franklioxygen/v2s

What it solves

v2s provides live, bilingual subtitles for macOS users during meetings, calls, streams, and videos. It eliminates the need to switch between browser tabs or full-screen captioning apps by overlaying a clean, two-line subtitle bar directly on the screen, allowing users to follow speech in one language and read its translation in another simultaneously.

How it works

The app captures audio from either the user's microphone or a specific macOS application. It uses Apple's Speech frameworks for transcription (preferring on-device recognition) and Apple Translation for on-device translation. For summarization, it leverages Apple Intelligence or an on-device extractive summary fallback. Voice activity detection is handled by the Silero VAD model running via Apple's Core ML framework.

Who it’s for

macOS users who need real-time translation and transcription of live audio sources (like video calls or streams) without leaving their active workspace.

Highlights

• Bilingual Overlay: Displays translated text on the first line and source text on the second.
• Flexible Audio Input: Can capture audio from specific apps or the microphone.
• Privacy-Focused: No cloud backend, no telemetry, and a strong preference for on-device processing.
• Menu Bar Integration: Operates as a lightweight menu bar app for quick access.
• Automatic Language Support: Dynamically detects supported languages based on the OS and model updates.