My blog

The mainstream press lately often features articles from the audiophile perspective that bemoan the apparent abysmal quality of mp3 sound. Life is of course a battle of constant trade-offs. What do you do if you’ve told yourself you only like blondes, then realise a brunette’s personality makes her your potential soul mate? Or if you always sleep on the side of the bed nearest the door, only for your new beau to take up that position? What characteristics have to be in play for you to break your ‘no pets’ rule?

In business, such polarity may be less life-challenging, yet equally emotional. There’s a classic business quote I love to use (usually when someone’s trying to beat me up on price!); “Quality, Price, Service. Choose the two you want.” The inference is plainly meant to be that you can never attain all three.

The critical compromise for the music consumer is typically one of convenience over sound to achieve cheap ownership. Obvious examples abound. The Cassette. How could music lovers take to such a concept, supposedly developed as a dictation solution, which to refer to as ‘technology’ is surely an affront to pioneers everywhere? Yet millions happily tolerated hissy background noise chained with the inability to dive straight into a track of choice because we could now listen to music on the move and Tapes were cheap. DAT and Minidisc initiatives failed because we eschewed their trumpeted technological innovation for cheaper and simpler trustworthy CDs. In fact sonic historians will possibly argue that each generational dominant force of listening platform has been what the ‘experts’ would deem the inferior option at the time.

Then mp3 mania took hold. In those late-90s, rogue-Napster, days, most files were encoded to “128 bits”. Whatever the detail of this signifies is largely irrelevant as most of us were aware that this meant your enormous number of ones and zeros that made up the music file on a CD were now compressed to a tenth of their original size, allowing disturbance-free albeit often muffled sounding tunes to clog up hard rives with several lifetimes worth of listening.

So what was the response of an industry trying to encourage legitimate ownership? One plank was to offer files at a higher encoded rate. 192 and 256 mp3s became available, but crucially, these were to command a price premium. The paying customer was expected to deduce that these gave a superior listening experience. But is this true? Thanks to the invaluable Coolfer, I learned from Electronic House that “A recent informal study at the science blog Cognitive Daily revealed that out of nearly 700 volunteers, only 33 listeners could reliably tell the difference between recordings encoded at 128 kilobits per second (kbps) and 256kbps.”

I became privy to rumblings of discontent crescendo from deep-space technoville inhabitants whose day job (undertaken mainly in the dark and at night) was once musicians had left the building, to make a track, shall we say for ease, “radio-ready”.

The phrase ‘poduction’ sprang up, referring to alleged knob-twiddling wizardry which allowed mp3s to sound fantastic on even the flimsiest, tinny headphones. A laudable endeavour. Yet I discovered that many Mastering Engineers muttered that people could potentially scam off the back of this. So, I duly allowed my first ‘launch’ (featuring 3 tracks as a “B-side EP”) to pursue a new route. Given that the instructions normally metered by record labels would be so totalitarian as to make even a Russian government official wince, my Mastering guru was delighted to knock himself out to produce something he considered ‘right’. Turns out he’s in league with the Turn Me Up crew. Bringing the dynamics back to music indeed.

The amazing result, was that I could tell the difference between 192 and his output even on my ipod with only twenty-five quid Sennheiser headphones. (And that coming from someone that couldn’t hear my ‘Dinosaur’ mozzie tones when the guys were working on them much beyond 14khz.)

So what was done? Well, I’ll now attempt the near-impossible; a technical explanation in layman’s language.

The studios ‘lame’ filter of choice developers undertook research with lots of blind-listening so that they could recommend settings to achieve two outcomes:
i) make an mp3 indistinguishable from a (CD) wav file, and
ii) compress the file to a similar size to that of a 128 bit mp3.
The main tool is Variable Bit Rate (VBR). The benefits of this approach come from two main capabilities:

You get different filters. Normally, the mp3 encoding means that say, anything above 16khz will get cut off. You might think that this shouldn’t be a problem, as only toddlers and dogs can hear above this level and neither are renowned music connoisseurs. The problem though, is that this strict parameter causes non-linearities further down the line. This permits weird modulations and phase-shifts to occur. In plain English, this makes the eventual file sound inferior because the 16khz constraint invariably means in some places the ceiling falls, making for instance, parts of the track suffer only a 10khz limit.

Data rate economy. The arbitrary filter figure is not set in stone. This means that when more is going on aurally, the compression can be less. When there’s not as much action, the compression can be more. Any split-second of sound can enjoy the optimum compression that’s uniquely right for it. And with the marvels of studio software, an auto-generated VBR file can be merely a starting point, with forensic examination providing further manual enhancement opportunities.

I suppose given the earlier mantra, if such ‘high’ quality and service exists, then the price should be higher. So, to test this theory out, my task now could be to find an mp3 store that’ll allow me to upload a VBR mp3, and perhaps let me go against the e-tail grain and charge a penny more than the omnipresent 79.