“Intel is adding speech recognition features to its system-on-chips (SoC) via a partnership with Sensory,” Nate Swanner reports for TNW.
“Any manufacturer using Intel’s Skylake, Anniedale, Broxton, Broadwell and Merrifield SoCs can get Sensory’s TrulyHandsfree speech recognition features. Sensory says TrulyHandsfree SoCs will ship direct from Intel,” Swanner reports. “The technology features a proprietary low power sound detection, which only wakes the voice recognition engine with keywords like ‘Hey Siri.'”
“Sensory also says TrulyHansfree is accurate when there’s a lot of background noise,” Swanner reports. “the implication we can draw is that this lowers the hurdles for Apple to add Siri functionality to [Macs].”
Read more in the full article here.
MacDailyNews Take:
[Thanks to MacDailyNews Reader “Fred Mertz” for the heads up.]
I want Apple to buy Nuance and actually make quality speech recognition software for the Mac. Voice to text is the future but it just doesn’t work as well as it should. Nuance seemingly feel satisfied with incredibly slow incremental improvements. Dragon for Mac is shonkey sh*t. The Windows version is far better.
I know Apple License Nuances speech engine but I really hope they have a skunk works dedicated to kicking Nuances arse.
Apple hasn’t improved PlainTalk speech recognition in DECADES. We’re lucking that third party companies bothered to create improved voices to work with PlainTalk.
Apple’s going to pull a skunk works manoeuvre to beat out what Nuance licenses? I don’t think so. That’s like hoping Apple will come up with a decent search engine to kick Google’s ass, when they can’t be bothered to put a decent search engine on their own website.
Not Gonna Happen. *sigh*
I wonder if MDN’s Take is deliberate… (silence?)
This is something that Apple should’ve already done long ago… their own voice-processing chip. V-series.
Good idea. I wish it would happen, but unfortunately that might go against Ive’s dream of paper-thin devices.
Siri today largely depends on a 24/7 link to Apple’s (or Google’s?) servers. Putting the processing power locally onto a Mac should be doable, especially for the desktop “trucks”. Amazon’s Echo spy machine is always pumping data to their servers, so it is likely Apple will chase after that by adding it to ATV.
But for portables, I don’t see Apple improving speech. Siri remains dumb as a box of rocks — just as spell correction and text prediction remains comically bad. To do a localized Siri would be a huge design impact. The processing power, high-quality microphone, and clear speakers could all require much more battery power and would likely necessitate a thicker device. Like Ive will ever let that happen — the new Apple sacrifices features, not thickness.
Um, huh? Siri is based on Dragon speech recognition, as owned by and licensed from Nuance. Siri is going to be rewritten to work with this new-kid-on-the-block speech recognition on Macs? I smell something fishy.
I think people are misunderstanding what this is. Looking at what the company says it does, it seems they only kick the process off and provide some audio clean up. The speech to text would still be done in the cloud by whatever service they want. This just allows the speech recognition to be triggered from a low power state and could allow features like customizing the trigger phrase. Apple enabled Hey, Siri to work on the newest iPhones and iPads, Apple makes use of the M9 motion coprocessor, which sips power very conservatively to listen for the trigger all the time. To make that work in a laptop without sucking the battery, they’d want a chip. Also, I presume their chip helps process out background noise, like room noise and whatever sounds are being played by the computer at the time. So the chip doesn’t have a library or speech to text processing program. That is all done in the cloud.
You can look at their website for an explanation. http://www.sensory.com/products/technologies/trulyhandsfree/
Thank You! There are some very good illustrative videos on their website.
Apple should have bought Nuance long ago. Then Apple could have made better text-to-speech voices and speech-to-text systems, long ago. Think of a Spanish voice that could ever rival Alex in English. Yeah.
Sent from my iPhone
>
Subscribed to your weblog, thank you