I've been building IchiPitchy, a vocal editor that runs entirely in the browser, no installs or plugins.
What it does
- AI assistant — describe the sound you want in plain language and it sets the mixing parameters for you
- Vocal separation — extract clean vocals from a full mix (voice and instruments in one take)
- Pitch correction — edit notes on a piano-roll and reshape pitch contours, right down to individual notes
- Auto-tune — set a key and scale and dial in how strong the snap is
- Effects — 3-band EQ, compressor, reverb, delay, volume, and noise reduction
- Projects — save your work and pick it back up anytime
The architecture
The core decision: all live editing runs client-side. Pitch shifting, effects, and the entire preview loop happen in the browser with zero network round-trips, so editing feels instant.
The backend only handles the heavy, one-off work — rendering the final high-quality export (lossless), running vocal separation, and powering the AI assistant. Real-time interaction stays local; the expensive jobs go to a queued worker.
The interesting challenge was keeping the in-browser preview perceptually faithful to the server-rendered export — same musical result, two engines.
Try it
It's live and free: ichipitchy.com
Would love feedback from anyone who's done real-time audio in the browser, especially on the client-preview vs server-export split.
Top comments (0)