STANFORD (CBS SF) – For many, using speech recognition on a smartphone has been at times a frustrating experience. Now, researchers at Stanford say the technology has advanced to a point where it is faster and more accurate than typing.
Researchers devised an experiment where participants took turns inputting about 100 different phrases, either through a standard iPhone keyboard or through Baidu Deep Speech 2 software.
Half of the 32 test subjects performed the experiment in English, while the other half conducted the experiment in Mandarin Chinese.
The researchers found the speech recognition software in English was three times faster than typing, while the error rate was more than 20 percent lower.
In Mandarin, the speech recognition software was 2.8 times faster, but the error rate was much lower (63.4 percent) compared to typing.
“So we expected speech to be slightly faster than typing on an onscreen keyboard. But we were surprised to find that it was about three times faster than typing on a keyboard,” Stanford computer science professor James Landay, the study’s co-author, said in a university video.
Landay said speech recognition has improved significantly in the last two to three years, benefiting from big data and deep learning to produce better results.
Researchers hope the results lead to better interfaces that take advantage of improving technology.