Apple’s Siri has given the world of mobile voice-recognition software a much-needed lift, as my colleague Ryan Kim pointed out earlier this week. But voice-recognition software still isn’t the easiest stuff to use, and there are many instances when one-finger navigation is simply a better option. Unlike touchscreens — which have become ubiquitous, thanks largely to the original iPhone — Siri and other voice-recognition software aren’t going to revolutionize how most of us interact with our phones anytime soon.
Don’t get me wrong: I’ve long believed that there’s an untapped market for voice-recognition technology in mobile, where navigating a phone with fingers can still be a hassle. Players like Nuance and Vlingo have quietly gained traction with third-party apps for smartphone users, and there’s a lot to like about Siri, as the recent flood of glowing reviews demonstrates. For Siri to change how we use our phones, though, a few roadblocks must still be addressed:
- There’s a learning curve. As Kent German notes in this CNET post, users must learn how to talk to Siri to make it truly useful. Queries must be spoken slowly and clearly, and they often must be phrased very specifically. That’s to be expected, of course — we are talking about computer software, after all — but it will slow uptake, especially among users who aren’t tech-savvy.
- It’s still far from perfect. Modern voice- recognition offerings are far more accurate than they were just a few years ago, but Ed Wrenbeck, a former lead developer at Siri, concedes that users “can expect a 60–70 percent return on average” when using the technology. And while its performance can be improved as users learn how to talk to it, Siri doesn’t truly offer the artificial intelligence that helps it understand what users are looking for. It can leverage powerful tools like context and semantics to deliver personalized results, but as Wrenbeck said, “Real AI can’t fit on a phone in our world . . . yet.” Indeed, “real AI” may come to handsets years from now, but until then Siri and other technologies will evolve gradually with incremental upgrades to the technology.
- The use cases are limited. Voice recognition is ideal for some mobile scenarios, like using a navigation app behind the wheel or accessing contact information quickly (rather than scrolling through lists of names). But it’s a poor substitute for touchscreens in a business meeting, say, or on the subway. Voice will eventually become one important option for operating the phone, but it won’t be the revolutionary feature some are claiming it to be.
- Lack of integration of third-party apps. Siri is integrated with core iOS functions like email, messaging, calendar and the web, but it’s uncertain whether Apple plans to release an API that would allow third-party developers to tap into Siri’s power. Just as the real power of the iPhone lies in the vast library of third-party applications, Siri’s true potential lies in integrating with those apps. But as John Gruber pointed out last week, third-party integration invites problems of its own: When a user searches for a friend, for instance, should Siri search the iPhone’s contact book or Facebook? It’s not clear that Apple plans to offer integration or how it would address the problems that would inevitably arise.
Voice recognition is a powerful tool for specific use cases: Developers of search, navigation and messaging applications, for instance, should make the technology a top priority. And the technical wrinkles will gradually be ironed out as Apple and others develop their software. But while touchscreens have changed the way most of us use our phones every day, voice recognition will mainly be used in specific scenarios and for specific functions.