Session File: Phish

The time: the summer of 1988. The place: the Roma Bar in Telluride, Colorado. In a few years, Phish will become the de facto successor to the Grateful Dead — a band more noted for its musically eclectic, super-improvisational, free form live shows than for its gold-status studio albums. Yet right now, in a small mining town venue that is figuratively a million miles from the arenas where they’ll one day perform multi-night engagements, the Vermont quartet of guitarist/vocalist Trey Anastasio, drummer John Fishman, bass player Mike Gordon and keyboardist Page McConnell are showcasing their songs to a half-dozen people. Among them is one Michael Lynch.

Over the course of five nights at the Roma, as well as another at the nearby and better-attended Fly Me To The Moon Saloon, Lynch illustrates his love and appreciation for the fledgling New England band by not only taking a feed from engineer Paul Languedoc’s soundboard and recording the shows on an assortment of Maxell cassettes, but also by taping over the likes of Pink Floyd’s Atom Heart Mother and the Grateful Dead’s 1978 Red Rocks Amphitheater concert. He won’t regret it.

Capturing Phish in its formative stages, Lynch’s recordings are entrusted to band archivist Kevin Shapiro about a decade later, by which point the musicians have evolved into a world-class outfit. And just under another decade later, following Phish’s disbandment, Shapiro transfers the well-worn analog tapes — played on a TASCAM 122 Mk III cassette deck without Dolby or pitch correction — to a Mac G4 using Pro Tools LE at 48kHz. He then compiles 36 tracks into a show-like collection, and delivers 24-bit/44.1kHz files on a Glyph GT050 FireWire hard drive to Fred Kevorkian for mastering into a three-CD set.

“Kevin was present during the mastering sessions and very helpful with his incredible musical knowledge of the band,” says Kevorkian, who spent nine years as a staff engineer at Sear Sound in New York after relocating there from Paris in 1989. During that time he began mastering, and this was the path he chose to go down when joining Absolute Audio in 1999. Five years later he set up on his own facility, establishing Kevorkian Mastering on the third floor of Avatar Studios. Since then, his album credits have included the White Stripes, Willie Nelson, Ryan Adams, Sonny Rollins, Iggy Pop, Cassandra Wilson, and the Dave Matthews Band.

“I master to an Audio Cube AC5,” Kevorkian explains, “and when I play files my source is the Sonic Solutions HD. Things go out digitally and most of the time they are converted to analog. I use all of my analog gear to get the sound I want, while the digital stuff is more for tweaking and surgical processing.”

First involved with Phish when producer Steve Lillywhite and mix engineer John Siket approached him to digitally sequence and edit the Slip, Stitch & Pass live album at Sear Sound in 1997, Kevorkian subsequently mastered a number of the band’s studio, concert, and solo projects prior to his work on Colorado ’88 in September 2006.

“Getting the overall tonal balance of the program didn’t involve unusual processing,” he states. “Most of the major problems we ran into while mastering Colorado ’88 had to do with balance, distortion, dropouts and tape hiss.

“The files were played back in real time with a Sonic Solutions HD system. The AES output was then converted immediately to analog by a Lavry 824 D/A, feeding directly into a Chris Muth 2020 mastering console. At that stage I can insert a variety of analog processors into the signal path, but for this project a Sontec MES 432 equalizer was first in the chain, with settings that were pretty broad and mainly used to ‘set the tone.’ The program was then encoded as an M/S signal, and fed to a Manley Variable-Mu compressor and Maselec MEA-2 equalizer.

“Because of the M/S configuration, I was able to EQ and sometimes compress the center differently than the sides. It’s a great tool, and it also allowed me to change the stereo imaging from song to song in order to create a more consistent soundstage. The signal was then converted back to regular stereo and routed to a Maselec MDS-2 de-esser that was set to control and smooth out the top end, just like the Neumann accelerator-limiter that is found on a cutting lathe.”

A matched pair of Chandler LTD-2 compressors was used occasionally to bring the average level up on the most dynamic songs. The signal was then converted to 24-bit digital with a Prism AD124 and fed into a Z-Systems parametric EQ, a Weiss DS-1 de-esser, and a Waves L2 Ultramaximizer.

