Has anyone here tried to shift vocal formants in Revoice Pro 4.0 vs. Variaudio 2.0?
I’m planning on laying down some vocals, and would like to shift their formants up (but not the pitch). My goal is to mimic the old tape trick I’ve read about where people would record vocals slower at a lower pitch, then speed up the tape to the desired pitch.
Has anyone tried to do anything like this in each, RevoicePro as well as VariAudio … and if so, would you be willing to share any thoughts about the pros vs. cons of one over the other?
The Elastique ‘Tape’ algorithms mimic that behavior. Keep in mind, that old technique didn’t independently shift the Formants. It shifted both the Pitch and the Formants up and down locked in a fixed relationship to each other. which is what the Tape algos do.
I was petrified I’d have to calculate frequencies vs. tape speeds to determine what pitch and tempo I’d have to record at so that after speeding up I’d achieve “higher formants” at just the right pitch and tempo.
I was so happy that both RevoicePro and Variaudio had a control for the formants independent of pitch!
Well i tried shifting formants in Variaudio, in Audio Process>MPEX Poly Complex and in Revoice Pro: All did poorly on a full piano piece when formants were shifted by more than 5-10% … maybe to be expected since all but MPEX is meant for single note melodies?
The only thing i can think of is slowing down my midi track so it takes twice as long, then using Elastique Tape to speed it up an octave, without preserving the formants, which is what @raino suggested, if i understood him correctly.
I’m after the effect George Martin got by speeding up the piano solo in “In My Life”.
I guess i could just find a harpsichord patch and get it done much easier haha!
This is close to the old technique, but it is missing a critical element.
When they used to slow down the tape to sing overdubs not only was the tempo getting slower, the pitch of the recorded Tracks also dropped lower. So the singer ends up singing their part a bit lower. Then when it all gets sped up to the proper tempo all the audio will play back at a higher pitch.
But when you do this with MIDI and slow that down the audio from the MIDI does not drop in pitch. Then if you record a vocal, set its algorithm to Tape, and speed the whole song up the vocal will rise in pitch while the MIDI audio will stay the same.
The easiest way to replicate the technique would be to render a temp mix of your backing tracks, set it to a Tape algorithm, and use it to record the vocals against while slowed down. Once you’ve returned to the proper tempo the vocals should be in tune with the MIDI and you can mute the mix.
I’d suggest creating a small test Project where you can play around with how this all works without risking something you care about.
FYI if you speed up a vocal tape recording to twice the speed - that’s how the Chipmunks were done. Which is a pretty extrema sounding. When The Beatles were playing with this stuff they only slowed it down a small amount (like 5-10% range).
It’s a piano part, and i cheated a bit: instead of playing it at half speed, for twice as long, I:
-dropped all midi notes one octave,
dragged the MIDI part with “size applies time stretch” so they filed twice as many bars
-used Elastique Tape at 50% on the rendered audio
I A/B’d the original audio with the Elastique Tape at 50%-processed audio:
It definitely sounded different, formant-wise, yayy!
It didn’t sound quite what like I was looking for, though. What i was looking for was the piano sound on George Martin’s piano solo on “In My Life” where he did exactly what you said … I understand he played it one octave down, and twice as slow, then after recording he sped up the tape to twice normal speed.
Probably i could get it a lot closer to that with some further processing like EQ, compression (and transient manipulation?), but i didn’t try any of that yet.
And maybe the Elastique Tape algorithm isn’t good enough to handle 50% or 100% stretch/scrunch changes? … I don’t exactly know.
Anyway, it’s been fun to play around, even if i didn’t get anything usable on tape (yet!).
Ha, that’s the example I was specifically thinking of. A large shift like that works because it makes the piano sound more like a harpsichord where the same shift on a voice will make it sound totally unnatural.
You might try the Standard Tape Algorithm & see how it sounds.
One trick when manipulating pitch by octaves is to do your pitch manipulation on a copy of the Audio (in general a good idea) and then mix in a bit of the unaltered Audio with the Pitch Shifted Audio.