Bench Talk

Hands-free computing is a growing technology for personal computers. Google Assistant, Siri, Cortana, and Alexa act as digital personal assistants, allowing users to control their devices through voice recognition. While previously limited to smartphones, these digital personal assistants are rapidly moving onto desktop operating systems. With the move, the assistants will have to conquer larger tasks, like running complicated software, and will have to be able to understand a voice from far away in potentially noisy environments.

When you think of digital assistants, you might imagine asking your phone to schedule an appointment or to map a route to somewhere you need to go. A lot of us are still just now experimenting with Cortana on our PCs. But considering that embedded systems are now in just about everything, and getting more powerful all the time, it makes sense that shortly we could be talking to computers more than we will type or touch them. Being able to communicate through voice with an intelligent system is vital for smart homes and autonomous cars to become a part of everyday life. I, for one, can’t wait to get into an argument with my fridge about my unhealthy dietary habits. But what kind of technological advancements do we need to develop before the world sees these hands-free systems become a reality?

Figure 1 Why touch your fridge when you can yell at it? Source: LuxuryLaunches.com

First, our computers have to be able to hear and understand us. Having Siri misunderstand your request for travel directions is funny, but when your driverless car takes you to the wrong destination and makes you late for an important event, it will probably be quite frustrating. Any far field system has to have extremely sensitive microphones to pick up voices from across a room as well as from close by. With such sensitive microphones, all sorts of extraneous noise are going to be picked up by the system and possibly processed as commands. To combat this, new filtering algorithms are being developed by companies like Sound Research to isolate voices from background sounds. At Intel Developer Forum 2016, Intel representatives showed off software which processed a recording of someone speaking to their PC while a TV blared in the background. After being filtered by the software, the recording no longer had any remnants of the TV noise, leaving the voice to be clearly understood by the digital assistant.

Figure 2 With better microphones and noise filtering, these people wouldn't be so angry.

Once detected, the computer then has to understand the message and interpret its meaning.  Speech recognition is an intensely interesting subject, as is the wider field of machine learning which is necessary for the sorts of systems I imagine would be assisted by giving vocal commands to a computer. Put all of this computer science and data analysis together, and we get technology which allows you to tell your Roomba to vacuum the floor. Unlike your kids, it’ll actually do it.

While humans will continue to use keyboards and mice in their day-to-day operation of most computing devices, the voice-controlled features that currently exist are a peek into the future of computing. Technophobia might see a dramatic downturn once computers are prepared to do your bidding with some simple voice commands. You may not care about talking to your work computer, but you can already tell your laptop to set a wake-up alarm while you’re brushing your teeth across the room. In the not-so-distant future, I would like to tell my smart coffee maker and smart chef to make breakfast for me from across the house without even getting out of bed. And those computers had better get my order right the first time, even with my alarm blaring over my voice.

While humans will continue to use keyboards and mice in their day-to-day operation of most computing devices, the voice-controlled features that currently exist are a peek into the future of computing. Technophobia might see a dramatic downturn once computers are prepared to do your bidding with some simple voice commands. You may not care about talking to your work computer, but you can already tell your laptop to set a wake-up alarm while you’re brushing your teeth across the room. In the not-so-distant future, I would like to tell my smart coffee maker and smart chef to make breakfast for me from across the house without even getting out of bed. And those computers had better get my order right the first time, even with my alarm blaring over my voice.