Cher can sing, but it’s commonly thought that the warbly electronic effect applied to her voice on her multi-platinum hit Believe was achieved with an infamous piece of pitch-correction software called Auto-Tune. Recording nerds debate whether it was that program or a less interesting vocoder-type gadget (Digitech Talker) that really did the business, but Auto-Tune can be used to get exactly that kind of sound if it is set up in an extreme way.

Auto-Tune is certainly used widely in a milder way to fix the shonky intonation of many a dodgy C-list celeb with a recording contract—not to mention a bigger name in the business who has managed to parlay tabloid confessions and cheeky boy-next-door ordinariness into a career of glorified karaoke accompanied by backing tracks pastiching various 80s and 90s musical styles.

Robbie Williams
Where are his lips?

Thing is, even applied with moderation, I always thought I could hear Auto-Tune compensating for singers’ shortcomings like this. There’s a “snatching” of the correct note that sounds robotic, a metallic change in the singer’s natural vibrato. Now that I have finally taken delivery of my own copy of a new computer-based pitch-correction tool, Roland V-Vocal, I realise how naïve I was. (I should point out here that I didn’t order this software specifically; it was a freebie with a larger music-recording package—not that I wasn’t itching to try it out myself.)

What I have learned, PooterGeekers, is that it’s the “Auto” part that is the weakness. Starting with a reasonable vocal, the “Tune” part can be applied manually to undetectable effect. (I realise that “starting with a reasonable vocal” is a little too much to ask from some of today’s performers, but bear with me here.)

I’ve had a sore throat over the past couple of days, so I’ve been unable to properly re-record a lead vocal I put down a few weeks back. [Yes, I split an infinitive, but it’s the best way to express what I wanted to say, grammar-nerds.] This presents an excellent opportunity to reveal to you the full horrible power of pitch-correction technology.

Do you remember my strange post a while back in which I was trying to get in touch with, er, a collaborator in France who goes by the handle “magmavander”? Magmavander lives in the south and makes all kinds of electronic music online using an amazing collection of free (that’s free as in “free beer”, but sadly not free as in “free speech”) tools called Buzz machines. We’ve been putting together a song about a Hollywood divorce, “Moving Day”—a process in which I have been something of a slow-coach; sorry, magma.

One of the many reasons for my slowness is that I’ve been sloppy about keeping noise out of those recordings, and not been very successful at cleaning them up—as you will soon hear. Having all those vocal tracks to hand is, however, useful for demonstration purposes. Listen to a (slightly re-arranged) extract of the original [MP3, 900K]. Focus your attention on the lead vocal which comes in at about 19 seconds and, in V-Vocal, looks like this:

a trace of Damians lead vocal
Damian gets a singing lesson from his computer
[click image to enlarge]

The yellow line is my original, the red line is what I should have been singing. Notice that I sang flat in the penultimate bar and, as I ran out of breath, sharp over the last few beats. Now listen to the pitch-perfect corrected version [MP3, 770K]. Can you spot any artefacts? I can’t and it’s my own voice that’s been manipulated. Evil, isn’t it? There you have it, listeners: against all odds I have found another reason to hate Robbie Williams.