Robots Weekly 🤖: Voice is Coming 🗣️
Months back, Google announced their new Duplex product and
terrified a lot of people received a lot of attention. Haven’t heard it yet? Check this out.
One of them’s a robot! In a nutshell, Duplex is a very human sounding AI that can call to book restaurant and spa/salon reservations on your behalf. Let’s give it another listen.
Crazy, right? I played it for my fellow Blue Ioners (Ionites? Ions?) and proceeded to blow their minds. And also terrify a few of them.
Duplex has lead to a few interesting questions:
1. Does it pass the Turing Test? ⚗
Short answer: no. Slightly longer answer: it can only function in a few narrow domains at the moment (one not mentioned above is calling businesses to ask about business and holiday hours to automatically update Google My Business listings). And even in those it’s easy to imagine it could quickly become derailed if the business deviates from the predicted script. The Turing Test requires a machine to trick a human into thinking it is also human when conversing about anything and everything.
2. Is it ethical? ⚖
Gotta be honest, this one surprised me. It seems a lot of people think this is unethical and Google is tricking people into thinking they are talking to a human. I understand the concern, but are we expecting that robot voices will remain artificial forever?
3. Was the demo even real? 🐀
Apparently a lot of tech demos are faked. I know, shocking. Who could imagine that a company would bend the truth when promoting their shiny new thing? But it’s a good question. Even if the demo was faked, the tech appears to be real (in Google commercials at least, wait a minute…).
Assistant Gets Chatty
Now Google looks to be using Duplex to supercharge what their Assistant can do as the voice battles heat up.
But a feature I like even more is the brand new call screen feature on the Pixel 3 phones.
Brand Your Voice! 🎙
Unimpressed? Then check out what Microsoft is up to.
You can now create a voice font. A what? “A recognizable, one-of-a-kind voice for your brand”. This Microsoft service allows you to upload audio files of the voice style you want with matching text files of the scripts. Then away you go! The system works its technomagic and generates a voice font that will learn and evolve over time.
I can see a lot of great uses for this in this era of podcasts and voice assistants. Brands could quickly turn around podcast episodes or Alexa skill/flash briefing updates without needing to coordinate talent and go through the recording process.
I can also envision actors and voice talent creating their own voice fonts to license out for cash money. It fits in with the current Hollywood penchant for sequels, reboots, and dead actor reincarnation too.