I’m working with an autistic singer-songwriter who cannot keep a beat to save his life. I spent a lot of time trying to help him lock in with a clicktrack, but he just couldn’t do it. So I finally just let him play, and am trying to correct timing with Audiowarp. This works fine on his vocals, and fairly well on guitar, but when I do it on his piano tracks I get a ghostly moaning sound.
I can reduce the effect somewhat by cutting an EQ notch around 250-350 hz. Does anyone have any other suggestions?
Use a midi keyboard triggering a virtual piano to capture the performance, then adjust the timing in midi.
Alternatively, if you have Melodyne or something with a similar capability, you can extract the midi information from the audio, use it to trigger a virtual piano, then adjust the timing in that.
Yeah, unfortunately if you are going to be tweaking the performance that much, MIDI is a much better tool than recording straight audio.
If it’s already recorded and it’s too late, you can try moving stuff around without stretching it and just crossfading the takes. If you do it right, it can be pretty transparent, but it is tedious. You have to put the crossfades up so that the waves line up so you don’t get strange phasing at each cut.
Thanks guys, I may see if he wants to retake it, and use midi next time. I’m afraid I’ve never really gotten used to recording/editing midi - midi tracks just behave so differently than audio…
Is it a real piano recording or keyboard? and Is it a terrible play or is it quantized on purpose?
I know cubase struggles with the time stretching , Samplitude/music maker does it a bit better. If you have it you could give that a try. Might have to deverb the piano a bit, or pull the sustain leaks back a bit for it to work
but I do like the idea of a retake on a midi keyboard.
It’s played on a Roland A-80, with audio output from a Boss DS-330
Recording midi piano makes sense when the pianist can play all the way through the song. In this case we had to stop and re-start numerous times, as he kept losing his place. And Cubase, for some reason I’ve never understood, doesn’t mute the previous take when retaking a track, and comping takes is a real pain compared to audio (or at least was last time I tried - have they improved this in recent versions?)
The artist’s speed variations are pretty drastic, and his playing is pretty clumsy overall. The project is not really for artistic value as much as it’s a social justice statement (for people with disabilities) and an expression of the artist’s own struggles with autism. I’m just trying to make him sound a little better than he sounds live…
