existing MPAA ratings, it's certainly conceivable that some people might not mind their kids being exposed to nonstop gore, but have a problem with foul language, or "inappropriate music" (I'm not kidding; it's one of their categories). Considered as metadata about the movies under review, this is an interesting approach -- one could easily conceive of categories being standardized, and then publishing more detailed ratings indicating the types of items which give offense.

While we're thinking about it, there are plenty of opportunities for having finely tuned metadata available to describe movie content (or books, or music, etc.). How many of us would have been grateful for reviews of this film that clearly indicated the presence of flashy CGI sequences, but also the complete and total absence of a credible plot or efforts by the actors? Of course, much of this can be obtained by reading current reviews, but those can be full of hedges and qualifications. Considered purely as an information-theoretic code, there might not be a less informative tool around than Siskel/Ebert/Roeper's thumbs-up/down. It can only convey two states! Extending that to even be separate thumbs-up/down for each of a half-dozen criteria would tell us a lot more about the qualities of a movie than simply whether it's good or bad. Some people just want to see extended CGI-fests; others, beautiful costumes (or people). And some of us even are concerned with whether or not there's a point. All it would take is one, maybe two lines of text in your local paper, where whole paragraphs go in now.

Of course, one perfectly reasonable objection is that any coding scheme short of natural language is going to be largely an exercise in oversimplification, not to mention the fact that this approach completely deprecates the writings skills of reviewers. To which SC says, "So what?". How many of these people were worth reading beforehand? It's been a while since your host can recall reading a review of anything broadly falling under the "pop culture" label that he considered to have literary/artistic merit in itself. That's not to say that such writing hasn't existed in the past -- it may still be produced today -- but the review as a literary form in itself strikes SC as having taken a miserable beating. A few ones and zeros could hardly be that much worse.

Real-time movie censoring

Courtesy of the Drudge Report, an article on a new DVD player that Wal-Mart will begin selling shortly. Incorporating software from a firm called ClearPlay, the machine will make use of predefined classifications to decide when/where to skip ahead in a movie, so as to avoid playing content which falls into one of four objectionable categories: violence, language, sex/nudity and explicit drug use. The player allows users to automatically skip any combination of these categories, so long as the movie has been prescreened by ClearPlay editors, who then create a file containing the needed filtering information, which can be downloaded to the player.

Naturally, the movie industry is not pleased, arguing that this software violates the copyright licenses granted to consumers who buy/rent copies of their movies. This point is controversial. Other firms, like this one, specialize in producing "edited" versions of current movies and music, cutting out scenes deemed objectionable, and releasing physically altered copies. Early entrants to the "edited rerelease" market engaged in what was indisputably piracy, selling copies of edited films without compensating the film distributors for each copy sold (as would be the case through normal retail). However, current editing services, such as the one linked to here, require you to purchase a legal copy, which they then edit (if VHS, which would be expensive to duplicate en masse) or replace with an edited copy (if a DVD). While this certainly alters the director's intent, so long as the copy has been legally acquired, this strikes SC as being protected by the same "fair use" laws that allow you to otherwise copy legally acquired recordings for personal use. Where Hollywood may have a fair gripe with present-day editing businesses is that this doctrine is only intended to protect end-user behavior, not people seeking to make a business of redistributing the movies.

The ClearPlay DVD player appears to remove the editing objection, because the consumer's copy of the movie is never altered. Effectively, all it does is fast-forward the movie at a set of predetermined points, and there is no way any court in this country will uphold the idea that fast-forwarding a movie in a private home viewing session is illegal. ReplayTV (a competitor to Tivo) already has demonstrated this point legally by suriving challenges to its commercial-skipping feature.

