Apple’s running battle with Amazon Alexa, Microsoft’s Cortana, and Google Home, has a lot going on behind the scenes. Greg Joswiak, vice president of iOS, iPad, and iPhone Marketing, and executive Alex Acero gave Wired a peek behind the curtains, and it’s interesting as can be.
Apple’s Long-View Dilemma
The trouble with taking the long view is dealing with the perception you’re falling behind. That’s been Apple’s dilemma with AI in general and Siri specifically. While Amazon built a platform designed around training its users, Apple has been building Siri to be a service that can truly understand us, and vice versa.
That takes time, though. I’ve written and podcasted about this before, but Apple’s approach has left a lot of room for Siri competitors to appear to be better, even if they’re built on gimmicks and short cuts. Even if customers have to train themselves to speak just so before their so-called AIs do anything.
Wired‘s piece speaks directly to this, and I imagine that’s part of why Apple set it up.
Making Siri Sound Human
From the beginning, Joswiak says, Apple wanted Siri to be a get-shit-done machine. It drives him crazy that people compare virtual assistants by asking trivia questions, which always makes Siri look bad. “We didn’t engineer this thing to be Trivial Pursuit!” he says.
The Right Voices
One of the tidbits discussed in the piece is Apple’s search for voice, including firing the company everyone assumed was Nuance. That company used to handle a lot of the voice recognition for Siri, but Apple developed its own machine learning technology to take it in-house.
Apple also interviewed hundreds of people to find the right voice. Then, it broke that voice into phonemes and reassembled back into speech. Apple’s goal is to make Siri sound human, or as Wired put it, to develop a voice with, “some ineffable sense of helpfulness and camaraderie, spunky without being sharp, happy without being cartoonish.”
The full piece talks a lot more about that process, including more about phonemes, stitching them together, and different ways even the same person says “watch.”
Lastly, we got some fun facts about Siri in the piece. For instance, Siri has been localized in 21 languages for 36 countries, more than all major competitors. Combined.
Siri is used by 375 million people monthly. That’s a lot, but far less than the almost-billion active users Apple is believed to have.