Summon a choir of blessed part-human part-machine voices to sing with you.
This is an autotune, harmonization, and formant-shifting synth/effect.
Here’s @nonverbalpoetry to explain it in a video:
- trans girl vocoder revolution (six)
- bon iver (fen)
- the documentary epsode Autotune from This is Pop (fen)
- not being able to sing well (both of us)
- having a hard time learning to sing well because of voice dysphoria (both of us)
We threw this together in like a week.
Norns, midi input (or grid and gridkeys mod)
Connect a microphone to the input of Norns, and a MIDI keyboard or other polyphonic controller to USB. In the main menu screen us E1 to navigate to the mixer, and lower the monitor volume to 0, since you’ll be monitoring through the script. Use headphones for cleaner input.
Now run the script. Open the Parameters menu. Set the
in range low and
in range high parameters to the furthest extents of your vocal range; this helps the pitch detection lock on to the right pitch.
Set the scale and root to your favorite key, and close the parameters menu. Sing into the microphone. A line will radiate from the center of the screen to a circle representing the note you’re singing, and a smaller line will indicate how accurately you’re singing that note. If you sing a tune, note how your voice gets pulled to the nearest in-scale note.
Now sing a note, and as you’re singing play a chord on your keyboard. Play more chords. Play along with your singing. Sing along with your playing. Enjoy.
The keys and encoders are currently unused by this script. There’s only the script parameters and your lovely voice as inputs. Probably other instruments would work I guess.
- This will not be the last script I want to build that uses this engine. It’s fun.
- I want to add an option for vocoder instead of PSOLA to the engine.
The PSOLA code was written by Marcin Pączkowski and is GPL.