Unreleased LP from 2021 on iDEAL, canned due to austerity measures at Boomkat. The following are the liner notes (sans footnotes) from said LP:
A Compressed History
By Jack Callahan
I started my career in electronic music at the age of 9 recording a toilet flushing with a microcassette recorder. Upon playback I discovered that I was able to make a recording slower or faster depending on the tape speed. I then played it back ad nauseum to my cousin Daniel, finding that slowing it down to half speed made it funnier. (This anecdote demonstrates to me that my initial attraction to the capturing and manipulation of sound was fueled by a visceral and pure response: laughter.) Subsequently I began to experiment with Windows Sound Recorder on our family computer (as many of my generation did), on which I could crudely speed up and slow down audio digitally, beautifully downsampling the audio with every iteration.
Around the same time I started taking music lessons, first on trumpet, followed by guitar and then drums. As I was learning to read music I got my first DAW: MTV Music Generator (a PlayStation game released on PC in 1999), bought by my father at Goodwill in 2000 for $1, with which I learned about note entry and multitrack instrumental arrangement. I then somehow found a free multitrack recorder application (the name escapes me, it was very rudimentary) and began recording myself covering songs using the musical instruments at hand with the microphone of a webcam. By the time I was in high school I had upgraded DAWs, sound cards, microphones and was recording bands for cash. Additionally I began mixing live sound at local venues and I even interned for a year at a recording studio in St. Louis (where I once met Bubba Sparxxx). Since then I have had the good fortune of getting many opportunities to hone my craft as an audio engineer on the road and in the studio.
Which brings me to this record. Since I began my life in sound at the turn of the millennium, a time when tools for and information about audio recording became available to the layman, there is one word that has become the bane of my existence as a technician, a term that is the focus of countless discourses around music in the digital age, a wellspring of confusion: compression.
Broadly speaking, compression is the process of making something smaller, as you would compress gas to fit in a smaller space. The first applies to the ratio of loudest to softest sounds, the dynamic range, and the process of reducing that range, making loud sounds softer (and conversely making softer sounds louder). This process has been around since the beginning of broadcast audio. The first dynamic range compressor was developed to protect AM radio transmitters from physical damage due to electrical spikes caused by peaks in the audio, a process that was previously controlled by a technician manually raising and lowering the level of the audio. Many analog hardware compressors developed through the years are highly sought after due to their particular behavior. If you want the “punchy” Motown bass sound you want the Fairchild 660, for The Beatles’ ''pumping'' drum sound you want the UREI 1176.
The second meaning of compression as it relates to audio is a product of the digital age: data compression of digital audio files, reducing the file size by omitting information, often using a codec (a portmanteau of coder-decoder), a program that implements an algorithm to help preserve the quality of the audio. The most well-known audio codec is MPEG-1 Audio Layer III, commonly referred to as MP3. The algorithms implemented in codecs generally utilize psychoacoustics to get rid of elements of the audio that are imperceptible to human hearing which take up unnecessary space. This type of compression is important in the storage and transmission of digital audio, especially in the context of the internet, as the smaller the file is, the faster it can be transferred and the less storage space it takes up.
These two types of audio compression are discrete processes with no relation aside from their etymologies. This however has not stopped their constant conflation in the popular unconscious. The confusion between the two types reached its apex at the intersection of two trends around the turn of the millennium: the loudness war and P2P file-sharing.
In 1994 the Waves L1 Ultramaximizer was released. Though not technically the first digital limiter (a type of extreme compressor), it was the most widely available and became an immediate hit with engineers. It utilized a process called “lookahead”, in which the limiter analyzes the incoming audio for signal peaks and adjusts the amount of gain reduction needed for the delayed output audio, allowing engineers to safely “brickwall” (increase the level toward its limit) audio without distortion to a point not previously possible, especially in the analog realm. According to mastering engineer Bob Katz, the headroom for peaks of the "hottest Pop CD" increased by 11 dB between 1990 and 2000, from -14 to -3. Journalists and engineers began calling this quest for louder records the “loudness war.” Albums that have been cited as key points in the progress of the loudness war are Oasis’s 1995 album (What's the Story) Morning Glory?, noticeably louder than any contemporaneously released album and often considered the war’s “the shot heard around the world”, and The Red Hot Chili Peppers’ 1999 album Californication, a milestone in loudness which was heavily criticized by engineers and fans alike for being so heavily limited that there is prominent and continuous digital distortion throughout the album, due in no small part to repeat excessive-loudness-offender Rick Rubin, the producer of the album who is known for strong-arming his mastering engineers into creating the loudest possible products, in this case fellow repeat offender Vlado Meller.
