As technology advances so must our means of interacting with it. Even today with Siri and Alexa we have begun to interact with our technology through speech. This method is more favorable than the use of touch by pressing buttons or moving dials. For this reason it seems logical that, eventually, merely speaking will become the primary method of interacting with our machines. In my previous post I mentioned how I was inspired by one way speech-based interfaces are used in cars. This got me thinking as to how this type of interface will be used in the future and what problems we may encounter if this technology isn't improved upon. For example, let's take a look back at it's integration into automobiles. In the near future we will probably all be using self-driving cars that will most likely rely on some form of speech interface for passengers. Let's say that, in the future, part of the information you store in the contacts on your phone is someone's address. When you enter a self-driving car it prompts you to access your contacts via Bluetooth and, in doing so, you can refer to this in your interactions with the car. If you wanted the car to take you to your friend Brian's house you would be able to say, "Take me to Brian's house", or if there are multiple "Brians" in your contacts then you could there first and last names and say, "Take me to Brian McGoober's house." Here you can see how having the ability to recognize the names of people can be convenient and useful. However, let's say that you don't want the car to access your contacts and instead the car will prompt you to give the street address. Even in this case it would still be important that the system can recognize a spoken name, but this time it would be the name of a street. Let's say you still wanted to go to your friends house but you are using the street address and you say, "Take me to 41 Bleeker Street." If the car's system couldn't recognize names or proper nouns then it could very easily misinterpret what you say. It wouldn't even know that, in this case, Bleeker Street is a name and would instead would think you said, "Take me to 41 bleaker street", which the car would not understand to be an address. Recognizing spoken names is one of the last big hurdles to overcome before speech recognition can reach the point where we can regularly use speech to interact with our technology. The sooner we perfect our speech recognition technology, the sooner we will enter the future.
Amol Kumar
Why is it important?
Updated: Sep 29, 2019