FLOSS speech recognition [LWN.net]

FLOSS speech recognition

Posted Jul 25, 2013 10:07 UTC (Thu) by stijn (subscriber, #570)
Parent article: FLOSS speech recognition

I wonder, is there a patent mine field around speech recognition, and if so, are the ramifications similar to other fields of software engineering? There must be considerable commercial interests in this particular area, and patents are the obvious levers businesses will try to push.

to post comments

FLOSS speech recognition

Posted Jul 25, 2013 14:03 UTC (Thu) by jmspeex (guest, #51639) [Link] (1 responses)

Patents are much less of an issue for speech recognition because there's no issue of compatibility. Standards attract minefields because with a single patent you can hold a company for ransom: pay us or else you can't be compatible with the standard and your product breaks. For something like speech recognition, it's hard to even tell someone is infringing (you don't know what the product is doing because it's not a public standard) and even if you did, there's no holding up for ransom because any details in the algorithm can be changed at will with no compatibility issue.

FLOSS speech recognition

Posted Jul 25, 2013 15:01 UTC (Thu) by stijn (subscriber, #570) [Link]

There are a lot of patents not to do with standards but just with functionality. For example Apple's 'bounce back' patent. There must be tens of thousands of them with a good number leading to lawsuits. I think Steve Jobs asserted Apple patented everything around Siri that they could. In compression there are a lot of patents, and the gzip authors had to be very careful around them, at a time when the current-day deluge of patents was presumably still some way off. The fear is that some stupid overly-general feature has been granted a patent.