Our Story To Date

Alert and geeky readers will recall that we just conducted a review of the-state-of-digital-audio, considering such issues as 96 kHz. sampling, 24-bit word-lengths, and the current situation regarding dither. What the alert and geeky readers don’t know is that behind the scenes, that article went through a whole lot of discussion, review and modification, because there are at present a lot of different conceptions of and viewpoints about this stuff, particularly dither. There is some serious, even perhaps obsessive, interest in how much dither, what kind of dither, how many bits, how little jitter, and so on, you really need, etc., etc. You may even have run across some of this on your very own.

Well, as luck would have it, I just got to listen quite specifically to a lot of this stuff, under controlled conditions. I’d like to share what I found with you – I think it’s worth knowing about!

AES Meeting at Northeastern Digital

The Boston Section of the Audio Engineering Society meets monthly, and the March meeting was supposed to be about High Definition CDs. However, that fell through at pretty much the last minute, and so Dr. Toby Mountain, mastering engineer par excellence and owner of Northeastern Digital, the Boston area’s most established mastering facility, stepped in to host the monthly meeting.

Toby offered to demonstrate the effects of various kinds of dither as well as the relationship between 20-bit and 16-bit recordings. This was all to be put together as only a good mastering facility can do it, i.e. with thoroughly rigorous and documented procedures and excellent, documented and controlled monitoring, set up to accommodate all us AES types.

For me, this was all too good to pass up, so I fired up the big Mitsu, blasted down I-495 to Toby’s place, snarfed down some refreshments he thoughtfully laid on, traded a few lies with my fellow members, and settled down in Studio A for an evening of serious listening. Ahhhhhh.

What Was Demonstrated

Toby played us several things. He started with an excerpt of a David Bowie re-mastering job from analog that he had first done in 1991 in 16-bit and now was re-doing using 20-bit technology. So, we got to hear both the 16-bit and 20-bit versions of the same analog copy (the source was a 15 IPS analog tape with Dolby noise reduction).

This was followed by a snippet of a CD mastering job Toby is currently working on that included a fade to “digital black” and then a fade-up of the next tune. Toby ran up the monitor gain some 40 dB for the critical part of this so that we could really hear the digital grunge – we got to hear the effects of dithered 20-bit, truncated 16-bit, 16-bit with normal dither, with Sonic Solutions’ Turbo-Bit dither, Super Bit Mapping II dither, and Apogee’s UV-22 dither. After we’d finished tormenting our ears with all this stuff, we listened to all of the same stuff at normal levels, i.e. some 40 dB softer.

We repeated this exercise with a 20-bit source recording of a string quartet in a fairly live hall, playing an extremely pianissimo passage (once again, signal about 40 dB below 0 dBFS). This time there was no digital black.

After we got done with all the usual questions and harumphing, I went across the hall to Studio B, where Toby’s other engineer, Laurie Flannery, was doing similar, if less tweaky, demos.

What Laurie had done was, she’d taken some standard 16-bit CD recordings and digitally attenuated them 40 dB (i.e., slightly more than 6 bits) and re-recorded them with dither on a 20-bit system. This she played back with the monitor gain turned up by 40 dB. The net result was an approximately 14-bit dithered recording. From this, she reduced the 20-bit playback to 16 bits with truncation and then with all the various dither schemes previously mentioned. The net result of these were recordings that were the equivalent of 10-bit recordings with and without dithers. Source materials included a choir singing music from the Renaissance, a rock ‘n roll fade-out, and a minimalist blues singer/acoustic guitar bit with lots of holes and dynamic range. After we got done, Laurie then played back the original 16-bit recordings at the original levels.

In summary, we got to listen to low-level stuff with a what you might want to think of as a 40 dB magnifier (100X, you could say) under good, controlled conditions. We also got to listen to the same stuff without the magnifier.

So, what did it sound like? I’ll tell you.