For years, people have been predicting that speech recognition will soon be the primary way we interact with computers, but it’s never really taken off. Speaking at Blur, a conference in Orlando, Gary Clayton, chief creative officer at voice-recognition company Nuance, talked about what’s still holding the technology back and why he thinks everything’s about to change.

People will always want to use whatever interface is fastest and most convenient, Clayton said. On the desktop, it’s still hard to beat the keyboard. In a mobile device, however, especially in a distracted situation such as driving, there’s finally a real need for an alternative way to input information. Clayton said he felt the environment shifting when he realized that dictation was the best way for him to send an e-mail from a smart phone.

“Speech does three things really well,” Clayton said, listing “command and control,” “dictation,” and “search.” All three applications are useful when driving, he noted, and cars could be the place where the technology really takes off. As far as Clayton is concerned, voice recognition software is more than accurate enough to handle this already. But many in the audience expressed skepticism based on problems with what they’ve seen make its way into cars so far–think GPS devices with frustrating user interfaces.

Clayton acknowledged that there are still problems with many commercial voice recognition systems, and said many of them come down to issues with hardware and design. “These should be baked into cars, with frequent updates, as part of the dashboard,” Clayton said.

Another problem is that many voice recognition systems require the user to push a button before speaking. This is necessary to preserve battery life, but Clayton says, “If you have to push a button, it’s already broken.”

Once problems like this are resolved, he believes voice recognition can finally take the primary role that’s been a long time coming.