Recorded sound is an illusion, and as I think I’ve said before, stereo is doubly so. In real life what we hear is sound coming from all directions, but conventional stereo comes from just two boxes, and the change in frequency response that we experience when sound arrives from different directions is not replicated. There have been attempts to emulate this using processing but none have been entirely successful, and while binaural recordings can sound wonderful on headphones, they don’t translate particularly well to loudspeakers because each ear picks up sound from both speakers. That makes it very difficult to judge how natural a loudspeaker actually sounds. We can make all kinds of precise measurements of frequency response, distortion and phase, we can derive waterfall plots and draw graphs, but what test equipment hears and what the human hearing system perceives doesn’t always seem to entirely correlate. To really tell how lifelike a speaker sounds, we have to listen to it. And that’s where the problems start.
The human voice makes a good source for testing loudspeakers as we are very attuned to the way it sounds, so anything that isn’t quite right tends to stand out. Great, so how do we get a human voice into the loudspeaker? We set up a microphone of course. But wait a minute, how do we know how accurate the microphone is? Well, we make measurements of frequency response, distortion and phase — we plot graphs... Hang on, isn’t this all starting to sound a bit familiar? The only way to tell if a microphone really sounds natural is to listen to it, which means listening to the output from the microphone via loudspeakers or headphones. You see the problem — the microphone vouches for the loudspeaker while the loudspeaker vouches for the microphone. It’s rather like two crooks providing alibis for each other.
Does this really matter? In some ways it doesn’t as we’ve become accustomed to the way recorded music sounds when played over loudspeakers, and indeed to the way the human voice sounds when played over less-than-optimal TV speakers. If you’re exposed to anything for long enough it becomes the norm, which is why naturally played music often sounds wrong to those listeners brought up on quantised, click-driven, pitch-corrected pop. Ironically, this could mean that if loudspeaker systems and microphones do evolve to the point that they really can translate sounds in a completely natural way, the listening public might not accept the results. After all, look at the way the vinyl and analogue tape generation reacted to digital recordings.