Hi Product Hunt 👋

I built AudioScript because I spend a lot of time watching foreign-language podcasts and technical talks, and I wanted subtitles that just work — without sending my audio to someone else's server or paying per minute.

AudioScript listens to your computer's system audio (or your mic), runs streaming speech recognition and translation entirely on your machine, and shows bilingual subtitles in a floating overlay above any app. Chinese and English are built in, and optional downloads bring that to 29+ languages — LMT-60 or NLLB-200 for translation, Nemotron 3.5 or Qwen3-ASR for recognition. Everything still works offline.

It's free, MIT-licensed, and available for macOS, Windows, and Linux.

I'd love feedback on:

• Accuracy and latency of the multilingual models on your hardware

• Which languages or model sizes you'd want tuned next

• Anything that breaks the overlay during real use

Thanks for stopping by!