A new study from a team of researchers from Stanford, the University of Washington, and Baidu suggests that the transition to speech-based interaction with consumer devices could offer some considerable benefits to users.
The study compared text-based entry to dictation in both English and Mandarin Chinese, with the text entered via the Apple iOS’s touchscreen virtual keyboard and the speech recorded using Baidu’s Deep Speech 2 speech recognition system. The experiment found speech-based user input both faster and less prone to error, in both languages.
For English speakers, the rate at which text could be inputted through speech was three times as fast as text entry; for Mandarin speakers, the rate was 2.8 times faster. Input errors were 20.4 percent lower using speech recognition for English, and a whopping 63.4 percent lower for Mandarin. As the researchers put it in a summary, the results “show that a significant shift from typing to speech might be imminent and impactful.”
The findings help to illustrate why major IT and tech companies such as Google, Apple, and Amazon have been investing so heavily in speech recognition recently. With its ease of use and improving accuracy, voice command could become the dominant user interface for a range of devices associated with the emerging Internet of Things, and if this latest research offers any indication, that should come as a welcome development for many users.
Source: Stanford University