TwincaptionsChoose models that fit your PC.
Windows desktop application
Twincaptions is a Windows desktop application, using the Windows WebView2 runtime for its interface. Current qualification was performed on the development Windows 11 system. A clean Windows installation and other Windows versions still need separate acceptance testing; this page does not certify every PC.
CPU recognition and translation
Parakeet is the included CPU recognizer, paired with compatible local translation packs. RAM and storage requirements depend on the installed models, active applications and language combination. An 8 GB low-power PC is a design target, not a guarantee of fluent bilingual performance or a measured minimum specification.
NVIDIA GPU recognition
Whisper large-v3 needs a compatible NVIDIA CUDA card and at least 5,800 MiB of free video memory when it loads. Its optional pack is about 3.7 GB and downloads after Setup on eligible systems. Other GPU applications share that memory. The app checks availability; a listed model does not mean it fits every card. See the visual model comparison.
Storage and first download
The current public Setup is about 1 GB. Allow additional space for installation and the optional roughly 3.7 GB GPU pack; installed footprint varies. Check the Download page for the exact release size and checksum.
Languages and practical setup
Available language combinations depend on both recognizer and translator support. Start with the included English–Spanish workflow, then install compatible packs for other supported directions. Test a non-sensitive sample with your actual call audio source before relying on the display.