Californication was released on June 8th, 1999 (and it happens to be the first CD I ever purchased). 7 days earlier, on June 1st, Napster was launched to the public. Though online file-sharing was well-established by then, Shawn Fanning’s and Sean Parker’s P2P file-sharing service was massive in expanding the prominence of MP3 in the popular unconscious. Developed by Karlheinz Brandenburg and his team at the Fraunhofer Heinrich Hertz Institute in Berlin in the early '90s, the MP3 codec was a huge step in digital audio compression due to its high quality and small file size. In 1994 they publicly introduced the first software to encode WAV files (the format used on CDs) to MP3, L3enc, which was swiftly reverse engineered and distributed for free, and within a few years FTP (File Transfer Protocol) sites like MP3.com increased its popularity, especially as internet speeds increased and broadband became more widely available. By the time Napster came along, “MP3” was the most searched term on the internet.
At the time of writing this, 20 years after the notorious Napster intellectual property trials which I happened to witness in person as a young man approaching his 11th birthday on a family vacation to Washington, DC (briefly meeting Shawn Fanning in between hearings at Napster publicity event, where I told him how avid a user of his platform I was), I think back on the countless conversations through the years I have had with many intelligent people who all made the same false equivalence which go like this: “I’ve been reading that music is getting more and more compressed these days which is reducing its sound quality but I heard Neil Young’s Pono allows you to listen to uncompressed files.” Recorded music being more compressed in its dynamic range for the sake of loudness and the files being low bitrate MP3s have no relation, they are two discrete phenomena. Yes, both trends were birthed by the possibilities of digital audio, pushing the extremes of (lack of) dynamic range and file size. This context aside, the two are unrelated.
This record is an attempt to work through my personal trauma stemming from this misconception.
Side A “Data Compression”
The first track, “Suzanne Vega 'Tom's Diner' MELPe STANAG4591 codec 600 bps”, uses Vega’s song from her 1987 album of the same name, famously used by Karlheinz Brandenburg to develop and test the MP3. As he tweaked the algorithm he used the song as a control and paid special attention to how it affected the quality of Vega’s unaccompanied voice. For this track I researched and found the lowest bitrate audio codec available: mixed-excitation linear prediction (MELP), a type of linear prediction coding (LPC). The specific codec I used is MELPe STANAG-4591, was developed in 2002 by the NSA and NATO and has been adopted as the NATO standard for encrypting speech (STANTAG stands for Standardization Agreement, the NATO standards and practices). 600 bps is the lowest bitrate available (though a MELP device capable of 300 bps was developed for DARPA), which is 1/320th the bitrate of a 192 kbps MP3.
The second track, “Madonna 'Music' difference between 32 and 33 kbps MP3s with mid channel removed”, uses Madonna’s song from her 2000 album of the same name, which was leaked before its official release onto Napster. An early and notable leak, it was picked up by the media, prompting Madonna's team to issue a statement threatening legal action. I used FFmpeg, an open-source, command-line-based software for processing media files, to convert the CD quality WAV of the song into two low bitrate MP3s, one kbps apart in rate. I then flipped the phase of one and played them simultaneously so that only the differences in the waveforms could be heard. Afterward I used a mid/side processor to remove the “middle” channel, the material common to both the left and right channels.
