A new report claims that Apple has decided to form its own in-house speech recognition team
to help create a future version of Siri, the company's popular virtual assistant. Motivation for this may have sprung from reports that Nuance, the company that currently powers the speech-recognition engine, was in talks with Samsung that may result in the Galaxy smartphone maker acquiring Nuance
. Apple has been poaching employees of Nuance and recruiting others to build a complete speech team for quite some time
, but new evidence has moved the story out of rumor status.
The hires are said to be working on a new version of Siri that does not rely on Nuance's technology, reports Wired
. The report quotes a postdoctor researcher at the University of Toronto who was wooed by Apple for the team as saying "Apple is not hiring only in the managerial level, but hiring also people on the team-leading level and the researcher level. They're building a very strong team for speech recognition research."
The "neural network" technique, also being used by Microsoft and Google in forthcoming projects, is an attempt to make processors work similarly to the way the brain communicates. Like Apple, Google uses neural network research to improve its own recognition engine, Google Now
. Microsoft is reported to be working on real-time translation
for use in a future version of Skype. Former MS researcher Alex Acero was hired away last year to be a senior director on the Siri team.
The new engine is said to be based on neural networking, and led by a number of former Nuance executives, including the former Vice President of Research Larry Gillick and current Siri Speech Project Manager Gunnar Evermann, who returned to Apple for a second time following a stint at Nuance. Experts from VoiceSignal Techologies are also said to be part of the team, which has been built slowly over time punctuated by some recent hires
. The majority of the work is centered at Apple's Boston offices.
The results of this in-house team are not likely to be seen in the short-term, with iOS 8 still relying on Nuance to provide the back-end of Siri's speech-recognition abilities. The technology has been slowly but mostly quietly improved, with more offline ability and more app integration and other features. In iOS 8, the system will be able to "listen" more actively so that users can trigger Siri by just calling its name. The forthcoming iOS update will also add Shazam music-recognition technology to Siri, as well as Siri assistance with compatible HomeKit and HealthKit products and other enhancements.