Just a quick note today to mention that I’ve updated the julius-voxforge package in Ubuntu Karmic from the two-years old 0.1.1~build726 to a recent daily snapshot. This new one has been generated using much more recorded speech (but on the other hand there are also some errors in the transcriptions which haven’t been fixed yet so it’s not as good as it could be; I’ll push some more snapshots from time to time and if it comes soon enough I’ll try to get in the next stable version which is expected in some months), but especially interesting is that it supports more phonemes now (remember the old days when words like “previous” or “computer” couldn’t be used?).

Anyway, I haven’t played much with it yet so if you try it out make sure to leave a comment, any feedback is welcome. The package can be installed safely on Jaunty and my last blog post on speech recognition has some explanation on how to use it. By the way, if you’re really interested in this, after the discussion at UDS a team (with an associated mailing list) was created to work in this area, so feel free to come by there.

Well, that’s it for now. Thanks for reading, and, if you’re a native English speaker, please consider donating some speech, it only takes a few minutes! Or you may also help translating the website into your language.