Over the past few years, we have seen computer software achieve what had previously been considered to be impossible. Cedar's Retouch made it possible to remove noises such as creaks and door slams from the middle of complex audio recordings while, more recently, Celemony's Melodyne DNA technology has given us the opportunity to change notes within an audio recording almost as easily as we used to do in synth parts, using a MIDI grid editor.
When I first started in this business in the mid‑'80s, I was told by several development engineers that we would never have enough computing power available to be able to 'sample' real reverberant spaces, yet here we are with convolution reverbs that take up only a fraction of the processing power of which our computers are capable. It seems that in the world of computing, we should never say never, which suggests that maybe we should hang on to all those old out‑of‑time, out‑of‑tune recordings, because one day soon way may be able to fix them up in ways never previously thought possible.
If so many 'impossible' things have already been achieved, what can we expect next? Nobody knows for sure, but taking the technology behind Melodyne DNA to its logical conclusion, it may eventually be possible to extract good-quality solo parts from a mixed recording — indeed, the program's current abilities hint at as much now. We may be able to quantise sloppy timing within a complex mix as easily as we do with MIDI, and how about being able to extract a vocal line, process it to sound like a completely different singer, complete with their trade‑mark inflections, then drop it back into the same mix? I also expect to see software that can use a set of built‑in rules to automatically EQ and balance the various tracks in a mix to give you a starting point where much of the work has already been done. But there are sure to be new things just around the corner that none of us can anticipate.
However, the majority of these new inventions, with the exception of convolution reverb, are aimed at fixing problems, many of which could be avoided in the first place, or at replacing human artistic skills. Recording music is, after all, about capturing a performance based on the interaction between musicians, and as the quantise button has already shown us, the more you try to perfect the pitching and timing of a recording, the more of the human element tends to get lost. Used sensibly, all of these wonderful new repair tools can save an otherwise flawed recording, but if you want to capture a true musical performance, rather than assembling music from loops and rigidly quantised phrases (which is entirely appropriate for some genres, of course), maybe it would be best to proceed as though none of these tools existed? That way, we could use them only as a last resort when something went wrong that couldn't be fixed faster and more musically by simply playing the part again.
Paul White
Editor In Chief