It seems that voice-controlled personal assistants are all the rage these days. Microsoft and Harman Kardon just revealed the Invoke Cortana speaker, Amazon put a screen on the Echo and called it a a Show, and Google just gave the Assistant more smarts, more features, and more places to be on. But one of the inherent drawbacks of relying on voice as a medium is that no one really speaks the language in the same way. Yes, we’re talking about accents, which has tripped up many personal assistants. Years after their initial debut, how far have these AI-powered agents come along? The folks over at WIRED made some rather comical tests to find out.
There is a reason why voice assistants like Siri, Alexa, Google Assistant, Cortana, and Bixby target the US market first. Aside from being one of the world’s biggest markets, it also makes sense to support one of the world’s most used languages first. But being the most used language means that people with different native tongues will sound differently when saying the same words or phrases. Which also means that the virtual assistants might not understand them as clearly as they do American English speakers.
The phrases WIRED used to test Alex on the Amazon Echo, Google Assistant on the Google Home, and Siri on an iPhone (sorry, no speaker yet) were meant to be both entertaining yet also indicative. Worcestershire sauce is one of the harder ingredients to pronounce, not to mention spell, and different countries have their own ways of saying it (correctly). Benedict Cumberbatch’s name trips up even native English speakers. And “Which writer wrote Westworld” is pretty much a modern-day tongue twister.
Despite being the youngest of the three, Google Assistant turned out to be better at understanding different accents. If you think about it, that’s not exactly surprising. Google has a treasure trove of technologies, data, and people working on natural language processing for years, be it text or, in this case, voice. Apple, on the other hand, has the advantage of being in this type of market far longer. This leaves Amazon at a disadvantage, with neither years of experiences nor a pool of resources to draw from.