The current methods of embedding pronunciation audio clips seem to be pretty primitive, eg on https://en.wikipedia.org/wiki/Tamil_Nadu there's:
The tiny audio icon links to the File: page of the audio clip, but the much larger "Tamil pronunciation" link goes directly to the .ogg source file, which doesn't play in Safari or IE/Edge as it bypasses our player compatibility stuff.
We may want a special thumb mode, or just a better convention for using existing thumb modes... or we may need to add JS to handle the direct .ogg links and bring up the player widget ... or something.