The purpose of this document is to discuss and explain some of the methods and processes used during the making of the OzComps project. It will also provide information about how certain issues and potential problems were resolved.

I: SELECTING TRACKS
Each compilation set in OzComps contains four sections, viz. 1) original album, 2) bonus album, 3) megamixes (on Stage I only); and 4) B-sides/album tracks.

A different set of criteria is used to select tracks for each of these sections.

1. Original album
   Selection of tracks for this part is, of course, limited to those on the original album release. Care must be taken to correctly identify versions of tracks. In most cases, it is the 7" A-side that appears on these compilations, but exceptions have been identified. There are a small number of sets which have one or two bonus tracks appended to their original albums; most of these are non-K-Tel albums in Stage II. This has been done to "spread the load", so that the bonus material does not require more than one bonus disc.



2. Bonus albums
   A very careful procedure is followed to allocate bonus material to the compilation sets. Here are some of the steps:
   
   1. Identify time window for each album.
      The "time window" is the period of time between the earliest and latest debut of tracks appearing on the original album of each compilation set. To find this information, David Kent's "Australian Chart Books" are used. This series of books also provides the date of debut for each compilation album itself, must be the absolute latest date for the end of the time window for the album concerned. Example: For "Thru the Roof '83", the earliest debut is Donna Summer's "She Works Hard For the Money", on 1983-07-18 and the latest is "No Sense" by Cold Chisel, which debuted on 1983-10-24. This is the "time window" for the album, which itself entered the chart on 1983-12-05. The assumption is made that the album's release date is the Monday of the week prior to chart entry, so in this case the release date is calculated to be 1983-11-28.
   2. Identify contributing record companies.
      As explained in other documentation, each album has a certain mix of contributing record companies. In Stage I, this mix remains rather constant except for three changes in 1986, 1990 and 1991. Stages II and III are very different, in that some compilations released by a major record company contain tracks exclusive to that company only, or one major may join forces with another and release an album containing tracks from both companies; other albums (such as the K-Tel compilations) contain tracks licenced from multiple companies (one or two K-Tel albums license tracks from all six majors on the same album!).
      Identification of contributing record companies is usually not difficult, because this information is often printed somewhere on the album's label copy. If it is not, perusal of the tracks contained on the album should provide enough information to confidently deduce this. David Kent's chart books provide the record company label and catalogue number information which assists in this process.
   3. Gather data from charts and record collection.
      a) Use of David Kent's chartological data
      As a starting point, David Kent's chart books were scanned and the data put into an Excel spreadsheet from 1965 through to 1993. Additional chartological information from the internet was also amalgamated with this data to create a super database.
      This spreadsheet also contains data about song version, composers, publishers, remixers (if applicable) and sound recording copyright owners. This additional information is added manually as it is discovered. If a physical or virtual copy of the single is available, information about the B-side may also be added to the database.
      Tracks already appearing on original albums were identified, then the list was sorted by date and every song which reached Top 20, but did not appear on an original compilation album was selected, along with a large number of non-Top 20 songs. The next step was to identify the record company associated with each song, and then use the "time windows" already identified for each compilation to allocate bonus tracks. In most cases, this was rather straight-forward, but there have been some complications.
      
      • There appear to be a few "gaps" in licensing; at least three have been identified so far: EMI 1972, RCA 1971/72 and WEA 1975/76. The compilation albums released at these times did not contain any tracked licensed from these companies, which creates a vacuum. EMI released an "Explosive Hits" compilation in 1970, and 1971, but not in 1972. Indeed, 1972 was a very quiet year for compilations in general, as David Kent's book shows only two reaching the national charts that year.
      • Some artists were so prolific in their schedule of singles releases that they overwhelmed the charts. There are numerous examples, but three notable ones are ABBA 1975/76, Madonna 1984/85 and Bruce Springsteen, also 1984/85. At these times, these artists released so many singles that it is just impossible to include only one on each three monthly compilation album. As a result, the one track per artist policy had to be relaxed in these and similar situations.
      
