Microsoft Discovers a ‘Breakthrough’ in Human Speech Recognition
>By James Lenhart
Before Google started their research on voice recognition it wasn’t so apparent how we would use it in our day-to-day lives. They began rolling it out in many ways, and eventually integrated speech recognition into the mobile world via Android. Speech recognition has been used for many years on telephone systems across the world, but are usually the thing you end up screaming at because it says something to like of “I’m sorry, but I am unable to understand you, please try again”.
Microsoft Research is stating they’ve found a ‘breakthrough’ in human speech recognition via a blog post, and we’re expecting to hear more details at this year’s Interspeech conference held in Florence, Italy, from Aug. 28 – 31.
Currently, speech recognition is done by either training the software over time, or pre-loading it with several phrases and vocabulary. However, it’s easy to see the flaws in this current model. Not enough data means higher failure rate, thus leaving the user discouraged and less likely to try it again. There must be a reward for the user if companies wish to use speech recognition. And the only reward you can offer is to make sure it’s extremely accurate, which is exactly what Microsoft is claiming they can do.
Check out the post to learn more, it’s a geeky read for sure.
via: Microsoft Research