Speech Recognition removed from Android
Friday, 7th March, 2008
Google have quietly removed the speech.recognition package from the Android API. I say quietly: the removal is noted in the API Diff specification for M3-RC37a to M5-RC14, released 15th Feb, but I haven’t been able to find any more public announcements – for example, it wasn’t mentioned in the m5-rc-14 release announcement. Google have also not responded to a couple of queries about android.speech.recognition on the android-developers mailing list.
Back in November, Nuance claimed to have packaged and embedded speech technology components for open-source distribution. If these components have been removed from Android, maybe they could be released independently.
[update: June 19th]
It’s outrageous that speech technology has basically been pulled from Android on the sly. I should say the official position (from Nuance) is that speech (recognition at least) has not been dropped: http://www.speechtechblog.com/2008/03/25/does-android-still-listen.
In August, Jerry Carter, Director of Speech Architecture & Standards at Nuance, will be speaking at SpeechTek 2008 about Speaking and Listening to Mobile Devices. The blurb says, “Speech technologies will be an integral part of the user interface on future phones using the Android operating system. Learn the basics for implementing Android applications that speak and listen to users.” Maybe Jerry will have news for us.
In the meantime, I see there are various efforts to port FreeTTS to Android. Does anybody know if any are bearing fruit? If there is scope for using C-based software on Android, there is also Flite. FreeTTS was based on Flite, but Flite focusses explicitly on issues like size, memory use, portability, etc.