The third track, “The Beatles 'I Am the Walrus' converted to a 128 kbps MP3 100x”, uses The Beatles’ 1967 song which was the first song I ever downloaded on Napster at the tender age of 10 years old. One could imagine a circumstance where 128 kbps MP3s (the standard bitrate at the time on Napster) are burned onto an audio CD (where they are converted to 16-bit WAVs) and given to someone to listen to (a common means of sharing music at the time). The recipient then rips the CD onto their computer and converts the WAVs into MP3s, which are effectively second generation MP3s. I searched “Napster” on Soulseek and found a user with a folder titled “Napster Downloads”, which contained a 128 kbps MP3 of “I Am the Walrus”, quite probably the same CD rip and MP3 conversion as the one I downloaded in 2000. I converted it to a 128 kbps MP3 99 more times using a custom FFmpeg script.
The fourth track, “Metallica 'I Disappear' (Napster Demo) difference between 4-bit WAVs at 1 & 1.1 kHz sample rates”, uses Metallica’s song written for the soundtrack to the 2000 film Mission: Impossible II. While still working on the song, the band learned in early 2000 that a preliminary version was being aired on US radio stations which was traced back to Napster. Upon further investigation, the band found their entire discography available for download on Napster. This prompted the notorious Metallica v. Napster, Inc. case (in which Metallica sought a minimum of $10 million in damages, at a laughably inflated rate of $100,000 per illegally downloaded song) that led to Napster’s eventual downfall. Again I used FFmpeg to convert a CD quality WAV of the song into two very low bit depth and sample rate WAVs, 100 Hz apart in sample rate. Again I flipped the phase of one and played them simultaneously.
The fifth track, “Radiohead 'Treefingers' MP3 artifacts at 96, 112, 128, 192, 256 and 320 kbps”, uses Radiohead’s song from their 2000 album Kid A. Three weeks before its official release, Kid A was leaked onto Napster and rather than being upset, singer Tom Yorke said Napster "encourages enthusiasm for music in a way that the music industry has long forgotten to do." I used iZotope Ozone’s “Codec Preview” function to solo the artifacts from converting a CD quality WAV of the song into MP3s at six of the most common bitrates. I then used a transient detector in Supercollider to randomly crossfade between the six generated files whenever a transient is detected.
Side B “Dynamic Range Compression”
The first track, “Suzanne Vega 'Tom's Diner' 5-10k de-essing artifacts”, once again uses Vega’s song, which features her solo voice. De-essing is a common type of frequency specific compression used by engineers to tame harsh sibilants of voices. I used the FabFilter Pro-DS de-esser’s “Audition” function to hear only the frequencies affected.
The second track, “Metallica 'Broken, Beat & Scarred' compressed 10:1 100x”, uses another Metallica song, from their 2008 album Death Magnetic. The album was criticized for its extreme over-compression and -limiting to the point of clipping throughout (another Rick Rubin production). In a personal email from the album's mastering engineer Ted Jensen, he claims that the "mixes were already brick-walled before they arrived" for mastering. I ran a CD quality WAV through the type of hard and fast compression used throughout the record in serial - 100 times in total - to attempt to one-up the original production.
The third track, “Daft Punk 'One More Time' difference between linear and natural phase in multiband compression at three common crossover frequencies (100, 1k and 10k) on the side channel”, uses Daft Punk’s song from their 2001 album Discovery. A famous example of their production style dubbed “French touch”, the song features heavy “sidechain” compression, where one element (in this case the kick drum) controls the amount of compression on another element (in this case everything else in the track). This creates an effect where the kick drum drastically reduces the volume of the rest of the track on every hit, referred to as “ducking.” I decided to get creative and ran a CD quality WAV of the song through a multiband compressor, a type of compressor where individual frequency bands are compressed separately from each other. I used three common crossover frequencies used for 4-way sound systems and processed the track once with the crossovers set to linear phase and once set to natural phase. I then took these two versions, again flipping the phase of one, and played them simultaneously.
The fourth track, “Red Hot Chili Peppers 'Californication' peaks with limiter at 0”, uses the title song from the Red Hot Chili Peppers’s 1999 album of the same name. As discussed earlier, the album is notorious for its compromised audio quality due to excessive brickwall limiting. I used the FabFilter Pro-L 2 limiter’s “Audition Limiting” function, which subtracts the processed output from the input audio, outputting only the audio which has been limited. I set the limiter at 0, which should normally mean that no audio is being limited, but because the CD master clips above 0 so frequently, by the end of the song you are even able to discern the melody in split-second overages.
