查看文章

This paper describes several feature extraction techniques, which will facilitate Automatic Speech Recognition (ASR) for Bangla speech. These techniques are applied on different sound-packets, which are essentially segments of Bangla speech. The key temporal regions in a sound-packet that contain vital information about the speech signal are identified. Some novel feature extraction methods are developed using the information contained within these key regions. It has been observed that a single feature cannot provide enough information to achieve successful automatic speech recognition; rather a combination of the features can be used effectively to increase the accuracy.