Most of the discussion around digital music has been piracy this, fair use that, and copyright control whatever.
As visual artists continually fight to stay relevant in the face of an increasingly non-visual medium, what's often ignored is the music actually stored in all those files. To songwriters' delight and engineers' grief, it's not the songs themselves people are forgetting, but the actual sound coming through all those headphones and computer speakers.
As MP3, AAC, and a host of other file formats overtake the compact disc as the preferred format for new music releases, more and more engineers and mastering technicians are finding they have to overcompensate for the perceived inferiorities of digital audio - yet many amateurs and inexperienced professionals are also compensating for the change in format in the entirely wrong manner. It's widely known that the key limitation of any digital audio compression format is how it handles the full spectrum of audio frequencies. The general tradeoff with MP3s, for example, is at the high end of the spectrum, with the elimination of any sound at frequencies at and above the 16 kHz register. The thinking in this is that some humans can't hear above this level, so it's safe to remove those sounds; the problem is that most humans still can hear at this range. Engineers know this, and also use this to accentuate a lot of higher-register instruments; flutes, violins, and upper registers of pianos, for example. In some respects, people are only getting 80 percent of the true audio.
The benefits of smaller file size and greater file portability notwithstanding, a lot of artists are finding their work reduced to a shell of their intended design in final form because of limitations in audio file design. What no one mentions is the fact that, much in the way no one would ever record for vinyl knowing full well they're releasing to CD, it's important for the modern engineer to know exactly how to approach the new version of the old medium. While there are no hard and fast rules for how to make a great-sounding MP3, there are some key considerations that every engineer can take into account.
One of the longstanding truths and shames of the recording world is that most listeners will not be enjoying the final product on equipment that can even hold a candle to high-end, professional-grade monitors. Despite the attention and care put into giving released music the highest possible quality, most of the efforts go unnoticed once people plug their $20 headphones into consumer-grade sound cards. So what's an engineer to do? Same thing they've always done: test it out across output devices, but even more than before.
For example, the most common media playing software programs are the following, in order of popularity: Apple's iTunes, Microsoft's Windows Media Player, VideoLAN's VLC, and AOL's Winamp. Each of these in turn use a proprietary coder and decoder (codec) to play back digital audio files, each tweaked for different responsiveness at different frequencies and, in the case of iTunes and Windows Media Player, to work best with their makers' portable music devices (the iPod and Zune, respectively). Each also supports a different suite of file formats, although the MP3 remains the only universally accepted format despite its serious limitations and loss-intensive compression algorithm.
The self-respecting engineer is probably using some very nice monitors to mix their tracks to perfection, but he or she also knows where those tracks are going to end up and tests out accordingly. Often referred to as "the shitty speaker test," most engineers will try out rough mixes in different settings and on different players; the key with a digital product in mind is to also play out through as many codecs and devices as possible. The differences between each media program are superficially subtle, but important nonetheless. Both iTunes and the iPod, for example, are quieter than their competition by design, but are also far and away the most popular media management and playback program. This only goes to magnify the mixing engineer's dilemma: who exactly are we tweaking all of this for? And why bother if it's all going to get lost in the conversion to weak formats anyway?
The solution, partly, is found by moving further up the chain of recording. Something's going to inevitably get lost each step of the way, so it becomes more and more important to perfect the recording stage rather than relying on the old philosophy of "we'll fix it when we mix it." In a way, this almost forces mixing to get back to its roots of audio enhancement rather than audio correction, and also turns mastering into a complementary step instead of just a ritual of album-wide correction (and industry-wide competition).
Speaking of which, mastering engineers are doubly punished by the move to digital audio files. Normally, theirs is an act of applying compression, or making the soft parts a little louder and making the loud parts a little softer, ultimately enforcing a tolerable fluctuation in volume throughout a track and throughout an album. These loud/soft dynamics are what ultimately give tracks their punch (the technical term is, in fact, "punchiness"), and compression is the great tool of mastering engineers.
However, converting a normal master of any audio track to almost any digital music format result in further compression of an already compressed audio track, meaning the gap between low and high is squeezed even more after the extreme highs and lows are chopped out of the track. Couple this with more and more bands' insistence on making their final tracks as loud as possible to stand out from other songs on radio, television, and other broadcast media, and you get to the root of the most common complaint of modern music. People aren't just saying that a lot of songs are all structurally and lyrically identical, they're saying they actually share audiological properties and elicit a similar reaction through uniform dynamics. Quite literally, they all really do sound the same.
With this in mind, there are three general approaches to get around this problem. One is to master with the high-end listener in mind and screw everyone else. The other is to actually apply less compression than desired to the final product under the assumption that the average listener or digital distributor is going to apply some compression of their own through their choice of codec.
The third, considerably more expensive and time-consuming option, is to actually create multiple masters to account for their ultimate medium. An album slated for release to both the iTunes store and to CD outlets, for example, could be mastered to both the relatively high fidelity of compact disc (320 kbps) and the comparatively low fidelity of Apple's AAC format (128 kbps). The iTunes-friendly master would sound at least slightly different than the CD master, but the pressed disc and compressed file could theoretically sound alike enough for even the most discerning listener to let slide.
The major issue with laying down any rules on mixing and mastering is just that: there are no rules. Engineers in the booth and at the console know what sounds good to them individually, and for a long time listeners' buying habits generally supported that. With the advent of inexpensive portable audio and widely varying distribution channels, the new complication is not only understanding what's going to sound good to other people but also how to make it sound good in a way that's consistent with where and how they're going to hear it.
Once upon a time, a nine-piece string section only had to sound good through studio monitors and listeners took it upon themselves to catch up to that sound. With the advent of digital audio distribution and plummeting prices on consumer audio devices, the tables have mostly turned, and suddenly that nine-piece doesn't sound as good when people are listening to it on their cell phone. With time, engineers may agree upon standards for their finished products, but for now the only surefire way to get it right is to employ the same time-honored process that's got them this far: listen, tweak, repeat.
Artwork by Joseph Devens