      b) Non-charting records
      Many worthy tracks did not appear in David Kent's chart books; and it is also true that some tracks appearing on original compilation albums did not reach the national charts (there are many examples of records performing poorly on a national level, while being big hits in certain regional areas of Australia). By far, the most substantial research in the entire OzComps project is that of discovery and consideration of non-charting records. Although OzComps contains around 10,000 tracks, it was never intended to be an exhaustive library of everything ever released by Australian record companies. To this end, limits have been placed on how much can be included. To some degree, the numbering code system places a limit of 9 discs per set maximum, with each disc limited to the standard CD length of 80 minutes. Some sets in OzComps have reached this limit, and a couple of those would have exceeded this, but there were no more codes available to allocate the excess tracks. Most sets in Stage I seem to be around 5 or 6 discs.
      Because these records do not appear in the chartological archives, their place in time must be deduced by using their record catalogue numbers and matrix (MX) numbers to assist with approximate pressing dates, although this only works with about 90% of records. Most records pressed in Australia contain at least two numbers; these being a record company's catalogue number, and a matrix number, which identifies the master plate used to press a specific record side. Each and every side of each and every vinyl record has a unique matrix number. Most of these MX numbers are sequential, which allows the pressing date of records to be estimated in most cases to the month. CBS/WEA shared a set of MX numbers, which continued uninterrupted from the 1960s to the end of vinyl in 1991. From the 1970's, each company had a "block" of numbers assigned to it, when these numbers were used up, a new "block"was started. Festival used its own MX numbers, which began around 1960 and continued uninterrupted until at least 1993. EMI used a different format of MX numbering, which is more difficult to explain. Eventually, EMI used catalogue numbers as the basis for their MX numbers, which renders these almost useless as a dating guide. MX numbers on records were assigned by the pressing plant, irrespective of the actual catalogue number assigned by the record company. The compilation albums themselves also bear MX numbers, and these numbers add reinforcement to the date of release. They also serve as convenient "dividing lines" for time windows. A compilation album can only contain tracks from records with lower MX numbers, since those with higher MX numbers did not exist when the album was physically pressed. Volumes more could be said about record company catalogue numbering, but this shall suffice for now.
      Having decided a record's approximate date, the process becomes the same as if the record appeared in the chart book.
      
      c) Confirmation of Australian release
      Some of the records used for OzComps were not Australian pressings. Before including them in the project, research was conducted to verify that these tracks were indeed released in Australia, and the approximate date of their release here. In most cases, local release has been confirmed, but the one notable exception so far is the single "Midas Touch" by Midnight Starr, which was a Top 10 hit in both the UK and US in 1986, but extensive research has so far failed to confirm an Australian release until, ironically, an Australian-produced remix appeared in 1996. Until a local release of the original is proven, this song cannot appear on OzComps.
   
   
   
3. Megamixes
   Megamixes are mostly confined to Stage I, although a few are found on Stage II. The concept of extended mixes of songs did not become popular until the start of the 1980's, but by the middle of that decade it was very common. Even by that stage, not all tracks received this treatment. Some were released on 12" as their album version, where this was longer than the corresponsing 7" release. In other cases, the 12" single contained bonus tracks, sometimes live performances or other rarities, where a remix was not produced. And there are also cases where no 12" single was released at all. For the purposes of OzComps, a "megamix" is defined as any version of a song that is different and/or longer than the version appearing on either the original or bonus section, except for instrumental, live or dub versions. According to the criteria, the megamixes appear on the same album set as their shorter cousins.
   One issue to be considered here is whether to include 12" only tracks. The protocol followed in these instances has been to include these tracks only if a 7" version also appears on the record; then both the 7" and 12" can be included in the set. If there is no 7" version, the track is not included in OzComps.



4. B-Sides/Album tracks
   a) B-Sides
   The inclusion of B-sides and album tracks is a unique feature of OzComps. It is hoped that much new, rare and interesting material will be discovered through this part of the project.
   More than any other section, this requires the most research, the most vinyl sourcing, and is the most likely to suffer from poor quality source material, simply because most of these tracks have never been released on CD, and a large number of them will never appear on CD. The only way to obtain a copy of these tracks is to actually get a copy of the original record and do it yourself. While most records are able to yield high quality results, some limitations are evident, and some records are worn more than others, resulting in audible defects in sound quality. Digital mastering of vinyl records is also time consuming, requiring a minimum of 3 times the actual playing time of each recording.
   Identification of a record's B-side is the first step in this process. If a physical copy is able to be located, this is quite easy. Alternatively, internet research using Google or Discogs may help. Care must be taken to ensure the data applies to Australian releases; often, a single is released in different countries with different B-sides. Other things to watch out for are the same track appearing as the B-side of more than one record, and a track being originally released as a B-side, then later released as an A-side (or vice-versa!).
   Having identified the B-side of a given record, it must pass an eligibility test to be considered for OzComps:
   • The A-side must appear on another part of the album set. "Orphaned" b-sides are not allowed.
   • The track must NOT be merely a remix, dub or instrumental version of any track included in OzComps. Certain other derivative versions may be included by discretion. Instrumental B-sides not related to the A-side are eligible. Live versions are not eligible.
   • The track must not have appeared elsewhere in this project (eg, as a previously released A-side). If the track passes these criteria, in may be added to the B-sides disc in the same set as its A-side.
   
