scorebored / scorebored-2-alpha

Scorebored, Version 2
MIT License
1 stars 0 forks source link

Improve management/compatibility of text-to-speech feature #9

Open chugcup opened 11 years ago

chugcup commented 11 years ago

We currently are using Google's TTS service for performing the text-to-speech feature for ScoreBored announcements. Not only that, but each time the announcer speaks a message it makes a brand new GET request to google for the appropriate .mp3. This results in some lag in speech, and obvious overhead

Want to look into some options for improving the text-to-speech feature. One option might be to cache all the "static" speech .mp3s, like "Point", "Match Point", "One", "Two", ... and only make requests to Google TTS for dynamic player names. This would allow for piecing together speech fragments to announce the score/game state.

It's possible doing this might make the text sounds "robotic", but it is worth a shot to see if it improves how responsive the announcer features are.

Also want to test this across browsers to improve compatibility. One major goal of ScoreBored is to allow for working on mobile devices. We kind of agreed that we don't care about all features being implemented on older browsers, but as long as things work on recent Chrome/Firefox builds its acceptable

mike-mcgann commented 11 years ago

FYI, I just found out that Google TTS returns a 404 is a Referer is sent in the request header. So it works fine when opening the page as a file but does not work when the page is served through a web server.