Sheffield Steel? Doug Sax Page 2

Harley: Is it similar to your modified Sony PCM-1610 at The Mastering Lab?

Sax: What we have is highly modified, but we haven't gone the whole route. There will be some point where you need a separate chassis and your own architecture so you can do exactly what you want. But there are enormous efforts going on, mostly by American companies, on the A/D side.

Harley: Is there is anything intrinsically wrong sonically with storing music as ones and zeros, or do you think the problems are in the implementation of current technology?

Sax: The latter.

Harley: Do you think digital will one day sound better than analog?

Sax: It should. Digital is analog. Digital is only the storage. The only way we have to evaluate it is to listen to it. Until we can plug ones and zeroes into our heads, we're going to have to listen to it. I don't know if you're going to get everything you need within the format. It's interesting because there are 18-bit D/A converters—ain't no 18-bit A/Ds. If there were, there's no way to store 18-bit numbers. And if you had a professional recorder that would store 18 bits in and out and really get some low-level resolution, the CD cannot be expanded to 18 bit.

Harley: We really are locked into the CD standard.

Sax: I don't know if it's true—I get mixed stories—but a DAT can [handle more than 16-bit data], from what I understand. Again, you're sitting at home with an 18-bit converter, 64x oversampling. It's like taking the finest ground-glass spectacles to look at a photograph in a magazine. Well, it's even clearer, it doesn't refract light, but it's still a photograph in a magazine. You can do all you want but if you really want to get the quality, the photograph itself has to be better. It has got to be pressed with more dots, or it's got to have something better done to it. I'm constantly amazed to hear guys say, "Well, I tried this new CD player and I think it really does it." It doesn't really do it because the recording doesn't really do it. "I got a new arm and cartridge and all my records sound fantastic." They can't. Only a handful of records sound fantastic.

Harley: How many records a year sound great to you?

Sax: About 10 a year of commercial things. Maybe that's high. I've been on the committee that does the nominations for best engineering, and I have no idea why anyone submitted these recordings for good engineering. It's hard to come out with five that really sound like something at the end of a year, and those are often not even the best five because there are nine voting chapters—Nashville, Chicago, New York, etc. Everybody goes in to make a really terrific record...What's your opinion of pre-emphasis?

Harley: From the playback side, I think it's better not to use it. Who knows what the de-emphasis circuit is like in some of these CD players? It's better not to have the extra electronics in the signal path.

Sax: Makes sense to me.

Harley: What do you think of pre-emphasis?

Sax: Well, your comments are right. It's an additional variable in the playback chain. But digital doesn't really handle the dynamic range as well as good analog. There are virtues to pre-emphasis technically—it takes low-level, high-frequency information that digital would miss and boosts it to where digital can capture it. I do hear more resolution in pre-emphasized product. It is quieter. You can actually hear more resolution. Especially on classical, where the resolution is more meaningful. Some of our CDs are emphasized. Digital doesn't hear down to the low levels. Obviously it becomes more coarse. If it were a 16-bit logarithmic system, I would probably be dramatically more impressed by it.

Harley: A logarithmic system is one in which, as signal level decreases, the amplitude between the quantization steps also decreases.

Sax: Basically, it would have the same distortion at –40 level as it had at 0 level. At 0 it would have higher level than what we now measure, but the level is up there only one or two or three times on the whole record, flash peak. Particularly if it is classical. But that distortion would be the same at –60 in a logarithmic system. It wouldn't become coarser and coarser in its abilities as you drop down in level. Linear digital is the opposite of analog. Analog becomes more and more linear as you drop down in level. The propensities to cause distortion increase at high levels, which would be saturation of the magnetic materials and hard working of the analog amps. But if you drop down 40dB, everything is sleeping. The particles are all very linear. There is lower and lower distortion with lower and lower level.

Harley: Distortion is particularly noticeable when you bring down a digital fader and get near the bottom of its travel.

