They already have tools to cut down based on silence. So, remove the laugh track to enable this silence, then cut based on the silence. If there was a filter for ffmpeg, you could just chain them together in a single command.