Silence Detection and Removal Method Based on the Continuous Average Energy of Speech Signal
Abderrahmane Adjila, Maamar Ahfir, Djelloul Ziadi · 2021
The speech signal processing is a very important domain in digital signal processing. This is because a variety of noise signals could degrade the original speech signal and make it unclear to user. This paper contributes to the literature by suggesting a method to detect and remove silence from the original speech signal based on the continuous average energy of the signal. Deleting the silence and voiceless segments from the speech signal are very beneficial to growth the overall performance and accuracy of the system in many domains of applications such as speech recognition and automatic speech segmentation. The results for a database which contains English, Arabic and French speech signals shows a better performance and robustness in noisy environment. The proposed method also has a less complexity compared to the recent method based on multi-scale product and its spectral centroid. In this research work the performance is evaluated using MATLAB tool.