   
   b) Album tracks
   The selection of album tracks is the most subjective part of OzComps. Not only do the albums themselves need to be selected carefully, but the tracks chosen from them need to be checked to ensure they were not released as single A- or B-sides (in which case they may be eligible for inclusion elsewhere).
   A policy was required to keep these album tracks "current" within the framework of each compilation album. It was decided that 12 months was an adequate window; therefore the policy was made that a given album is permitted to spawn as many tracks to OzComps as possible within 12 months of its Australian release date.
   Review of this policy was made necessary by the realisation that many big hit albums have stayed on the charts for much longer than 12 months (some have lasted two years or more). Limiting these albums' contributions to just 12 months would have the effect of later singles released from them being included on original compilation albums or under the bonus disc policy, while other tracks would be barred because they are deemed "too old". The policy was therefore revised to read "tracks are eligible for inclusion as an album track on a given OzComps release if the OzComps album is released i) within 12 months after the date of original release of the album from which the track is taken, or ii) not more than 3 months after the end of the album's original chart run, whichever is the later. Albums which did not reach the Australian charts are in category i)".



5. Dub/instrumental versions (MP3 edition bonus material)
   A feature of many 12" singles released in the 1980's is the dub mix. These special mixes give a different perspective to the main track they are related to and are usually omitted from reissue releases. OzComps includes many dub mixes as a special bonus on the mp3 edition of the project (Stage I only). There are no plans to release the dub mixes on CD because there are usually not enough of them to justify the release of a disc dedicated to this purpose. The MP3 edition is included free when an entire album set is purchased on CD, or it can be purchased seperately.

Sourcing tracks for OzComps is not an easy task. It is not merely a matter of ripping a few CDs or downloading files from the Internet, although these methods do provide a large portion of the required material for the project.

There are two main difficulties to deal with when sourcing files. These are quality issues and version issues.

Quality refers to the medium on which the sound recording resides. Obviously, it is preferable to source a lossless digital recording, such as a CD or a WAV/FLAC file. Unfortunately, it is not so cut and dry as this. Many recordings have never been issued on digital media, and in some cases, licensing restrictions prevents them from ever being officially digitised. The next best thing is to create a digital audio file from an analog medium, such as a vinyl record. Cassette tapes may also be used, but their sound reproduction is not as good as vinyl.

Even if a recording has been issued in digital form, a lossless copy of that recording may not be available for inclusion in OzComps. Therefore, the use of mp3 files becomes necessary. The consideration of using mp3 was resisted because of the lossy property of the format, but it was unavoidable. The next hierarchy of quality is therefore bitrate of mp3. Fortunately, 320kbps is now very common (a few years ago it was not so easy to get 320kb mp3 files), but there are some files which are only available at lower bitrates. The "standard" bitrate of 128kb is the absolute minimum acceptable for mp3 files. Files encoded this low are noticeably lacking in sound quality.

Here is a question to consider... which is better: a WAV file sourced from a vinyl record, or a 320kb mp3 file sourced from a CD? While it is obvious that a WAV is far superior to the mp3 in terms of fidelity, some may argue that because the CD is natively digital, it is clinically better than any analogue medium. Others see this latter point as irrelevant.

The version issue is much more difficult and subtle. This refers to the logical test: is the recording in question identical to that issued on the Australian 7" single? This question is not always easy to answer. Some songs appear on their respective albums in exactly the same version as released on the single. In some cases, a long intro/ending may have been edited shorter by deleting or short fading. In other cases, the album version is in fact the same as the 12" single, with the 7" being an edited version of that. Still others are remixes, so in these cases the album version is not useful at all. The version of a song is not often specified on compilation CDs, and, indeed, some compilations are sometimes guilty of prematurely fading tracks in order to fit more songs on each disc, or they feature alternative versions of songs. Budget compilations are notorious for later re-recordings of hit songs, with a disclaimer along the lines of "recorded by one or more members of the original artist".

