Listening to some vocals I recorded yesterday, and noticing loads of out-of-tune bits that completely escaped my attention yesterday. (I find it very hard to hit notes accurately at the lower end of my vocal range.)
@brucelawson What I tend to do is record maybe 4 takes all at the same time (one after the other I mean lol) and vary things like strong or mellow tone, standing or sitting, whether I'm smoking during the take. Then chop-and-paste the best phrases together.
Then I do selective dabs of auto-tune -- specific notes only, slight envelope with tolerance so it doesn't apply at the start of any note, and isn't super strict about tuning variations, which makes it less apparent and more natural.
@siblingpastry I did 2 takes and they both sounded in tune (at the time) so I chopped the best phrases timing-wise together. Today, they sounded sooo off-tune. Autotune is too instrusive for me; I'm sure i hear robots in it (perhaps because I know it's there; I've never heard robots in your stuff for example)
@MoritzGiessmann I think it's because I didn't warm up properly and hadn't sung for a while (I had a cold so had been short of breath) but I will try that. @siblingpastry
@brucelawson I've generally found that I get better results from sitting down and singing at room-volume.
Conventional wisdom says you need to stand and project to get the best out of your diaphragm, but when I do that it just sounds a bit, well, performative.
@brucelawson The robotic sound is a resampling artifact. The way to (mostly) avoid that is to minimize how much it applies, with a combination of envelope, tolerance, and the MIDI data that triggers it. Then it only applies to the cringe notes :-)
Vocals are never precisely in time, and if the MIDI notes are quantized, the tiny timing differences can surface as audible robotic glitches. So defining the MIDI data with intentional timing 'errors', to match the audio, also helps to minimize that.
Add comment