According to the YouTube blog post, translated captions require a caption track first, so not all videos will have translated captions. But once a caption track is added, there is now a handy “request translation” button that crowdsources accurate translations. Once the button is pushed, YouTube creates “first draft” caption translation documents based on Google Translator that the uploader can can edit himself or share with other translators. The documents have a nifty interface that embeds the YouTube video next to the translation so the captions have a bit of human input with regards to timing.
So while translated captions won’t be automatically generated, the amount of translated captions available on Youtube will go up significantly. Sure, a video with only a few views probably won’t have translations for less common languages, but a viral smash like Gangnam Style definitely will. As for me, I’ll finally find out what those shrieking Japanese girls in the background of cute cat videos are saying.