Reduce the size of the LPK installation package

Automatically summarize the meeting title based on the content

AI models that distinguish human speech use streaming sampling and support audio of any length.

Initial store release: live recording and speaker identification, local and Lazycat Drive import, complete transcripts, speaker aliases and audio samples, summaries, notes and six export formats. Version 0.14.3 preserves the transcript header border and corners during processing and shows the application version in Settings; includes final speaker and timing correction.