Voice Recognizing games is used to play games without the help of any physical devices such as keyboards, mouse, game pads, etc. Games can be controlled using human voice, which is easier, portable and efficient than any other physical device. The goal of the project is to develop a system which recognizes voice of any user with any accent by training the system so as to increase its confidence level for the user. Voice commands are recognized with a high confidence level and the recognized voice command is linked to a particular word which is then written into a file. By using X11 programming, an event driven programming, the recognized words which are written into the file acts as events to trigger a task, such as a keystroke. Thus by using both, a trainable speech recognizer and X11 programming, we are able to use voice instead of a keyboard and thus play games using voice.
Automatic speech recognition (ASR) has been used in variety of real time applications. Successful implementation of ASR system can replace pads and keyboards as computer games controlling devices. A successful computer game should be intuitive, interesting and player friendly. For a game to be player friendly, user interface should be easy to use like ASR system. At first, the system must recognize the command in a short interval around one second. Secondly, the accuracy of speech recognition should be very high. For this project, we are looking at 90% sentence recognition accuracy. An ideal game using ASR system as user interface, should be able to recognize sound of every speaker in common acoustic conditions, including various accents, non native speakers and even for people with speech deficiencies.
Factors that can impact robustness of speech recognition are
- Low quality microphone
- Noisy environments
- High processing power and memory requirements of an ASR system.
There are a lot of research done to improve robustness of speech recognition against noise. Speech enhancement algorithms are one of the most commonly used techniques. In this, before passing to an ASR system, speech signal undergoes a denoising process like spectral subtraction. Another way to reduce noise is to design new auditory models which are less sensitive to noise. To show case voice recognition capabilities, a game called Rally Navigator is created in which the player will compete in the race as a navigator. Objective of the Player in the game is to provide the driver with information about the route and track elements like curves and straights. In order to make the game more complex, speed control and gear switching is done based on player inputs. The aim of the game is to win the rally. Precise and well timed information supplied by the player, helps the car to finish the game. Above functionalities will be controlled by ASR System.