Live subtitles & translations for any audio on your Mac — right from the notch. On-device, free, open source.
See the code
Live subtitles and translations for any audio on your Mac, shown in a panel that drops down from the notch.
Free, open source and fully on-device.
Click the notch and NotchSub starts listening to whatever your Mac plays (YouTube, Zoom, podcasts, streams). It shows the translation in large text with the original line underneath.
One line (recommended):
curl -fsSL https://wutlu.github.io/notchsub/install.sh | bash
Or download NotchSub.dmg and drag the app into Applications. The app isn't notarized yet, so the first time you open it, go to System Settings → Privacy & Security and click Open Anyway.
Requirements: Apple Silicon Mac (M1 or newer), macOS 15 Sequoia or later.
On first launch NotchSub downloads the Whisper model (~630 MB) and compiles it for the Neural Engine, which can take several minutes. Later launches take a few seconds. To get translations, install the language pack once in Settings → Translation pack → Download.
git clone https://github.com/wutlu/notchsub && cd notchsub
./scripts/build-app.sh --run # builds build/NotchSub.app and opens it
Requires Xcode 26 / Swift 6.2.
| Path | What's inside |
|---|---|
Sources/NotchSub/Notch | Notch panel (NSPanel), shape, click handling, subtitle UI |
Sources/NotchSub/Audio | System audio capture (process tap), 16 kHz resampling, VAD |
Sources/NotchSub/Transcription | WhisperKit engine and streaming partial/confirmed segmentation |
Sources/NotchSub/Translation | Apple Translation session |
Sources/NotchSub/Pipeline | Audio → text → translation → UI |
scripts/localizations.py | All UI strings in every language (regenerates Resources/Localization) |
design/make-icon.swift | The app icon, drawn in code (scripts/make-icns.sh) |
docs/ | Website (GitHub Pages) and install.sh |
Test the speech pipeline without using the speakers:
say -o /tmp/t.aiff "Hello, this is a test."
.build/arm64-apple-macosx/release/NotchSub --selftest /tmp/t.aiff
./scripts/release.sh 0.2.0 --publish # builds the DMG and zip, then creates a GitHub release
Issues and PRs are welcome. To add or fix a translation of the interface, edit scripts/localizations.py and run python3 scripts/localizations.py.
Live subtitles & translations for any audio on your Mac — right from the notch. On-device, free, open source.
See the code
Live subtitles and translations for any audio on your Mac, shown in a panel that drops down from the notch.
Free, open source and fully on-device.
Click the notch and NotchSub starts listening to whatever your Mac plays (YouTube, Zoom, podcasts, streams). It shows the translation in large text with the original line underneath.
One line (recommended):
curl -fsSL https://wutlu.github.io/notchsub/install.sh | bash
Or download NotchSub.dmg and drag the app into Applications. The app isn't notarized yet, so the first time you open it, go to System Settings → Privacy & Security and click Open Anyway.
Requirements: Apple Silicon Mac (M1 or newer), macOS 15 Sequoia or later.
On first launch NotchSub downloads the Whisper model (~630 MB) and compiles it for the Neural Engine, which can take several minutes. Later launches take a few seconds. To get translations, install the language pack once in Settings → Translation pack → Download.
git clone https://github.com/wutlu/notchsub && cd notchsub
./scripts/build-app.sh --run # builds build/NotchSub.app and opens it
Requires Xcode 26 / Swift 6.2.
| Path | What's inside |
|---|---|
Sources/NotchSub/Notch | Notch panel (NSPanel), shape, click handling, subtitle UI |
Sources/NotchSub/Audio | System audio capture (process tap), 16 kHz resampling, VAD |
Sources/NotchSub/Transcription | WhisperKit engine and streaming partial/confirmed segmentation |
Sources/NotchSub/Translation | Apple Translation session |
Sources/NotchSub/Pipeline | Audio → text → translation → UI |
scripts/localizations.py | All UI strings in every language (regenerates Resources/Localization) |
design/make-icon.swift | The app icon, drawn in code (scripts/make-icns.sh) |
docs/ | Website (GitHub Pages) and install.sh |
Test the speech pipeline without using the speakers:
say -o /tmp/t.aiff "Hello, this is a test."
.build/arm64-apple-macosx/release/NotchSub --selftest /tmp/t.aiff
./scripts/release.sh 0.2.0 --publish # builds the DMG and zip, then creates a GitHub release
Issues and PRs are welcome. To add or fix a translation of the interface, edit scripts/localizations.py and run python3 scripts/localizations.py.