I need a desktop application built in Python that runs smoothly on both Windows and macOS and converts live speech into text. I do not want to rely on cloud APIs such as Google Speech Recognition, IBM Watson, or Mozilla DeepSpeech; instead, the program should bundle or download a native/offline model so users can work without an internet connection. Core requirements • Real-time transcription that starts as soon as a microphone and audio is detected • Automatic punctuation and sensible sentence-level formatting • Consistent recognition of different English accents while remaining extensible to additional languages in the future (the goal is broad “all languages” support, starting with English) Deliverables 1. Executable installers for Windows and macOS, plus source code. 2. Clear instructions for adding or swapping language models. 3. Brief README covering environment setup, build steps, and usage. 4. A short screen-capture demo that proves real-time transcription, punctuation, and accent handling work offline. The project is complete when both platforms install without errors, launch cleanly, and transcribe speech in real time with punctuation, matching or exceeding 90 % accuracy on everyday English conversation. It should work with low latency. Should capture mic and audio and provide live transcript