“The digital EQ was used to tighten the low end and to bring out some of the lost presence,” Kevorkian explains. “The Weiss was mostly dedicated as a one- or two-octave band compressor, around 2 or 3kHz, taking care of the extra midrange added by the EQ, and lastly we used the L2 to reduce the overall dynamic range by a decibel or two without significant side effects. The output of the L2 was routed to the Audio Cube AC-5 workstation, capturing the mastered digital audio in real time, and typically a new file would be created in the AC-5 for every single song, as well as for specific sections.”

Compiled in chronological order from the shows recorded by Michael Lynch at the Roma Bar on July 29 and 30, 1988, the Fly Me To The Moon Saloon on August 3, and then back at the Roma on August 4 and 5, Colorado’88 challenged Kevorkian in terms of addressing the sonic discrepancies that were evident from one mix to the next.

“With all the files in the system we were ready to sequence the album,” he says, “but while this would normally be a fairly easy process, in this case it became our biggest challenge. As mentioned before, the tape hiss was a real problem — not only was it loud, but the band’s wide dynamic range would expose the noise during the very soft sections of the songs. Thankfully, the AC-5 has some pretty nice restoration tools, and so its ‘De-Noiser’ was used to remove some of the tape hiss.

“We decided to leave the loud sections of the songs alone. There the hiss was quite acceptable, but in most cases a noise reduction of 3 to 4dB was applied to the quieter sections. A new de-hissed file would then be created, synced to the original file, and crossfaded slowly with the loud sections. This worked out really well, but unfortunately you can’t just apply the same setting to all the noisy sections, so we had to analyze and process every one of the hissy parts separately. There were a lot less side effects that way.”

Still, the real challenge was to transition between two songs from different shows.

“We were dealing with a lot of hiss on each side of the edit,” Kevorkian says. “Each one had a different characteristic, so as soon as you’d crossfade them there would be some nice phasing. Usually you can cover some of those imperfections with crowd noise, but here I was, with an audience of only six people! What’s more, they were pretty quiet. So, in order to get those transitions right, we used a little more ‘De-Noiser’ to match both the hiss level and some additional internal EQs. Then we had to experiment with some uneven manual crossfades to create a smoother transition — tedious but efficient.”

Dropouts presented another problem, courtesy of poor tape-to-head contact during the digital transfers (and perhaps some shedded oxide) due to said tapes being old and worn after having been played numerous times.

“Some short dropouts could be minimized by a quick ‘envelope ride’ to compensate for the energy loss, as well as a burst of HF boost to bring back some of the top end,” Kevorkian says. “Then again, there were also times when we were lucky enough to find a section that was identical to the problem area, in which case it was just a matter of copying and pasting. Otherwise, we had to leave things alone because there was no successful remedy.”

Finally, there were the balance and distortion issues, such as a John Fishman trombone solo on “I Didn’t Know” that was so loud, it appeared to clip everything in the signal path.

“Distortion is very difficult to fix,” asserts Kevorkian. “We were able to smooth out the distorted signal a bit by using the Audio Cube’s ‘De-Clicker’ — that took some of the edge off. However, at that moment the trombone was about 20dB louder than Mike Gordon’s bass. The balance was really out of proportion. By chance, John was mostly in the left channel and Mike in the right, and during the previous mastering pass I already had a special setting for that particular section. The L2 wasn’t linked and it was set to heavily limit only the left channel, so we decided to pan the stereo signal about 3dB more on the right side to give the bass a little more weight and tuck the trombone a little more into the mix.

“That worked, but it wasn’t enough. We therefore had to split the program into a pair of mono files, and while we left the right channel alone, we drew an envelope around each trombone note in order to get it more even. This was the most extreme balance problem we had to deal with, although we did use a similar technique to pull back Page’s piano solo on ‘Flat Fee’ when it was sticking out a little too much.

“In the end, we felt we had accomplished something really unique. Colorado ’88 was one of the most involved and complicated mastering projects I’ve ever worked on. But it was also challenging and interesting, and I’m just glad that we were able to preserve, restore, and document the early days of a really fantastic band.”