Hi Bernhard!
I took another look at this and have prepared a pull-request which I just submitted to add SoundSprite functionality.
I used sox to convert, trim silence, mono, truncate and fade the samples.
Then I used a newly forked node.js package:
https://github.com/realbluesky/soundsprite - which stitches the samples together (with 0.25 second gaps) and generates compatible json file and primary audio formats.
In the pull request I know there is a bug with AudioElementSound and SoundSprite. Would you please take a look? If you let each sample finish playing then it runs fine, but if you try to start another sample while the first is still playing it seems to botch the AudioElement pool logic and reports the Object is no longer available or not able to be used.
Also, you'll notice I switched the Sound url parameter from String to dynamic. The soundsprite.json format matches the one used by
https://github.com/goldfire/howler.js/ and as such provides a list of the generated formats. So instead of adding all supportedTypes and try/catching, it simply removes the non-supported types from the array list and goes from there.
From my testing this PR has no breaking change for the existing audio implementation, but you'll want to confirm :)
I hope you like it - if it needs some cleaning let me know.
Thanks!
Alex