I've done some testing now at home on my Mac, and neither Mac Phoenix nor Mac Internet Explorer correctly display the Unicode IPA extensions. Safari displays most of them, but is missing some critical symbols, like the 'er' sound in 'her'.
It seems that a solution that works entirely correctly for the majority of browser users is really the only acceptable solution. Since most browser users use IE, just using Unicode IPA isn't really going to cut.
Since we cannot rely on browsers to correctly render IPA, we'll have to render ourself, at the server-side. Since I've done my testing, I really think option 1 from my first message is the best:
1.) Auto-detect the browser and send IPA Unicode to browsers that support it and TIPA LaTeX images to those that don't. (Pros: attractive display of IPA for all users. Cons: lots of programming)
This way, as certain browser/OS combinations come to be known to be reliably reproducing IPA, we can let them get the Unicode IPA, and everyone else gets LaTeX'ed IPA or if necessary, SAMPA.
I guess I should put my code where my mouth is and learn more about how the math TeX extensions work with the Wikipedia back end and make it go myself. In my copious free time.
On a mostly unrelated-note, perhaps explaining my obsession with the topic, I work for a company that makes TTS (Speech Synthesis) software. I work with phonetic representations of words all day long. Something that might be cool for some pages on Wikipedia and definitely for all of Wiktionary would be to have TTS-generated samples of how things are pronounced. And before you complain about how robotic and wobbly TTS sounds, you should listen to some of the most modern voices out there. They sound very natural. Check out [[Speech synthesis]] for a list of good voices with free web demos. We could probably negotiate a deal with one of the companies wherein we include their TTS samples in the Wikipedia in exchange for clearly marking where the TTS samples came from. Since it costs virtually nothing to generate the samples, it would be essentially free advertising for the company. And for all the people who see pronunciation schemes as indecipherable Greek, a good sound sample clarifies any phonetic confusion, and doesn't force poor Wikipedia users to listen to crappy home recordings of our geeky voices.
Cheers! - David [[User:Nohat]]
Steve Vertigum wrote:
This is probably the most well thought out addressing of this issue ever done on wp. I must say this is impressive and inline with the consensus of
No unicode IPA on IE?? Hmm. Well, considering the expensive workarounds you listed -- as necessary to accomodate IE users -- for a fix that entirely in Microsoft's domain, I would lean toward calling the IPA Unicode as "standard" anyway, and let the ?? or Xboxes be the problem of the IE end user. This is already the case for any character sets that arent loaded up anyway -- (I have yet to load a Hindi character set for example. ;) Soon afterward someone will write a hack to accomodate IE no doubt, but theres no reason not to push the Unicode IPA as the standard right now.
But that still doesnt deal with the problem of easy input via a Roman character set. A little conversion hack from the pseudovalues (/s/) to their IPA equvalents should be a first priority , and I would do it myself if I had the time, or could program a little better (late bloomer ok..)
As always with apologies to the hackers, -S-