Although Julius is only distributes with Japanese Acoustic Models, the VoxForge project is working on creating English Acoustic Models for use with Julius, and other Free and Open source Speech Recognition Engines (disclaimer: I am the VoxForge admin).
I wonder how this will affect Speech Recognition on Ubunto. There are some people on the Ubunto Speech Recognition page ( https://wiki.ubuntu.com/SpeechRecognition ) who believe that, from an accessibility perspective, Dragon Naturally Speaking on Wine is the only option available.
>Could this be why some people still prefer newspapers and magazines to the Web?"
To answer the question... I prefer newspapers to the Web because the size of articles (and their position in relation to the front page) gives me a sense of the importance of a story. With Web articles, you can have an almost unlimited amount of information attached to a topic - some/most of it of questionable value. Because newspaper real estate costs so much, reporters are required to get to the point succinctly, otherwise the newspaper would go out of business. If a story is worthy, it gets more room. And as a reader, you can decide at a glance whether you want to read the whole thing. You don't get that sense with web articles.
Makes sense... the fact that Google released its n-gram models (a training corpus of one trillion words from public Web pages) last November (for a nominal price to researchers) seems to support the assertion that although Google might have a few people working on Speech Recognition, developing their own engine is not part of their core strategy.
Slashdot Mirror
Julius Speech Recognition Engine
Although Julius is only distributes with Japanese Acoustic Models, the VoxForge project is working on creating English Acoustic Models for use with Julius, and other Free and Open source Speech Recognition Engines (disclaimer: I am the VoxForge admin).
I wonder how this will affect Speech Recognition on Ubunto. There are some people on the Ubunto Speech Recognition page ( https://wiki.ubuntu.com/SpeechRecognition ) who believe that, from an accessibility perspective, Dragon Naturally Speaking on Wine is the only option available.
>Could this be why some people still prefer newspapers and magazines to the Web?" To answer the question ... I prefer newspapers to the Web because the size of articles (and their position in relation to the front page) gives me a sense of the importance of a story. With Web articles, you can have an almost unlimited amount of information attached to a topic - some/most of it of questionable value. Because newspaper real estate costs so much, reporters are required to get to the point succinctly, otherwise the newspaper would go out of business. If a story is worthy, it gets more room. And as a reader, you can decide at a glance whether you want to read the whole thing. You don't get that sense with web articles.
Makes sense ... the fact that Google released its n-gram models (a training corpus of one trillion words from public Web pages) last November (for a nominal price to researchers) seems to support the assertion that although Google might have a few people working on Speech Recognition, developing their own engine is not part of their core strategy.