Broadcast audio playout for macOS. Built for live radio, podcasts, and events — free and open source.
The on-air console: what's playing, what's next, and exactly how long you have left to talk.
Every feature exists because something goes wrong in live audio — a track that starts too slow, a gap that shouldn't be there, a licensing report due Monday.
Set a cue point on any track and the ON AIR panel counts down to it — going red in the final five seconds. Talk up to the vocals, never over them.
Per-track crossfade with linear or equal-power curves. The outgoing track fades as the next one rises — automatically, at the right moment.
Drag handles on the waveform to skip a long intro or cut an outro. The scrubber, show-ends clock, and countdowns all respect your edit.
Drop a pause anywhere and playback waits for you. Assign a looping music bed and it fades in underneath the live segment.
Segue scans each file and normalises playback to the EBU R128 broadcast standard. No gain-riding between a quiet interview and a loud jingle.
Every start, finish, skip, and fade is timestamped. Export the session as CSV for APRA, PPL, BMI, or whoever collects in your country.
Playlists auto-save and restore. Tracks on unmounted network drives are flagged and skipped — never dead air. Undo covers every edit.
Export a playlist with every trim, fade, ramp, and tag colour intact. Import it on another Mac and the show is ready to run.
No project files, no rendering, no learning curve you don't have time for before a show.
Drag files from Finder straight into the playlist. MP3, WAV, AIFF, FLAC, M4A, AAC — anything macOS can decode.
Reorder by drag, set trims and ramps in the waveform editor, insert pauses where you'll talk, tag tracks by colour.
Segue advances the playlist on its own, crossfading where you asked, stopping where you told it to wait.
Everything that matters during a live show has a key. Press ? in the app for the full list.
Segue was written entirely with Claude Code. I can't code in Swift — this project is an experiment in describing the tool I wanted and having AI build it. It runs my own shows, and I'm sharing it in case it's useful to yours.
It's presented as-is: the app isn't notarized with Apple (right-click → Open on first launch), and bug fixes go through the same conversation that built it. The source is MIT licensed — fork away.