Pencereleme tekniklerinin konuşma tanıma sistemi için karşılaştırılması

Abstract

Speech recognition, which provides the ability to manage intelligent devices, aims to improve the prediction skill of these devices by extracting the necessary information from the speech signal. In this study, with examination of past researches, a speech recognition system with high quality for efficient speech recognition is suggested, and performance analysis was performed with different experimental sets. In addition, prior to the feature extraction, different windowing techniques were compared on the sound signals in the pre-processing stage to evaluate the situation. For each windowing technique, separate experiments were carried out under the same conditions and the results were discussed, then it was stated which technique was efficient.