Live bilingual subtitles for meetings, calls, streams, and videos on macOS.
v2s turns microphone input or app audio into a clean two-line subtitle bar so you can follow speech in one language and read it in another without leaving the screen you are already using.
- Follow live conversations with translated subtitles pinned at the top of your screen.
- Capture from your microphone or from a specific macOS app instead of your entire system mix.
- Keep the original speech and the translated line visible together for fast context switching.
- Stay in a lightweight menu bar workflow instead of juggling browser tabs or full-screen caption apps.
- Menu bar app built for always-available subtitle access.
- Live subtitle overlay with translated text on the first line and source text on the second.
- Audio source selection for microphones and running macOS apps.
- Speech transcription powered by Apple's Speech frameworks, preferring on-device recognition.
- On-device translation powered by Apple Translation.
- Transcript summarization powered by Apple Intelligence, falling back to an on-device extractive summary when Apple Intelligence is unavailable.
- Overlay styling controls so the subtitle bar stays readable on top of real work.
v2s asks Apple's Speech and Translation frameworks which languages the current Mac supports, so the choices automatically follow OS and model updates. Regional variants are collapsed in the UI, while meaningful script variants such as Simplified and Traditional Chinese remain separate. Apple Translation availability is also checked for each source/destination pair before a session starts.
- No account, cloud backend, analytics, or telemetry.
- v2s has no cloud backend and does not send audio or subtitle text to its own servers.
- Translation uses Apple's on-device Translation framework. Some language packs may need to be downloaded first through System Settings.
- Speech recognition prefers Apple's on-device models, and v2s picks a language variant that has a local model whenever one exists.
- Some languages have no on-device model on a given Mac — this is common on Intel Macs, and for languages outside the modern Speech stack. Those run through Apple's server-based recognition, which needs a network connection, is subject to Apple's service quotas, and sends captured speech to Apple under Apple's privacy terms.
- Voice activity detection runs the Silero VAD model through Apple's system Core ML framework; v2s bundles no third-party inference runtime, and the conversion is reproducible.
brew install --cask franklioxygen/v2s/v2sv2s is not notarized by Apple yet, so macOS quarantines it after download. Clear the flag once, then launch the app:
xattr -dr com.apple.quarantine /Applications/v2s.appIf you already have v2s.app in your Applications folder from a manual install,
add --adopt so Homebrew takes over the existing copy instead of refusing to
overwrite it.
Updates arrive through the in-app updater. To let Homebrew handle them instead,
run brew upgrade --cask --greedy v2s.
- Download the latest
.app.zipfrom Releases. - Unzip and move
v2s.appto your Applications folder.
- Launch v2s — it appears as an icon in your menu bar.
- Select an input source (a running app or microphone).
- Choose your input and subtitle languages.
- Click Start.
v2s will ask for permissions on first use:
- Speech Recognition — to transcribe audio into text.
- Microphone — when using a microphone as the input source.
- Audio Capture — when capturing audio from another app.
- Speech transcription and translation require macOS 26 or newer
- Apple silicon and Intel Macs are supported. Which speech languages are available, and whether they recognize on device, depends on the Mac and the selected language.
git clone https://github.com/franklioxygen/v2s.git
cd v2s
open v2s.xcodeprojOr from the terminal:
xcodebuild -project v2s.xcodeproj -scheme v2s -configuration Debug buildTo build specifically for an Intel Mac:
swift build -c release --arch x86_64MIT