Sax: The analog equivalent of that is the volume control. When you bring that way down, it's fine—no distortion. But [in digital] these low-level components that you can isolate when you take the fader down to –60 are going on all the time. In a system with a recording of that clarity, it is covered up by the loudness of the signal, but it's still not resolving the low level correctly. You put in a –60 tone and it sounds like a buzz saw, it sounds like a kazoo. You open them up on a live signal, like a concert hall, you hear it gating on the room noise. It's like "I hear it, now I don't, now I do." It's a joke. It has a long way to go.

Look at it from the standpoint that says it sounds pretty good now. After the way it started its journey, it sounds pretty good, and if you know how many areas in which the format can be improved, if you know how bad some of the areas are now, you can visualize the fix. It's going to have to sound better. As I said, what it is capable of is just like your home TV compared with a studio monitor...

Harley: That's an interesting analogy. People are more discriminating visually than aurally since they get a reference of reality every waking moment. If people were exposed to real musical events more often, their acceptance of what is good sonically would change.

Sax: Take a string quartet. You've got two violins, cello, and viola facing each other. If the room has any size and you back off to the fourth row, you'll be surprised how small the string quartet is. If you sat in the 15th row and shut your eyes, it would sound like mono. Now you go on to the perspective of someone in the audience, not at Carnegie Hall but in a 1000-seat hall, and you sit in the eighth row and you tell me the angle between the string quartet and the listener. Now, you come up and you mike them. You put the violin over here and the cello here, a couple of mics, nothing esoteric. You take this home where you'll be 10' from the speakers and the violin now goes way over here and the cello goes way over there and it sounds fantastic. But does it sound the way the string quartet did when you were in even a good seat in the audience? Not remotely. And if you made a recording that conveyed the true size of it, even at 20' out of your speakers, nobody would like it. Because all of the other recordings have a cello that's as big as this room, and it sounds as big as an orchestra.

In fact, when you put on an orchestra, which is in reality 60' across, it doesn't sound any bigger than the string quartet. And they want it live so they record it in a barn and here you get these bigger-than-life instruments with five seconds of decay and people sit at home and say "Here is a string quartet."

The recording I'm most proud of is my new upcoming album of a string quartet. People don't realize how small a string quartet is. I have two recordings of Michael Newman, solo guitar. I put him in the Mastering Lab between my speakers just the way he would image, turned all the equipment off and let him play. Do you realize how quiet a classical guitar is? It's a delicate, gorgeous, quiet instrument. It's a big problem because I can play it out of the speakers 20dB louder than he is. It sounds wonderful. Loud clarity. But you go in front of the Harry James Band and that band is 20dB louder than my speakers can play with a 300W amp. You're afraid to get close to the band with mics. So conversely, the dynamic range of a quiet instrument without a big image comes out overblown, enormous, like a solo piano that goes wall to wall.

Listen to a solo concert grand piano from 15'. Shut your eyes and tell me how big that piano is, how much left and right. You hear recorded piano, even well-recorded piano, and it seems to go off the speakers. But when a live pianist goes from the top to the bottom of the piano, the thing doesn't move, it comes from the same place. It sounds about this big [holds hands about two feet apart]. Sure it picks up room and size, but it's almost mono. Yet if you made an almost mono piano recording, if you made something that would play out of big speakers that were 10' apart that sounded the way it really did in size to you sitting there at 10', you would be panned out of the world for having a recording that has no dimensionality (footnote 5).

So in recording, there's reality and then there's how much fudging that is acceptable. You're used to having the small instrument sound enormous and used to having the enormous instrument, the orchestra, sound small. Because you have a speaker that is much bigger than a guitar and much smaller than a symphony orchestra, you've got problems.

Footnote 5: Those who have listened to the Stereophile flute and piano LP or my piano recording on the HFN/RR Test CD will realize that accurately capturing the true size of live instruments is a particular obsession of mine. It is also one of the reasons why I think Doug Sax's recordings of Michael Newman are among the finest acoustic guitar recordings extant.—John Atkinson