If you want to build a product with speech recognition capabilities, Nuance has been the default choice for some time. The company's technology powers Apple's Siri and Samsung's S-Voice as well as car computing interfaces from BMW, Chrysler, Ford, and many other automakers. Google has had its own voice recognition service for some time, but previously it was only used in Google-branded products like the Google app, Google keyboard, or Google.com. Now that voice recognition technology is being opened up to developers. At its NEXT cloud platform conference, Google announced the Cloud Speech API.
The new API will bring Google's voice technology to the masses, and it seems to work pretty much the way it does in Google products today. Speech is streamed up to the cloud and back in real-time, including partial "type-as-you-speak" results. The transcribed text can be dumped into an input field for voice transcription or used for a "command and control" feature, like bossing around a robot. Google's speech API can handle 80 languages and variants, while its now-rival Nuance only seems to support 38. As part of the Google Cloud Platform, we'd imagine it needs a constant Internet connection to work.
For now, the cost of the Cloud Speech API, which is only in a "limited preview," is free. Google says it "will introduce pricing in future phases." The preview seems to be invite-only, but interested developers can fill out this page and hope they get accepted.
Read on Ars Technica | Comments
No comments:
Post a Comment