Does it make any difference if you DO NOT use drag and drop to insert the voiceover onto the slide? (I'm asking because perhaps something doesn't get set up properly via the method you are using.)
Does it make any difference if you make sure to have a slight gap between the start of the slide and the start of the audio and another gap at the end of the audio before the end of the slide? (I'm asking because if there is no gap between the start and end of the audio voiceover and the start and end of the slide, then Captivate will output the audio files as one big contiguous audio file, and that may be causing a significant delay in the audio beginning to play. If there are gaps before and after the audio clip, Captivate leaves them all as separate clips and they should begin to play quicker.)