How to sync dual-channel transcripts via OpenAI Whisper (VAD silence stripping destroys absolute timestamps)

i am actually kind of stuck and dont know exactly how to resolve this issue with the above constraints.Thanks for the suggestions tho