The main project page touched on some of the procedures from the acquisition phase (see timeline).   The big challenge there was to get my hands on CD recordings of all the tracks I was looking for.  I was able to get most from friends' collections, but there were still more than forty that had to be obtained from libraries, and another handful that had to be purchased outright.  Furthermore, acquisition had to be done twice for many tracks because of the discovery of the end-of-track clipping that the ripping software was doing.  This further extended the acquisition phase. 

Late in the project, I acquired an updated version of Adaptec's Easy CD Creator software (v 3.50).  It was not known at the time whether this version suffered from the same shortcomings as the one I'd been using (v. 3.01).  I chose not to install nor experiment with the new version until the Project was complete, as the workarounds I implemented for the old software might have led to even more problems if the new software behaved differently. 

The most time-consuming aspects of acquisition were research, travel, and inspection.  I had to find out where tracks could be obtained from, go and get them, reproduce them, and take them back to wherever I got them from.  During acquisition, I listened to each track at least twice through to assure that the original disc-to-WAV extraction performed properly, and then to check that the CD-R archive-copied audio was all there. 

Before the project started, I naively thought that once acquisition was over that production would follow close on its heels.  As it turns out, the crafting of the master discs from which all distributed discs are copied was a task equal in length to the acquisition. 

The main objective of the master creation phase was to take the CD-R archive tracks made in the acquisition phase and arrange and record them in the proper order on the masters.   It turns out that this could not be managed entirely by the Adaptec software as originally thought, because some end-of-track clipping does occur even when inter-track gaps are present on the source disc.  For this reason and others, many intermediate processing steps were taken.  These steps were performed on each and every one of the 143 songs that wound up on the masters. 

First, the track was ripped from the archive CD-R on which it had been saved during acquisition.  There are nine acquisition CD-Rs, Alfa through Igloo, filled with 80s tracks.  The ripping process wrote the audio to a WAV file on the hard drive.   Then the WAV was read into CoolEdit for inspection, editing, and playback.  The most important aspect of the editing was to properly prepare the end-of-track portion of the audio for the production phase.  First, any noise burst at the track end had to be removed.  For some reason, there is something at the ends of tracks of the CD-Rs that doesn't show up in normal playback but does produce a big static burst in the ripped WAV.  This, and the dead time that follows the burst, was removed.  Further, any other dead time (period of zero audio) at the tail end of the song was clipped off, so that the end of the song coincides with the end of the file to within 50 ms. 

At this point, the level of the audio at the end of the track was inspected to ensure that it did not end "suddenly."  Audio at greater than about -73 dB from the peak level sounds unpleasant if it just cuts out.  It sounds unnatural, and can cause speakers to make a popping noise as the signal terminates.  If it turned out that the track didn't fade well, whether the fault of the original recording or clipping of the original recording, then a fade was implemented over the last two or three seconds of the audio which gradually attenuates the signal and brings it to a graceful close. 

Next, the end of the song was buffered with 1000 ms of silence.  The buffer was needed because this part of the track gets clipped off during the production phase by Easy CD Creator v. 3.01.   Tests have indicated that this buffering will assure that all of the audio on the masters gets reproduced onto the copies. 

The track was then listened to (often several times) for two purposes.  First, to guarantee that there were no skips or dropouts that would necessitate re-acquisition.   (Pops and crackle are regrettable but unavoidable on some tracks.)  Second, to assure a thorough review of the song's lyrics file.  By the beginning of the master creation phase, more than half of the lyrics files had been reviewed in some detail already.  Reviewing every file again improved accuracy and enforced a uniformity onto the style of the transcription. 

Once the listening was completed, the edited WAV file was saved back to the hard disc and then moved to temporary storage on SparQ cartridge along with other WAV files to be incorporated onto the same master.  A typical WAV file was between 30 and 55 megabytes in size, so the saving and moving of them took anywhere from 45 to 120 seconds on the 350 MHz Pentium II. 

Typically, tracks could be prepared for the masters at the rate of three tracks per hour.   After five to six hours work then, it was time to burn another master disc.   The layout was usually fairly straightforward, the most tedious part being the editing of the track names for the master jewel case notes.  The actual burn itself took about 45 minutes on the Hewlett Packard CD-Writer Plus 7200i.  The master was then labeled and the audio reviewed.  At this stage, it was sufficient to check just the beginnings and endings of each track to assure that the WAV files did not terminate abnormally.  This was also the time that the MP3 files were produced from the WAV files on the SparQ.  The MP3 files were then moved to their own SparQ cartridge for storage until the collection was complete and the MP3 disc could be produced.  Once each master's MP3s are made and the master disc checked, the WAVs were deleted from the SparQ to free room for the next master's WAVs.