Way back in 1999 when Betty Boards were beginning to circulate beyond the DAT trading community (I have never owned a DAT machine) and onto FTP servers, I acquired a copy of the shows (digital clones of the What's Became Of The Bettys project) and burned them to CDR. I had more free time on my hands back then and was discovering a lot of the Dead's music for the first time. I was pretty excited about being able to finally trade music without dealing with nth generation cassettes! Anyway, I popped in 2/18/71 and immediately became aware of the chronic noise problem. It gave me a headache to listen to the show. I had recently bought a copy of Sound Forge and the Sonic Foundry Noise Reduction Plugin and thought that, perhaps, I could notch out the noise without adversely affecting the audio.
I have some experience in audio production, Fourier theory, Nyquist theory and do a lot of listening on a halfway decent stereo. I knew any digital mucking with a recording will affect the sound, and one should always apply the lightest touch possible when it comes to any digital processing of audio. To my dismay I quickly discovered that the center frequency of the whine varied throughout the show. This may have to do with the reel speed changing, or perhaps the interference itself was moving around (I have yet to get a definitive answer as to the source of the interference).
I realized from trial and error that if you applied too wide a filter and just applied it to the whole show, you'd get a very hollow sounding show. Realizing this, I decided instead to apply a whole bunch of narrow filters. Here is my old 1999 page I wrote about this.
It's almost ten years later. My interests have changed. I no longer have the free time I had. I am more into studio recordings and building amplifiers and that sort of thing. I still love the Dead but have drifted away from regularly listening to their live shows. However, occasionally I will go to bt.etree.org and see what new shows are making the rounds, or better releases of old shows, that sort of thing.
Not too long ago I discovered the Port Chester shows were remastered with an impressive set of equipment, and at a higher bit depth and resolution that I had access to in 1999. I am specifically referring to this mastering from the "Green Mountain Bros." I was honestly expecting to be blown away by this new transfer given the impressive care which was obviously given to the transfer. However, due primarily to a brute force approach to dealing with the whine (not the first time this has happened), the show sounds really bad. There has been a huge hole carved out of the audio spectrum. The following two figures compare my 1999 NR approach to the Green Mountain Bros. approach. I have sampled some of the space between songs where the tape was rolling but the band was not playing, and done a spectral analysis which compares amplitude to frequency:
The "bite" out of the spectrum is where the notch filter was applied in both cases. In the Green Mountain Bros. mastering, they have applied a much wider filter than I did, and, as far as I can tell from sampling other bits from their release, applied it to the whole show.
I wan to be clear here: I do not think that my 1999 attempt is the final word in this show. I do, however, think it is the best sounding filtered version of this show in circulation. I have to re-iterate that it took me days and days of squinting at the screen to painstakingly apply the narrowest filter possible to each song. It was a colossal pain in the ass. I think if you listen carefully, it does sound like there is a slight amount of hollowness to the sound on occasion. I may have applied too much attenuation, taking out more than just the whine (its amplitude did vary also, however). The hollowness I hear in my version might be there in the original too, though; I don't have copies of the unfiltered shows any more. Not all of the Betty Boards are the final word in audiophile nirvana, in my experience.
I do believe that a better job can be done. I believe the Green Mountain Bros. have the right idea by doing all digital manipulation in the 24/96 domain. I honestly do not know what bit depth the Sound Forge Noise Reduction Plugin worked in; I would presume double precision floating point data, but I cannot be sure that it wasn't single precision. I did not apply any dithering after processing, and did not upsample to 96 KHz, etc. Hell man I was running a 233 MHz Pentium running Windows NT for crying out loud!
What I would recommend to anyone wishing to give the Port Chester shows (or any other 71 shows with this problem - there are a few more) yet another go is that you use the narrowest filter possible, and keep an eye on the center frequency.
I find it somewhat interesting that the only official release from the Port Chester shows is Rhino Records' release of 2/19/71, which also happens to be the one show I didn't work on, for the simple reason that the Betty Board was recorded in mono and I felt that it wasn't worth my time since I'd never listen to a monaural show anyway! I would be very interested to know whether the Rhino release is in stereo, and also whether they cleanly notched out the whine.
The remainder of this page is as it has been since 1999 when I first wrote it.
In August I seeded the etree with Grateful Dead 2/18/71 and 2/21/71, from What Became of the Bettys DAT source. A month or so later I seeded 2/23 and 2/24 of the same year, same source. These are all from the Port Chester run. I am currently working on the remaining two show from this run. The shows had a high-pitched whine in them which I removed with the Sonic Foundry Noise Reduction plugin (for more on this check this link out). I used many narrow filters to follow the drifting whine. I found that the NR made the shows listenable, and received positive input from the etree trading community.
Another version of 2/18/71 is circulating now. This version sounds bad. Instead of making many narrow notch filters, whoever processed the show hit it with one very wide notch filter.
Below are two frequency response curves of intersong noise (at the end of Bertha from 2/18/71). The top one is mine, the bottom one is the newly circulating one.
I say this as a warning... you probably don't want to burn the newly circulating stuff. It sounds really hollow. I'm not saying someone couldn't have done a better job than I; with much patience and better software I know it could be done.
A final note. A month or so ago I contacted one of the folks who announced that they were going to start circulating all the Port Chester shows, staring with 2/18/71, to tell him that I had already cleaned them up and that he might want to save himself the hassle of reinventing the wheel. His reply to me was, in its entirety:
We are not using any of your tapes. Never have, never will.
The 2/18 I am warning you about is the result of their efforts. See the etree shn circulation info on existing versions of this show and many others. Both versions, mine and the other one, are from the excellent WBOTB source, but with different post-production techniques.
Anyway, CAVEAT EMPTOR... --Leigh Orf, 
Fig. 1: Frequency response curve for a ten second sample of intersong "silence" after Bertha. The flatness of the response from ~5000 to ~17000 Hz is a testament to the fact that the sample is mostly tape hiss (there is some tuning / talking going on; that's the signal at lower frequencies). The area to focus on is at roughly 7800 Hz. The sharp peak is the whine, the valley is there because I had to make the filter slightly wider than the whine because it didn't stay at 7800 Hz all of the time; it wandered.
Fig. 2: Roughly the same sample as above, taken from the newly circulating source of this show. Notice the large chunk taken out of the curve. Instead of creating many narrow notch filters to follow the whine around, whoever processed this show hit the entire show with 1 wide (over 1000 Hz with -20 dB attenuation) filter. This leaves the entire show sounding hollow.
For more notes on how I processed the early '71 shows, see this link. For feedback on my hiss noise reduction query to the etree, see this one.