macOS menubar app for retroactive audio recording with transcription.
The key idea: Ripcord runs a circular buffer in the background. When something worth keeping happens, you save it — and the audio that already happened is already there.
- Retroactive circular buffer — 1–15 min configurable; save audio that already happened
- System audio + microphone — mix into one track or split into stereo (system L, mic R) for post-production remixing
- Mic device selection — choose any input device from the menubar
- Silence auto-pause — automatically pauses recording during silence, with configurable RMS threshold and duration
- Built-in transcription with speaker diarization (via FluidAudio)
- Speaker identity persistence — voice profiles saved across sessions with automatic matching
- Video file transcription — transcribe audio from MP4, MOV, AVI, and other video formats
- Re-transcribe — re-process any recording with different settings
- Filler word removal — strip um, uh, etc. from transcripts
- Transcript formats — txt, md, json, srt, vtt
- Real-time streaming transcription in a dedicated window
- Remote control socket — Unix domain socket for external tools to receive live transcript data and control the session
- Tunable parameters — adjust diarization sensitivity, speaker count, and display options live
transcribeCLI for batch transcription — bundled inside the app, usable standalone
curl -fsSL https://raw.githubusercontent.com/unthingable/ripcord/main/install.sh | bashThis downloads the latest release, installs to /Applications, and clears the quarantine attribute.
Or download Ripcord.app manually from Releases and move it to your Applications folder. The app is unsigned, so you'll need to clear quarantine:
xattr -cr /Applications/Ripcord.app- macOS 14.2+ (Sonoma)
- Apple Silicon
On first launch, macOS will prompt for:
- Screen & System Audio Recording — required for capturing system audio
- Microphone — required for mic input
Launch Ripcord from the menubar. It immediately starts filling a circular buffer with system audio (and mic, if enabled).
- Drag the capture scrubber to choose how much buffered audio to keep (30s up to the full buffer)
- Click Record (or press Cmd+Shift+R) to save — the recording includes the selected buffer plus any new audio going forward
- Click Stop to finish — the file is saved to the output directory
The waveform shows live audio amplitude. During buffering, the highlighted region shows how much audio will be captured; during recording, the entire waveform turns red. The level meters on the right show two bars — blue for system audio, cyan for mic.
After recording, your files appear in Recent recordings at the bottom of the panel. From there you can:
- Transcribe — pick model, format, and speaker settings, then transcribe
- Copy transcript — copy the full transcript to the clipboard
- Re-transcribe — re-process with different settings
To transcribe an external audio or video file, click Transcribe file at the bottom-left of the panel. Model downloads are one-time and happen through Settings.
Bundled inside Ripcord.app/Contents/MacOS/transcribe.
transcribe <audio-file> [options]
Example:
transcribe recording.m4a --format md -o transcript.md --num-speakers 2
Requires Swift 6.0 toolchain (Command Line Tools).
git clone https://github.com/unthingable/ripcord.git
cd ripcord
make install # build, bundle Ripcord.app, copy to ~/Applications
make test # unit tests (buffer, writer, interleave)
make test-e2e # end-to-end (requires audio permissions)
swift run TranscribeKitTests # merge pipeline tests
Sources/
Ripcord/ # menubar app (SwiftUI)
TranscribeKit/ # transcription library (shared)
transcribe/ # CLI executable
Tests/
test_components.swift # unit tests
test_e2e.swift # end-to-end (requires audio permissions)
TranscribeKitTests/ # TranscribeKit unit tests