Analysis of the Multidirectional Single Channel Speech Quality Improvement Using Dyadic Discrete Wavelet Transform in Comparison with Discrete Wavelet Transform for Improved Accuracy

Innovative dyadic discrete wavelet transform techniques are used to improve multi-directional single-channel speech accuracy. Materials and Methods: The experiment used a Kaggle dataset, which is open-access. The clinic calc tool was used to calculate with a 0.05 alpha, 0.80 G-power, and 0.2 last beta. Group 1 contains Stft and group 2 contains Dwt, each with 376 samples. Thus, 752 samples were used in this experiment. To measure accuracy, 80% of images are trained, 10% tested, and 10% validated, and performance analysis validates SPSS software's better methodology. Results: Dyadic discrete wavelet transform achieved 65% accuracy in speech quality detection and 55% in improvement, with a significance value of 0.001 (P<0.05), resulting in improved sensing performance. Conclusion: Dyadic discrete wavelet transform works better than discrete wavelet transform for speech quality improvement detection.