I think it started a few years back with Apple’s Siri, the “intelligent” personal assistant and knowledge navigator built into, among others, iPhones. Of course, the competition quickly followed the example and today we also have Microsoft’s Cortana, Google Assistant, and Samsung Viv. They all have their quirks and strong points, but we can expect them to improve a lot in the (near) future. Surprisingly, the assistant currently having the best chances of beating them all is Amazon’s Alexa.

Implemented by the highly successful smart, voice-enabled speaker ‘Echo’ and its relatives ‘Echo Dot’ and ‘Tap’, Alexa was present in more than 30 new products shown at the 2017 Consumer Electronics Show (CES) held every year in Las Vegas. Voice-enabled here does not mean that it can also make vocal sounds audible, but that it is controlled by voice. You tell it to play Elvis, and it will play Elvis.

The success of Alexa is due to its openness. Where Siri and Cortana were initially closed, Alexa has been an open platform right from the start, actively inviting developers to add “skills” to it. A skill is a new function, like ordering a pizza hands-free or changing the color of an LED simply by saying so. Google Assistant is also an open platform, allowing third party “actions”. Forced by the success of Alexa, Microsoft and Apple have recently added developer access to Cortana and Siri too.

Contrary to Siri and Cortana who use the processing power of their host to do the job, Alexa uses Amazon’s cloud services for the heavy lifting, making it a light-weight voice-control system for all sorts of applications. Cloud-based implies an Internet connection, without one it will not work.

A system using Alexa starts listening when it hears the wake-up word ‘Alexa’ (can be changed by the user). It then starts capturing audio and streams the data over the Internet to the server where it is analyzed and interpreted. It is also stored there for later reference, to improve its voice recognition capabilities, and this is the point where privacy concerns pop up. Indeed, a device capable of streaming private conversations to the cloud where they are stored is a potential privacy threat. Authorities, or hackers for that matter, might want to tap into such an interesting database for more or less justifiable reasons.

 
Photo credit:  (c) Can Stock Photo / Subbotina, text by CPV

According to Amazon, up to now requests to access this data have always been refused, but nobody knows what may happen in the future. Users can delete their recordings, but doing so will cripple the intelligent assistant, so few people will probably do this.

Mattel’s Aristotle is an extended Alexa system with a second wake-up word ‘Aristotle’, allowing it to interact with children. It captures not only audio, but video too, streaming it to smartphones.


Are you ready to give up your privacy so you can order a pizza without dialing a number? Or are you going to join the EWF Voice Privacy Alliance?