Microsoft has gained patent rights to a technology for censoring speech.
An automatic censoring filter, available in real-time mode or a batch mode, processes an input audio data stream containing speech and then alters “undesired words or phrases” to make them “unintelligible or inaudible”.
The software giant was awarded patent number 7,437,290 by the US Patent and Trademark Office last Tuesday (14 October).
Presumably the tech is intended to soothe the ears of sensitive souls who prefer a beep to a word such as ‘fuck’ slipping out during a live broadcast. Or in call centres, if a shirty employee goes off the rails and starts swearing at the customer, the software could censor anything deemed offensive or to not be toeing the company line.
This tech in fact moves on from the crude beep sloppily employed to mask what Microsoft repeatedly refers to as “undesired words” heard in live recordings.
Instead, the auto censoring filter uses a lattice comprising either phonemes and/or words derived from phonemes for comparison against corresponding phonemes or words included in undesired speech data.
“If the probability that a phoneme or word in the input audio data stream matches a corresponding phoneme or word in the undesired speech data is greater than a probability threshold, the input audio data stream is altered so that the undesired word or a phrase comprising a plurality of such words is unintelligible or inaudible,” said MS in the USPTO filing for the new patent.
It's not entirely clear where Microsoft's interest in this technology lies. But such a move could be looked upon favourably by the Chinese government, which consistently attempts to censor much of the web.
Indeed, in August 2007 the company, alongside Yahoo!, inked a web pact with the People's Republic government in which it agreed to record the identities of bloggers and censor content. ®