![]() | Only 14 pages are availabe for public view |
Abstract Speech recognition is a knowledge based process for automatically extracting various information from speech ultterd by human beings. Speech recognition technology has made steady process in its 40-years history and has succeeded in creating several substantial applications. The goal of speech recognition research is to produce a machine which will recognize accurately normal human speech from any speaker. One of the most important achievements for the feature extraction part and speech recognition techniques is the linear predictive coding (LPC). In this work LPC is used for extracting the formants of the Arabic digits from (0 - 9) and DTW techniques is used for time normalization. The artificial neural networks are used for speech recognition as a classifier for two main systems which are dependent system and independent system for speakers. The recognition accuracy for dependent system is 98.8% but the recognition accuracy for independent system for males only is 88% and the recognition accuracy for females only is 87.8%. When combined groups are used, the recognition accuracy is 78.8% so a new system is used for improving this result after dividing the Arabic digits into groups and prepared as inputs to the system where three neural networks are used to classify the Arabic digits. The system improve the recognition accuracy into 95.5%. |