Aggravating issue with audio and timing

Captivate 9.0.2.421

I my Captivate lesson, I have every screen has audio narration. Where there is only narration and a "click next to continue" button, I use the standard add audio to the timeline with no problems.

However, some screens are set up to simulate software interaction with a single button on the screen. For those slides, I have used the "On Enter: Play Audio" action. When the user clicks the correct button, it executes a shared action that hides any prompt text and proceeds to the next slide. I also have an invisible "catch all" button if the user does not click on the correct area. That button executes a shared action that displays an extra prompt and plays audio associated with that prompt. More importantly, it stops the slide audio if it is playing so the two audio files do not overlap.

Here's my problem:

When testing my lesson published as HTML5, if the timeline on those interaction screens is not a few seconds longer that the audio being played by the On Enter action, the button pauses are ignored and when the audio is finished, the lesson automatically moves to the next slide. This happens RANDOMLY, not always on the same slides, so it is difficult to track down and fix.

What I don't get is why audio not on the timeline still requires the timeline to be stretched out to cover the triggered audio being played.

Again, aggravating!