Google and many other IT companies are doing their best to make the devices we know today more human. Various programs have been installed on smartphones and tablets to enable them to bear a normal conversation with their owners, such as, Apple’s Siri, but many more changes will be enacted in the future.
Google engineers have studied the human neural network algorithms at length before they created their Android OS. The program is, thus, capable of recognizing the human voice and correctly interpret commands due to a computerized learning system that resembles the human brain. Tests have shown that the new program has significantly reduced the voice error rate as opposed to the previous versions of software.
The modifications that have been operated are rooted in the manner in which the two data transmitted to the device is stored and interpreted by the software. Unlike the previous programs, Google’s Android is using a voice recognition program that chops and stores the information to eight different computers belonging to Google’s wide array of servers. After the information is stored, the program processes the data using the human-inspired model created by Vanhoucke and his team.
He explained that it is not just the layers that are formed as we speak that help us create a good perceptual system, but rather the velocity with which we interpret the data contained in the layers. According to developers, Google’s software begins the interpretation of the information at the lowest level possible by analyzing first the vowels and consonants that make up the individual parts of speech. Once the first layer has been deciphered, the software passes on to another layer until it can completely understand what the user has stated.
The voice recognition system based on neural networks was first started in 1980, but real progress was registered only in 2006 with the discovery of deep neural mappings and the introduction of low-cost graphical processing units. The model inspired by neural network algorithms has been used in the creation of many products, but the Jelly Bean speech recognition software is the most advanced.