France Telecom claims patent on Opus, gets rebuffed
Reply #1 – 2013-02-06 21:55:01
Looks that way, though while I have some patent experience, I am not a lawyer. And if anyone nervous of getting even close wanted to be REALLY safe they have only to implement the simple patch suggested in that 'defending opus' link to the encoder's SILK layer code (one line change) which has negligible effect on PESQ scores around 12kbps, and virtually none from 14 to 16kbps upwards according to the graphs. It's possible they want to make the declaration of a potentially relevant patent 'just in case' it's applicable. It COULD be applicable to an encoder that is not identical to the reference encoder and does manage to implement claim 1 of that patent in a way that still decodes properly. Although, come to think of it, the part about computing the difference between the synthetic signal and the original speech signal is not what Opus is trying to do (one of its innovations), so it would probably be pretty stupid to compute that, but it might be possible to infringe if you really tried to. The IETF rules requires declarations of IP that could be relevant from companies involved, so it might be nothing more than somebody realising that it's an option to implement an Opus encoder that way. It's good to see Mozilla keep us informed so soon and make a completely convincing case that this particular patent cannot possibly apply to a decoder (because by definition a decoder doesn't have access to the speech input signal). I notice that the Opus-codec.org license page mentions a lawyer's opinion of the disclosed IP (but doesn't make it absolutely plain what date the opinion was stated - given that companies may acquire IP in the future or discover potential relevance of other IP.... external counsel Dergosits & Noah has advised us that Opus can be implemented without the need to license the patents disclosed by Qualcomm, Huawei, or France Telecom.