Soft Computing Implementation of Automatic Speech Recognition: Isolated Words Recognition

This work assimilates an introduction to the Human speech production system, Fundamentals of speech recognition, the methods for data analysis as well as the theoretical background of past research, Soft computing, implementation as a form of Methodology, Results and Conclusion & future work . Soft Computing includes a family consisting of many members, namely Genetic Algorithms (GAs), Neural Networks and others. With this text, you will find a balance between theory and implementation that allows you to build your understanding of the basic concepts, Methodology and a results obtained using various neural networks with Recognition percentages of isolated words in two different languages and Confusion Matrix for radial basis similarly for probabilistic. This work also shows the Comparative analysis of results from the Recognition using GA and without GA. The use of GAs for feature optimizers is a novel concept in Automatic Speech Recognition hence the system becomes more robust to noise. Emphasis is laid more on a soft computing rather than on a hard one.