williamcopper
3) the best 'detunes' are those that take into account the harmony --- if your target melody is on a note that is the third of a major triad, go for detune down significantly --
4) if your target melody is on a note that is the fifth of a major triad, detune up a very small amount ---
5) if you are on a minor chord and your target melody is on the third detune up significantly.

I assume you're suggesting these directions to put the 'de'tuned voice closer to just intonation? I'd never thought about this before but it does make a lot of sense. But is it practical to do this on a per-note basis? I wouldn't expect so.

Well ... think of it this way: is it practical to rely on musicians' ears?  Surely.    Way better, anyway, than anything that is basically random, digital, and non-musical.
 

williamcopper
3) the best 'detunes' are those that take into account the harmony --- if your target melody is on a note that is the third of a major triad, go for detune down significantly --
4) if your target melody is on a note that is the fifth of a major triad, detune up a very small amount ---
5) if you are on a minor chord and your target melody is on the third detune up significantly.

But that isn't really a "fattening" process, is it?  What you are describing is "just intonation", the principle that the most "in tune" pitches are not what an equal tempered tuner says is "in tune".  If I am not mistaken, you can tell Melodyne to apply just intonation for the key center you are working in.
 
Harmonies will really pop when they are lined up with "just intonation".  This is why Barbersop quartets sound different from other vocal groups.  They really work those harmonies for just intonation.  As I understand it, the "fattening" effect is really the opposite of getting the pitch clean and perfectly in tune.  Fattening actually spreads out the pitch as if a bunch of tone deaf musicians are going at it (which might be exactly the effect one wants to achieve.)
 

williamcopper: I think you have a different definition of 'practical' to me! Normally, detuning or adding some sort of doubling or chorus is a single effect you apply to a whole track. You're suggesting applying different pitch shifts to different notes. I don't doubt that it might sound better in some circumstances but I also expect it will take 100x longer. Is it worth it? Maybe if you're tweaking a chorus of a pop song that has to be completely perfect.
 
cparmerlee: If a second track is added and the pitch is slightly different to the original then it will perform the purpose of fattening up the sound. Moving it in the direction of just intonation rather than in an arbitrary direction improves the chance that it'll sound good. If you need multiple detuned voices then maybe it would work well to cluster them between the equal temperament pitch and the just intonation pitch. Again, I'm not sure whether it's worth the effort though.

How to do a quick, simple, and effective Automatic Double Tracking effect.

Kylotan
You're suggesting applying different pitch shifts to different notes. I don't doubt that it might sound better in some circumstances but I also expect it will take 100x longer. Is it worth it?

Melodyne does this automatically.  You can fix an entire track with one mouse swipe.

 
In Melodyne, identity the root scale, then select Tuning - Just intervals.  Then click on Correct Pitch.  Done.
 
But again, just intonation and fattening are two different things -- almost the opposite of one another.  Just intonation makes the harmonics (overtones) of the triad line up arithmetically, such that each note amplifies the next.  That's how barbershop groups get their chords to pop out.  Fattening goes the opposite direction, making the pitch center LESS clear.  In the old days, we simply double tracked (Recorded each take twice) or literally flanged the track (ping-pong the track, riding your thumb on the flange of the tape to create intonation variations.)

The difference between the manual method and using Waves Doubler is that the latter lets you modulate the shift amount, and modulate each "doubled" copy separately.
 
This is supposed to mimic the random interplay of double-tracked vocals. It doesn't. Real vocals are neither random nor sinewave-modulated, and this is no more realistic than randomly moving drum hits in an effort to "humanize" a drum track. That doesn't mean there's no place for the effect; just recognize that it's a gimmick and decide if the performance benefits from it or not. I'd urge you to not adopt it as part of your default vocal treatment.
 
What the pitch shift does is generate intermodulation distortion. Normally, IMD is considered a bad thing, but for vocals it can work as long as the effect is kept subtle. IM distortion results in new frequencies being injected into the vocal, comprised of the sum and difference frequencies between the two parts. Perceptually, we hear those new high frequencies as a glossy sheen atop the vocal. As long as we can't zero in on them (because they're low volume and moving around) they aren't perceived as separate tones, much the way reverb modulation prevents us from hearing individual resonances.
 
I've gotten both good and ugh-puke results from this method. Depends on the vocal, and how natural-sounding you want it to be. If you find that your 1-cent (1 cent = 1% or 1/100th of a semitone) adjustment is too subtle, start raising the pitch amount one cent at a time until you can clearly hear the effect. Then back it off by 1 cent. You may find the sweet spot at 3 cents or even 7.
 
I would also suggest a band-pass filter on the cloned-and-shifted parts. Most algorithms introduce artifacts that are mostly in the high frequencies, and low frequencies don't benefit from it and only get muddy. Treat your "doubled" tracks' EQ like you would a reverb send. As with a reverb send, it may also help to route your doubled parts to a separate bus so you can easily set their levels relative to the main vocal. Most of the time, you'll want them 3 to 12 dB down.
 
Have fun!

He needs that 30 yr old 910 Harmonizer! 
If the Boz Imperial Delay wasn't so goofy expensive (I guess cheap compared to the 910 hardware), I'd recommend that too.

bitflipper
The difference between the manual method and using Waves Doubler is that the latter lets you modulate the shift amount, and modulate each "doubled" copy separately.
 
This is supposed to mimic the random interplay of double-tracked vocals. It doesn't. Real vocals are neither random nor sinewave-modulated, and this is no more realistic than randomly moving drum hits in an effort to "humanize" a drum track. That doesn't mean there's no place for the effect; just recognize that it's a gimmick and decide if the performance benefits from it or not. I'd urge you to not adopt it as part of your default vocal treatment.
 
What the pitch shift does is generate intermodulation distortion. Normally, IMD is considered a bad thing, but for vocals it can work as long as the effect is kept subtle. IM distortion results in new frequencies being injected into the vocal, comprised of the sum and difference frequencies between the two parts. Perceptually, we hear those new high frequencies as a glossy sheen atop the vocal. As long as we can't zero in on them (because they're low volume and moving around) they aren't perceived as separate tones, much the way reverb modulation prevents us from hearing individual resonances.
 
I've gotten both good and ugh-puke results from this method. Depends on the vocal, and how natural-sounding you want it to be. If you find that your 1-cent (1 cent = 1% or 1/100th of a semitone) adjustment is too subtle, start raising the pitch amount one cent at a time until you can clearly hear the effect. Then back it off by 1 cent. You may find the sweet spot at 3 cents or even 7.
 
I would also suggest a band-pass filter on the cloned-and-shifted parts. Most algorithms introduce artifacts that are mostly in the high frequencies, and low frequencies don't benefit from it and only get muddy. Treat your "doubled" tracks' EQ like you would a reverb send. As with a reverb send, it may also help to route your doubled parts to a separate bus so you can easily set their levels relative to the main vocal. Most of the time, you'll want them 3 to 12 dB down.
 
Have fun!

U mek Beepster's brane gooderer.
 
:-)

@bit: Rightclick-save :)
Thanks!