Fix vocal pitch in Sonar without nasty correction artifacts.
The critics are right: pitch correction can suck all the life out of vocals. I proved this to myself accidentally when working on some background vocals. I wanted them to have an angelic, 'perfect' quality. As the voices were already very close to proper pitch anyway, I thought just a tiny bit of manual pitch correction would give the desired effect. (Well, that and a little SofTube TSAR1 reverb, which can give a lovely, diaphanous hall sound.) I was totally wrong, because the pitch correction took away what made the vocals interesting. It was a failure as a sonic experiment, but a valuable lesson, because it caused me to start analysing vocals to see what makes them interesting, and what pitch correction takes away.
And that's when I found out that the critics are also totally wrong, because pitch correction — if applied using the following techniques — can enhance vocals tremendously, without anyone ever suspecting that the sound has been corrected: there's no robotic quality, it doesn't steal the vocalist's soul, and pitch correction can sometimes even add the kind of imperfections that make a vocal sound more 'alive'.
The key to proper V-Vocal use is knowing what to avoid, and the prime directive is 'don't ever use any of the controls grouped under the Pitch Correction or Formant Control boxes' — unless you want the hard correction/hip-hop sound. Only use manual correction, and even then only if something actually sounds wrong. Avoid any labour-saving devices; don't use the line tool or the option to add LFO vibrato, but instead use the pencil tool to draw a line or draw in vibrato. It will take more effort to get the right sound (and you'll become best friends with the Undo button), but the human voice simply does not work the way pitch-correction software works when it's on auto-pilot. By making all your changes manually, you can ensure that pitch correction works with the vocal instead of against it.
One of my synth programming tricks on choir and voice patches is to add short, subtle upward or downward pitch shifts at the beginnings of phrases. Singers rarely go from 'no sound' to 'perfectly pitched sound', and the shifts add a major degree of realism to patches. Sometimes I'll even put the pitch envelope attack time or envelope amount on a controller so I can play these changes in real time. Pitch correction has a natural tendency to remove or reduce these spikes, which is partially responsible for pitch-corrected vocals sounding not quite right. So it's crucial not to correct anything that doesn't need correcting.
Consider the 'Pitch Spikes' picture, bearing in mind that the orange line shows the original pitch, and the yellow line shows how the pitch was corrected. Each note attack goes sharp very briefly before settling down to pitch, and 'correcting' these removed any urgency the vocal had. Also, all notes except the last one should have been the same pitch. However, the first note being slightly flat, with the next one on pitch (it had originally been slightly sharp) and the next one slightly sharp added a degree of tension as the pitch increased. This is a subtle difference, but you definitely notice a loss if the difference is 'flattened' to the same pitch.
In the last section, the pitch centre was a little flat. Raising it up to pitch let the string of notes resolve to something approximating the correct pitch, but note that all the pitch variations were left in and only the pitch centre was changed. The final note is an interesting case: it was supposed to be a full tone above the other notes, but the orange line shows it just barely reached pitch. Raising the entire note, and letting the peak hit slightly sharp, gave the correct sense of pitch, while the slight 'overshoot' added just the right amount of tension.
Another problem is where the vibrato runs away from the pitch, and the variations become excessive. The 'Vibrato' screen shot shows a perfect example of this, where the final held note was at the end of a long phrase, and I was starting to run out of breath. Referring to the orange line, I came in sharp and settled into a moderate but uneven vibrato, but then the vibrato got out of control at the end.
Bearing in mind the comments on pitch spikes, note that I attenuated the initial spike a bit but did not flatten it to pitch. Next came redrawing of the vibrato curve for more consistency. It's very important to follow the excursions of the original vibrato for the most natural sound. For example, if the original vibrato is going up in pitch, the redrawn version should track it, and also go up in pitch. As soon as you go in the opposite direction, the correction has to work harder, and the sound becomes unnatural. This emphasises the need to use V-Vocal to repair — not replace — troublesome sections.
Also note that, at the end, the original pitch went very flat as I ran out of breath. In the corrected version, the vibrato goes subtly sharp as the note sustains. This adds energy as you build to the next phrase. Again, you don't actually hear it as sharp, but you sense the psychoacoustic effect.
Sometimes a vocal can be perfect except for one or two notes that are really off, and you're loath to make changes. V-Vocal can do drastic fixes, but you'll need to 'humanise' them for best results.
In the 'Before & After' screen shot, the pitch dropped like a rock at the end of the first note, overshot the pitch for the second note, and finally the vibrato fell flat (literally). The yellow line in the top image shows what typical hard pitch correction would do: flatten out both notes to pitch. On playback, this, indeed, exhibited the robot vibe, although at least the pitches were now correct.
The lower image shows how manual redrawing made it impossible to tell the notes had been pitch-corrected. You should never have a 90-degree pitch transition; voices just don't do that. Rounding off transitions prevents the warbling hard-correction sound. Also note that, again, the pitch was redrawn to track the original pitch changes, but less drastically. Be aware that, often, the 'wrong' singing is instinctively right for the song, and restoring some of the 'wrongness' will enhance the song's overall vibe.
Shifting pitch will also change the formant, with greater shifts leading to greater formant changes. However, even small changes may sound wrong with respect to timbre. Click on V-Vocal's 'F' button, and you can adjust formant as easily as pitch.
In the Formant Editing screenshot, the upper image shows that the vibrato was not only excessive, but its pitch centre was higher than the pitch-corrected version. The lower pitch didn't exactly give a Darth Vader timbre, but didn't sound right in comparison to the rest of the vocal. The lower image shows how the formant frequency was raised slightly. This offset the lower formant caused by pitch correction, and the vocal's timbre ended up being consistent with the rest of the part.
When you create a V-Vocal clip, it's an entirely new clip; Sonar mutes the original, and hides it underneath the V-Vocal one. After doing your editing, turn on Track Layers and you'll see the V-Vocal clips and the original muted clips.
If you're sure the V-Vocal edits are what you want, you can delete the original clip. However, the muted vocals generally won't take up much space, so there's no harm in leaving them if you think you might want to revisit them later.
To hear these kind of pitch-correction techniques — or more accurately, to hear a song using the pitch correction techniques where you can't tell that there's pitch correction — check out the music video at www.youtube.com/watch?v=q9eb32yjGHo. This is a cover version of Mark Longworth's 'Black Market Daydreams', and there's quite a bit of touch-up on my vocals. But listen, and I think you'll agree that pitch correction doesn't have to sound like pitch correction.