v2.3.3

Improvements

• Progress reporting during audio mixing is now smoother and more accurate
• Custom transcription servers no longer require an API key during configuration, e.g. for servers inside your own network
• Fixed an error that caused custom servers to reject empty audio segments
• Fixed schema validation failures when custom servers return slightly different response formats from OpenAI
• Toasts with buttons in the upper right are now clickable - previously toasts in the wrong location would drag the window instead of clicking
• Fixed a repeated error toast when opening or saving the transcript editor for a meeting with no detected speech