We’re guessing a large number of people were screaming profanities at their Nexus One handsets this weekend, not out of frustration with the phone itself but to test out Google’s apparent voice recognition censorship.  Testing of the Nexus One’s speech-transcription systems revealed that swear words are not directly translated, but instead replaced by strings of hashes.

“We filter potentially offensive or inappropriate results because we want to avoid situations whereby we might misrecognize a spoken query and return profanity when, in fact, the user said something completely innocent.  Ultimately our goal is to return results that show exactly what you said, and we’re constantly working to improve the technology to better fit our users’ needs” Google statement

Android’s speech recognition relies on Google’s servers remotely processing voice commands and then feeding back the results to the user’s handset.  Google reckon the censorship isn’t a matter of wanting to avoid legitimately fruity language, but in fact the opposite: accidentally mis-transcribing something and inadvertently sending back obscenities to a user.

[Thanks to everyone who sent this in!]

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.