I built Connecting Captions, a live captioning and translation overlay for macOS that runs entirely on-device. Nothing goes over the network: no API calls, no audio leaving your machine, and it keeps working offline.
I wanted live translation I could use in meetings and conversations without worrying about connectivity, latency spikes, or sending sensitive audio to a third party.
The technically interesting part is a low-latency, self-correcting pipeline for speech-to-text and translation. It has two modes:
Sentence mode: buffers until a sentence boundary, then translates with full context. It’s more accurate, especially between languages with different word order. Streaming mode: translates word by word as speech arrives, trading grammatical accuracy for minimal delay.
Current state:
Platform: macOS Models: Whisper Latency: ~200 ms Languages: 25
Code: https://github.com/chrisswimlee/connectingCaptions
I’d love feedback, especially on the self-correction approach and which use cases matter most to people.