What's Next For Google Voice Actions? A Discussion With Google's Mobile Team (TCTV)


Why go to a press conference, when the press conference can come to you?

This morning Google held a press event in San Francisco to unveil a few new Android features including Google Voice Actions. Currently comprised of 13 “actions,” the sleek app allows user to verbally command their phones to perform a wide array of tasks such as to call a friend, find directions, dictate and send e-mail/text messages, call any business on Google Search, find songs, etc. As my colleague MG Siegler writes, it’s “awesomely fast and accurate” and only available on the Android.

After the event, several key members of Google’s mobile team— including Hugo Barra (Director of Product Management), Mike LeBeau (Sr. Software Engineer), Fernando Delgado (Product Manager) and Dave Burke (Engineering Manager) — dropped by TechCrunch’s SF headquarters to give us a deeper dive into Android’s flashy new toys. Barra and LeBeau, who you will probably recognize as the star of the adorably dorky demo video, discussed the future of Google Voice Actions and its technology on TechCrunch TV. See video above.

There were a few interesting disclosures:

Google believes Voice Action will be big in SMS. Although voice command web queries have seen a surge in popularity on the Android platform, with 1 out of 4 queries via voice search, Barra believes Google can achieve similar numbers on the SMS front. “I would love for it go to 1 out of 4 for SMS, right, wouldn’t it be great if this technology was so good— and I think it is— that I’ll feel comfortable sending the majority of my text messages from it.”

There will be many more Google Voice Actions available in the future. Expect the next announcement to focus on openness and integration with third party developers. “We think it is the tip of the iceberg…We’ve only done an inkling of in this release is integration with third party apps…We are definitely thinking about ways to make this as open as possible,” LeBeau says. If you want another hint for what Google has in store for Voice Actions, channel Captain Kirk. LeBeau, the architect for Voice Actions, says he was heavily inspired by Star Trek (unclear, whether he was referring to the original series, Next Generation, Deep Space Nine, or one of the feature films).

Google Voice Actions and Google Voice’s transcribing services share similar technology. Google Voice transcription (which turns voice mails into text) has been less accurate because the path from voice to app is less direct (i.e. talking directly into your phone enhances accuracy, while previously recorded content is not as clear). “They definitely share a lot of the same core technologies some of the big differences are that in the case of Google Voice mails or YouTube transcription, the circumstances that we’re trying to transcribe the audio are actually much more difficult. So a user picking up their phone, pressing a button, knowing that they’re speaking to the system, is a very different experience and much higher quality experience in terms of audio and speaker quality in terms of the way that they’re trying to annunciate and things than something you find in a voice mail or something you find on YouTube with lots of speakers…we’re sharing learnings from both of those to build core technologies that do both of those things really well,” LeBeau says.

According to Barra, Google has not started the process of making Google Voice Actions compatible on other operating systems (i.e. the iPhone), but it will happen. “We haven’t started working on bringing Voice Actions to other platforms quite yet but we are very interested in doing that. There will be constraints, there’s a lot of things we won’t be able to do on other platforms because they’re not as open or as easy to work with as Android is but as we do with pretty much every single one of our products we try to bring them to as many users as possible and of course that means going to all the platforms out there. That is work we hope to start in the near future,” Barra says.