Go take a look at this page with your Android browser:

http://www.jplayer.org/HTML5.Audio.Support/

If this page shows your browser does not support the <audio> tag, Audio() object, or MP3 then you are currently sunk and this issue is NOT Adobe or Captivate's fault.

HTML5 is quite frankly a nightmare to support at the moment because everyone seems to somehow expect it to be as good or better than Flash SWF output, even though the browsers across this new landscape are a dogs breakfast of incompatibilities.

Bottom line: If you are planning to try and use HTML5 output from Captivate, TEST, TEST, TEST, TEST, in ALL browsers used by your target audience BEFORE you promise anybody anything.

UPDATE: Here's another good link for testing purposes:

http://html5test.com/

Yes but do ALL FOUR browsers actually have all the codecs they need to play that audio? 

One of the things I've noticed when reading lots of blogs and forum posts about the myriad of HTML5 issues being experienced is that even when the browsers DO supposedly support the new media tags for HTML5 they don't necessarily have everything they may need to play the files.

So, in one sense you are correct, it's NOT as simple as just looking at a test page.  But if your test suggests that the browser SHOULD be able to play the media, then you need to dig deeper to find out why it actually does not do so.

Something just occurred to me.  Captivate will normally concatenate audio clips in a project if the audio on each slide is set to begin at the very beginning of the slide.  If you delay the start of the audio clip by only a fraction of a second on each slide, Captivate will split the audio into separate small clips, one per slide.

In your HTML5 project how many audio files does it show?  Can you check to see if you have only one big audio file for the entire project or if there are numerous audio files corresponding to each slide?

If it turns out that you have only one large audio file, then perhaps your issue with the delayed start to the audio is that the browser is downloading the entire clip before starting audio playback.  Breaking up into smaller clips might give a better user experience.