If the single version is based on the same mix as the album version, it is easy to edit the album version down, as long as any copy of the single version can be found, regardless of quality. Sources such as YouTube assist greatly with this. The video from YouTube can be downloaded and the soundtrack extracted to a WAV file. This WAV file can then be placed alongside the album version WAV in a wave editing program, which will allow the differences in the two files to be seen. The album WAV is then edited to match. Even a low bitrate mp3 is useful for this purpose. Usually, this process results in a file virtually identical to the 7" single, which is the ultimate goal of the exercise. But beware the subtleties in mixing! Cases exist where certain short phrases of music in the single version do not appear in the longer album version. Sometimes a sound effect has been added at an editing point, which does not appear in the album version. The inability to source these sections from the higher quality file presents a problem. Sometimes, it has been possible to simply "patch" the unique section in directly from the lower quality file, as long as the transition points are not noticeable. If this works, it is difficult to hear the transition, even when it is known exactly where those points are. As a result, these files become less than 100% lossless, but at least the version issue is settled.

All files have been edited to ensure they have 500ms of silence at the beginning. This is because some media players do not play the start of a file properly, or at all. The volume of each file is also adjusted by applying a sensible amount of digital compression using a hard-limited amplifying algorithm set to a maximum level of 99% (-0.16dB). It has been an unfortunate practise in recent years for many CDs to be mastered to 100% (0dB), a technique known as "brickwalling". The 99% (-0.16dB) limit is non-negotiable, and any files exceeding this limit are brought down to this level.

All filenames in OzComps are prefixed by a 2-part code inside a pair of braces, for example {113212~95}. The first 6-digits are the OzComps reference number, which are followed by a tilde (~) and a 2-character source code, which identifies the original source type for that file.

The standard 6-digit OzComps code uses the format XSSDTT, where:

• X is 1 for Stage I, 2 for Stage II or 3 for Stage III
• SS is the Set number within each Stage
• D is the Disc number within a Set
  This field is set to "X" for Dubs + Instrumentals
  This field is set to "A" for Artwork
• TT is the Track number on the Disc

• Sets use the first 3 digits, eg 113
• Discs use the first 4 digits, eg 1132
• Tracks use all 6 digits, eg 113212

• Stage I contains 95 Sets numbered 101 - 195
• Stage II contains 56 Sets numbered 201 - 256
• Stage III contains 36 Sets numbered 301 - 336

• 113000 = the entire Set 113 "Thru the Roof '83" (all 5 discs 1131 to 1135 inclusive, and 113X and 113A)
• 113200 = Disc 2 of the Set 113 "Thru the Roof '83"
• S00t00 = All discs of type t in Stage S, where:
  t = 1 for "original" Discs;
  t = 2 for bonus Discs;
  t = 3 for Megamix Discs;
  t = 4 for B-sides/Album tracks Discs;
  t = X for Dubs + Instrumentals (this is only valid for Stage I); and
  t = A for Artwork

When an OzComps track is referenced in liner notes, its reference number is quoted inside a set of braces like this: {113212}. This is an easy way to locate related tracks within the project.

After the 6-digit reference number is a 2-character source code. This code identifies the original source type of the recording contained in that file. Some of the more common source codes are:

99 sourced from a compact disc (digital lossless file) e.g. {113212~99} 95 sourced from a FLAC file (lossless source) e.g. {113212~95} 80 digitally mastered from a vinyl record (analog source) e.g. {113212~80} 60 digitally mastered from a cassette tape (analog source) <60 sourced from an MP3 file (lossy file) 22 128k MP3 from original digital source e.g. {113212~22} 36 160k MP3 from original digital source e.g. {113212~36} 45 192k MP3 from original digital source e.g. {113212~45} 52 224k MP3 from original digital source e.g. {113212~52} 56 256k MP3 from original digital source e.g. {113212~56} 58 320k MP3 from original digital source e.g. {113212~58} 32 224k MP3 from original vinyl (analogue) source e.g. {113212~32} 50 320k MP3 from original vinyl (analogue) source e.g. {113212~50} 40 digital stream capture from rara.com Refer to the separate page for a full list of source codes. Source codes will not appear in OzComps disc tracklistings, but a database listing the sources of all files will posted on a web page to be created later. OzComps Home Page This page © 2012 - 2014 d-Wizz Pty. Ltd. All rights reserved.