Before going on about the technical issues involved, let's stop to dispense with one objection; namely, that if consumers don't like a movie, they should just not watch it. Hollywood is free to make whatever they like, and none of the people buying censorware are doing anything which interferes with the ability of the studios to disseminate R-rated films. This is a matter of what people want to be able to view privately, in their own homes, and especially about parental control. While SC would not personally purchase movies which have been edited by censors, no matter how well-intentioned, nor would he buy products designed solely to automate censoring of content, he doesn't think that the studios are on very solid ground by demanding that people not do what the fast-forward button has enabled for years. Furthermore, the censorship involved is on a very selective basis; this is not like the Websense episode, where filters meant to block one type of content had the side effect of blocking significant amounts of material that was clearly not targeted by the policy. The fact that Internet-filtering software is grossly inadequate to that particular task does not invalidate the right of employers to set policies about appropriate use of their property, and the same is true of parents at home. Assuming that some capability for parental control can be provided, SC doesn't see how anyone's rights are being violated. Nor, as long as the studios are paid the market price for a license to watch their films, how any economic injuries are being done.

This brings us to the question of whether linguistics could provide a solution that satisfies content providers as well as people seeking to reduce their exposure to obscene material. Present-day speech recognition is virtually real-time, but still lags the millisecond-level synchronization between audio and video streams that would be necessary to block obscene speech during playback. Your host is ignoring the fact that such a system would inevitably also censor some words that merely happen to rhyme with the desired stoplist; right now, we're just talking about technical feasibility. It might be possible to set up a DVD player with a 3-4 second buffer, run speech recognition on it, and insert beeps with the benefit of a second or two to ensure adequate processing. This would probably raise the cost of a DVD player by $50-60, to incorporate both the additional memory, as well as a special-purpose DSP, and would still only address the filtering of speech, not video. While this cost is probably more than acceptable to people who would otherwise spend even more on edited copies of movies, it doesn't accomplish as much as a system like ClearPlay can do right now, covering both audio and video. Doing the same kind of real-time, or even delayed, recognition of video is much more problematic, and your host won't even try to estimate the costs or the technical barriers that need to be overcome in order to provide hardware-based censoring of video content.

So finally, granting that editing of content is likely to be superior to any near-term technical solution, this raises the question of whether or not permitting such editing ought to be accepted as a matter of public policy, or market behavior. As I've already indicated, selling customers a service that essentially automates pressing the fast-forward button is more likely to survive legal challenges than other forms of editing out there. From Hollywood's perspective, the question is whether it's better to continue the fight, and risk setting a precedent which could later be interpreted to expand end-users' rights in ways not even contemplated right now. Or whether they might as well acknowledge the existence of demand for less-violent/profane entertainment, and work to create official versions which allow them to continue to sell directly to the consumers in question, perhaps even at the premium they've already demonstrated a willingness to pay. There are no technical barriers to doing so, at least for DVD releases -- the original DVD specification always included provisions for multiple camera angles and soundtracks, in part to allow for movies to be played according to different preferences, not all of which were the "original vision" shown in theaters. Extras like this are even considered a selling point for "special edition" releases of the same movie over and over and over. So the idea that showing a version of the movie at home which was not the director's original theatrical vision is bunk; it might be more interesting to find out why studio executives feel that it's legitimate to spend additional funds to create home releases featuring more violence, nudity and profanity, but not less.

Only Canadians need apply

SC is always on the lookout for the latest and greatest in audio/video equipment. So every now and then, he visits the websites of various manufacturers to see what's new. He was mildly peeved to discover one that doesn't seem to want his business (or maybe they just know he's not really in the market right now).

InFocus, a leading manufacturer of video projectors, has apparently come out with a new DLP projector which may finally be possible for your host to watch without becoming dizzy. (DLP makes use of a multicolored wheel in place of the colored lenses common to traditional tube-based displays -- some people see floating spots on the screen just like you get when you rub your eyes.) Since this was interesting to me, I wanted to find out where I could see it demonstrated.

So off to InFocus' dealer information form. The only states/provinces listed? All Canadian. The only North American country listed? Canada. This, from a company located in Oregon. Perhaps SC will be interested at some future date, just as soon as California is annexed to British Columbia.