Apple's Siri gets smarter with AI and a natural, 'expressive' voice

Dive Brief:

Apple’s Siri is getting several new artificial intelligence-based enhancements, including the ability to conduct language translations and better understand what users say and want, according to Apple VP Craig Federighi, who introduced the new features Monday during the company's annual Worldwide Developers Conference in San Jose, California, CNET reports. The translation feature will support English translations to Chinese, French, German, Italian and Spanish. More languages should come later.
The Siri enhancements, which will be released as part of iOS 11 this fall, will also allow third-party programmers to use Apple AI abilities including speech recognition and image processing, according to the report.
Siri in iOS 11 will also be more proactive and have a less robotic voice. "We’ve used deep learning to create a really natural and expressive voice for Siri," Federighi said, VentureBeat reports.

Dive Insight:

Advances in natural language processing could mean big things for smart speakers like Apple's new HomePod, Amazon Echo and Google Home. As language becomes more natural and devices become smarter, digital assistants are likely to become more useful and potentially a native addition to both homes and businesses alike.

Though Apple appears a bit behind in natural language processing improvements and home-based smart speakers relative to its peers, experts say the company is working hard to get the tech right. It’s possible that with the latest Siri advancements, Apple may leapfrog some of its competitors.

But Apple does have some work to do in that respect. An April assessment of digital personal assistants conducted by Stone Temple found Google Assistant on Google Home appeared to be the smartest, answering 68.1% of questions, with more than 90% of the answers complete and correct.

Siri, meanwhile, answered just 21.7% of the questions with a 62.2% accuracy rate. Still, considering a digital personal assistant that could talk to users was a pipe dream not long ago, the tech has already come a long way.