Dr. Pedro J. Moren of Google talked about the process of developing speech recognition for over 30 languages including pig Latin:
Abstract--"The speech team at google has built speech recognition systems in more that 40 languages in little more than 3 years. In this talk I will describe the history of this project and what technologies have been developed to achieve this goal. I'll explore a bit some of the acoustic modeling, lexicon, language modeling, infrastructure and even socialengineering techniques used to achieve our ultimate goal, to build speech recognition systems in the top 300 languages of the planet as fast as possible." For me the technical aspect is simple but running a company is more complicated: lots of solutions to what seem to be non-trivial messiness in speech signal are simply to ignore them, as we have to constantly think about whether it's worth it to spend lots of money and time to incorporate some feature that won't even make that much of a difference. This I have never considered in academia. And apparently, even Google has to consider money!(LDC data is too expensive!)
0 Comments
Leave a Reply. |
NEWSLOG
|