Isovocal

Isovocal was built for Drop Frame's own live sets: a way to pull the vocal out of the track that is playing and lay it over the next one, or drop a track to its instrumental, without bouncing stems or adding tracks. Your clip stays a clip.

Drop it on an audio track, launch a warped Session clip and press Process. The song is separated once, at home, on your Mac: on Apple Silicon the GPU does it in about twenty seconds. It is cached from then on, so on stage the device only reads from that cache and costs next to no CPU.

Then one dial does the work. Left is the instrumental, the middle is the original track and right is the bare vocal, and every point between is usable. Map it to a controller like any Live parameter.

It reads the clip's warp markers, tempo and transpose through the Live API, so the vocal stays locked to what Live is actually playing, loops and tempo changes included. It never writes anything back: your clips and your project stay exactly as they were.