A fully offline recording & transcription desktop app.
Wisp captures your microphone and system audio (the other side of a call) at the same time and transcribes both on-device. Audio and text never leave your machine.
macOS 26 (Tahoe) is the primary supported target. Windows support is in preview with
Windows.Media.SpeechRecognitionand local-model setup wiring. Linux support is coming soon.
- Fully offline — Audio and transcripts stay on your device. Wisp works with Wi-Fi turned off.
- On-device transcription — Uses
SpeechAnalyzer, the new API in Apple's Speech framework on macOS. Windows preview builds can useWindows.Media.SpeechRecognitionor prepare a local model from setup. - System audio + microphone capture — Uses macOS 14.4+ Core Audio Process Taps to tap meeting-app output without prompts, mixes it with your mic input, and merges both sides into a single transcript. Windows local-model work is structured around WASAPI mic + loopback capture.
- Built in Rust with a GPU-rendered UI — The UI is built on GPUI, the framework that powers the Zed editor. Native-feeling responsiveness and smooth scrolling.
- Simple local storage — Recordings are stored as WAV and metadata as SQLite under
~/Library/Application Support/dev.mokmok.wisp/. Easy to export and analyze later.
Wisp is a small Cargo workspace with cleanly separated concerns:
| Crate / target | Responsibility |
|---|---|
apps/wisp-desktop |
GPUI desktop shell. Renders recording state and the transcript view. |
crates/wisp-core |
Shared, platform-agnostic types (Session, Segment, IDs, SourceLabel). |
crates/wisp-audiokit |
Safe Rust wrapper around platform audio/transcription backends. |
crates/wisp-audiokit-sys |
Raw C ABI bindings to WispAudioKit. |
crates/wisp-storage |
Session/segment persistence on SQLite (bundled rusqlite). |
native/WispAudioKit |
Swift package handling Core Audio Process Tap capture and SpeechAnalyzer transcription. Linked into the Rust binary as a static library. |
Roughly, data flows like this:
Core Audio Process Tap ─┐
├─► WispAudioKit ─► wisp-audiokit ─► wisp-desktop (GPUI)
Microphone input ───────┘ │ ▲
└─► SpeechAnalyzer ────────────┘
│
└─► wisp-storage (SQLite + WAV)
- macOS 26 (Tahoe) — Wisp relies on
SpeechAnalyzer, Core Audio Process Taps, and the new Metal Toolchain, so macOS 26 is required for now. - Xcode 26 — for the Swift 6.0 / macOS 26 SDK.
- Windows 10/11 preview — uses
Windows.Media.SpeechRecognitionfor the platform recognizer route and offers a setup route to download a local Whisper-family model under%APPDATA%\dev.mokmok.wisp\models. - Rust 1.96 — pinned in
rust-toolchain.toml. - Microphone and system-audio recording permissions. macOS will prompt on first launch.
A Nix flake is included, so the dev environment is one command away:
# Enter the dev shell
nix develop
# Run a debug build
cargo run -p wisp-desktopIf you'd rather use Rust + Xcode directly:
cargo build -p wisp-desktop --releaseSee .github/workflows/release.yaml for how the release .app bundle is produced — pushing a v* tag builds Wisp.app on a macOS 26 runner.
Set WISP_OUTPUT_DIR to override where recordings are written. When unset, Wisp uses ~/Library/Application Support/dev.mokmok.wisp/recordings.
- Windows support — preview setup and
Windows.Media.SpeechRecognitionroute are in place; WASAPI loopback + local-model transcription is the remaining hardening path. - Linux support — exploring PipeWire monitor sources paired with a local Whisper-family model.
- Copy transcript to clipboard and export as plain text (.txt).
- Export to Markdown / SRT / JSON.
- Speaker diarization within a single channel.
Issues and pull requests are welcome. Before sending a PR, please make sure cargo fmt, cargo clippy --workspace --all-targets, and cargo test --workspace pass under the same conditions as CI. For the Swift side, make -C native/WispAudioKit runs the equivalent checks.
TBD (will be added before public release).