A Brief History of the Shed Sessions Restoration.
written by Brian Robinson
I canít remember exactly when or how Pat Martin and I got into a conversation about his old Unicorn 10" 4-track tapes, but I was intrigued and I needed a project to get me off the streets again. He wanted them copied to CD as he was aware that the tapes were shedding oxide. The "Shed Sessions" seemed an appropriate name for my new project! An hour later Pat turned up with three Tesco carrier bags. My heart sank when I looked at the tapes and I could smell the warehouse where theyíd been stored. Upon opening the boxes I thought to myself "thereís no way these tapes will have survived twenty five years of being poorly stored, lost and forgotten". Beyond that, usually most four-track recordings made in someone's dad's garage in back 1977 wouldn't be suitable for a CD release!
There were about ten reels containing a total of 37 songs. Pat had managed to obtain an old Tascam A3440 4-track recorder complete with the usual colony of spiders. I demagnetised the heads and "lined up" the machine. I was surprised at how well it performed. I spooled up the first tape and watched the oxide fall away from the take-up tensioner! There was definitely something still on the tape, but I knew Iíd only have limited plays before the Unicorn songs ended up as a pile of brown oxide on my studio floor. I knew that it wouldnít be possible to merely transfer the songs to CD as the tape noise was prohibitive and a manual mix "on the fly" was not going to work.
I recorded the material using Cubase VST 32. I decided to record all four tracks straight onto my hard drive and do all the mixing/editing on the PC. This produced 148 separate files, which I could edit separately. I had to stop after each track to clean the heads. I regret not making notes at the time, but I do remember after a short time realising that there was a good chance that this material was salvagable. Tape noise was the first thing to deal with and I used a noise-reduction plugin within "Sound Forge". The NR plugin works by creating a "noise print" from a sample (usually taken from just after the recording starts and consisting of tape noise only) analysing the result then subtracting this from the whole track. This is much more effective than domestic Dolby, etc., because the noise is removed completely from the whole recording, not just the quiet parts.
This was the single most important element of the restoration" process, and also the easiest although there was some experimentation. Some tracks needed more processing than others did and mandolins, tambourines etc. seemed to lose a little "presence". Iím told that tape noise itself can be a psycho-acoustic enhancement with some sounds. In these cases I either manually "floored" the tape noise on either side of the instrument (tedious) or used a fast gate.
The second stage was to remove all unwanted recorded noise such as Pete Perryerís handclaps (hundreds of them - he really enjoyed the backing vocal sessions!), "pops" and breathing sounds, mechanical knocks and bumps etc. The sound editing software produces a visual "wave" that can be magnified in time and amplitude. I soon learned to recognise an intake of breath and other unwanted material and manually muted them. I had to do this on 148 tracks, from start to finish. This was extremely tedious and although I tried noise gates and other techniques nothing worked as well as my ears and eyes. This stage alone took me about a month.
The third stage was to get the levels within mixable parameters. Anyone whoís ever used a four-track recorder knows how difficult it is to get all the parts you need in the mix. For example, a mandolin, a vocal track and a harmonica were recorded on the same track - all at different levels. I selected the loud parts, reduced the levels and vice-versa. Again a labour-intensive task but I didnít want to speed too much time grappling with automation within Cubase when it came to the final mix.
The fourth stage was pre-mix. This is where the final picture started to come into focus. I went against my "prime directive" not to change the material when I noticed a whole backing vocal section missing in "Is That a Shame" so I cut and pasted an existing one. I also managed to find Pat Martinís only mistake in the whole of these sessions - a missing "D" in "Restless" lasting about half a second. I inserted the missing note using my trusty old bass and now claim my rights to "Guest Artist" status. On "Open Sea" the ending seemed a little too abrupt, so I cut and pasted an acoustic guitar (you can hear it kick in about 20 seconds before the end) and this is the lone guitar you can hear fading out after the guitar solo "outro" finishes.
Basic "structural" errors were corrected during this process but I was starting to enjoy myself as I was hearing (almost) the finished songs. I know all the words, folks!
Fifth stage - mix down. The tedious stuff was done and I was now grappling with the limitations of a four-track tape recording. There was a "bounced-down" stereo track and two tracks with backing vocals or additional instrumentation on every song. The main problems that present themselves are (1) the "bounced" stereo track is always inferior in quality to the other two tracks (2) the "bounced" stereo track is difficult to mix-fix if the stereo image is off. Using a combination of frequency and level manipulation, as well as automated facilities within Cubase I did my best to achieve a balance. I saved the songs in 16-bit format ready for final mix.
Sixth stage - final mix. This involved "trimming" the start and end of the mixed tracks. This had to wait until last as if Iíd "trimmed" individual tracks during any of the above stages, Cubase would have loaded them from the "new" start - badly out of sync! A couple of songs ended in spontaneous "jams" which were great fun but maybe only of interest to the playing, rather than the listening public. At this stage collaboration with Pat became important. Fade-outs were the only way to end some songs and he decided how and when this process should take effect.
When I started to think about "enhancements" (reverb, stereo imaging etc.) Pat and I experimented with "T-racks Finalyser" which provided some stereo enhancement and final mix level "polish" to some songs. Simple reverb treatment seemed to bring out the best in others.
A track that wasnít working was "Canadaís A Long Way". I figured that it was because the kick drum was low in the mix and not "locking" with the bass, a Unicorn trademark. I manually superimposed a "kick" over the track as e.q. didnít deal with the problem. If you listen closely you will notice that Pete didnít line up with me 100% Guest artist credit is due again!
"Rainy Season" cried out for some bad weather - my Roland Sound Canvas Pro obliged here.
I know how hard I worked on the Unicorn tapes but that alone doesnít explain how they "polished up" so well. They have an amazing history including being lost, forgotten and poorly stored over a long period of time. They have no right to sound this good after twenty five years even given the advantages of digital processing and dogged patience. Can Agfa, TDK and Maxell recording tapes have Guardian Angels??
It was labour of love that Iím pleased to see was not wasted, in the shape of the "Shed No Tear" CD release. Pat was very kind to use the word "ingenious" on the CD notes to describe my work, but it was more perspiration than inspiration. The fact that these tapes survived to be released is an amazing thing, and I am proud to have been a part of that.