Thanks to Marco Arment I’m now aware that Neil Young is promoting the most recent in a well-intentioned, but misguided, quest for a better audio format. These come from a true sense that most people these days don’t hear audio in all its glory and, worse, that it can be very hard for even people who do have good equipment to get recordings that make the most of it.

Unfortunately, all these formats, for the most part, solve the wrong problems. This excellent article by Monty from Xiph.org, and even more so the followup video,¹ go into the technical details as to why. In extreme brief, yes it is, in fact, possible to perfectly reproduce any analog waveform you might want with 16 bits sampled at 44.1 kHz (CD quality). Properly conducted listening tests bear this out. So just increasing the sampling frequency won’t get you anything (and might even hurt).

So why am I sympathetic, then? Because most available music is made in such a way that it doesn’t come close to exploiting the quality available in “CD quality” sound. This means that music mastered for a new—scarequote—audiophile format might end up being better, even if the format isn’t any better, simply because the people doing the mastering know it is being targeted at people who want to play the audio on good equipment in a quiet room and get the most out of it.

The biggest problems with audio today come from a combination of the loudness war and the fact that engineers, rightly, target their music for playback on a wide variety of equipment, with the knowledge that many, perhaps most, will listen to it through pack-in earbuds and car stereos. High dynamic range is not your friend under these circumstances, and if you most of your listeners won’t hear the subtle bits unless you compress them² it makes sense to do that and sacrifice some quality for the tiny percentage of listeners sitting in a quiet room with big speakers. This is a large part of the reason why vinyl LPs can sometimes sound better than CDs (beyond issues of subjective preferences). They’re often a different master, encoded in an inferior medium, but targeted at superior equipment.

So what would be a better solution? The problems of targeting and preference are never going to go away. In fact if a new, high resolution format were to take off, it would eventually get just as bad as CDs, because engineers would stop being able to make assumptions about the people who buy it. But we can now do much of that work on the fly at the time of playback. What I’d like to see is a format that encoded (at standard 16 bits/44khz and with high quality, high bitrate, lossy data compression) uncompressed, neutrally balanced, audio but was designed to play back with presets based on the equipment used. So the default could apply compression and an engineer supplied equalizer setup that mimicked current audio recordings, targeted at small speakers in loud environments. Play it back on a computer or phone without knowing anything or changing any defaults and nothing changes. However, you could dig in, uncheck a box, or select “high quality headphones” or “large speakers” or something of the like³ and you get the version with full dynamic range and equalizer tuning for those situations.

The advantage of this is that it doesn’t take any more space than we currently use and people who don’t care can keep on not caring and still buy the same music. Sound engineers, who no doubt would love to listen to music not mixed for $10 earbuds, can make the mix that they would like to listen to available without producing a version that will get them complaints from 90% of their audience. And we consumers who would like to be able to put on a pair of nice headphones or sit in our quiet living rooms with big speakers, and listen to every bit of the music we can get out of them, can play the same music through a car stereo and hear a mix that’s designed to make the right compromises.