Tighten a vocal to the beat — sliced and moved, never stretched.
You give it two files: your vocal and your beat. Grid Polish figures out the beat's hidden grid — the exact moments notes are supposed to land at your tempo — and finds the exact instant each word or syllable of your vocal begins.
For every word that's only slightly off the grid, it makes a tiny cut just before the word starts, slides that piece onto the grid, and joins the gap with a soft crossfade. That's it.
Slice and slide, not stretch. Stretching audio to fix timing bends it, and it starts to sound warbly or robotic. Cutting and moving keeps every syllable exactly as you recorded it — same tone, same texture.
Only the small stuff. Anything far off the grid is left alone on purpose. That's usually swing, a pickup, or a style choice, and snapping it would kill the feel. You control how far it's allowed to move and how strongly.
Just before the word. Cutting right before a word starts (not in the middle of it) is what keeps consonants crisp and stops clicks.
It doesn't change pitch, fix wrong notes, or clean up noise. It needs the vocal alone and the beat alone, exported from the same project so they start together. And nothing is uploaded — it all runs in your browser, on your device.
Use the vocal alone and the beat alone, exported from the same project so they start at the same moment. WAV works best.
Strength is how far each word moves toward the grid (100% = fully on it). Max nudge is the biggest move allowed — anything further off is assumed to be on purpose and left alone.
The two players above are quick preview mixes, volume-matched so the comparison is fair. “Download polished song” is the finished track: your vocal and beat together at their original levels, just re-timed (if the two together would clip, it's lowered just enough to fit). The vocal-only file is there if you want to drop it back into your project.
Everything runs inside your browser. Your files are never uploaded anywhere.