Siasca
A vocal studio that fits in a pocket. Import a song, pull the singer out of it, and record your own voice over what's left. Everything runs on the phone.
- Role
- Design & engineering
- Stack
- Swift 6, SwiftUI, AVAudioEngine, Core ML
- Year
- 2026
Pick a song.
Any song in Files, or any video in Photos. The audio comes out in the background and the project shows up in the list while it’s still working. Start three at once. Nothing blocks.
Pull the singer out.
Tap Separate. A model on the phone splits the track into a vocal stem and an instrumental one. No upload, no account, no server. Tens of seconds, once per song, and the editor stays usable while it runs.
Sing over it.
Hit record. Playback and the mic share one clock, so your take never drifts from the track. Trim it, split it, drag it. Add reverb and EQ. The song you sang over stays locked, so you can’t nudge it by accident.
Nothing leaves the phone.
Songs, takes and stems live in the app’s own storage. Settings shows exactly how much each takes, and one tap clears what no project uses any more.
Three apps and a lot of exporting, folded into one screen.
Sia is the front of the Thai เสียง, sǐang. Sound. Sca is the back of the Swedish skapa, to create. I grew up between Bangkok and Sweden, so the app got a name from both.