To most ears, a recording of a song sounds like a reasonably accurate reproduction of the band or artist performing in front of them. Of course, most people nowadays know that many of the components of this finished product had to be played again and again until they sounded ‘right’ – and that what we hear all together on the recording probably wasn't all played at the same time. Nevertheless, a suspension of critical judgement is made and the song is perceived as a true representation of the artist's skills. But is there a magical ‘black box’ that can make anyone sound good? Since the nineties there have indeed been ‘boxes’ – in actuality, mostly software, but sometimes packaged up in a purpose-built box – that can adjust intonation and timing; but this wasn't always the case.

In the days before digital recording, when the hits I produced were made, the processes were more complicated. The method I used mostly was to ‘comp’ the vocal. This involved recording lots of takes of the vocal performance and keeping the better ones on separate tracks (we used multitrack machines usually with twenty-four tracks, but sometimes more), and then patiently sifting through each take line by line, sometimes word by word. The good ones were bounced onto another track, compiling one complete vocal track out of many. This could be an extremely tedious process at times, taking many hours to produce a three-minute vocal; but if done well, the result would be a near flawless performance that still sounded completely natural. And I have used this method even on very capable singers, because my view is that with a recording you want to hear the best version, the definitive version, one that will stand repeated listening. Even the finest singers in any genre will admit that it’s very rare for a performance to be perfect.