SPEECH RECOGNITION ROBOT

Main Article Content

Ashwini Prasanna, Abhishek Damodaran, Brijesh R, Harshitha TR, Bandavya M

Abstract

Speech recognition software has advanced greatly since it had been first invented, but still it's several big problems that prevent it from being employed exclusively as a way of transcription. one in every of the foremost basic speech recognition problems is thatthe quality of the input devices getting used. if a microphone isn't sensitive enough or is overly sensitive then it can create audio information that's difficult for the software to decipher. Difference in pronunciation, accents, and speaking cadence combine to make one in every of the more pervasive speech recognition problems. When one word is often pronounced in several ways, the software can become confused and misinterpret what's being said. the identical happens when an individual speaks slower or faster than the program expects. an analogous problem stems from background that may be problematic to filtrate the most speech and might cause inaccurate translations when included within the speech processing. Speech is one amongst the foremost important way of communication for people. Using speech as an interface for processes has become important. This project it's implemented to regulate a robot with speech comments. The speech recognition system is trained in such a way that it recognizes defined commands and also the design system will navigate supported the instructions through the speech commands. The received voice signal Is identified and also the voice signals are quantified, sampled and also the discretization process takes place. To the get feature extraction of the speech signal we use MEL-Frequency Cepstrum Coefficients (MFCC) method and KNN (Nearest neighbor classifier).

Article Details

Section
Articles