Electromyography has been used for Human-Computer interactions (HCI). Gesture recognition such as hand and finger movements is helpful to have a better HCI experience. This study investigates methods used on a publicly available dataset. To the best of our knowledge, this dataset has never been used with wavelets previously. This study uses Discrete Wavelet Transforms (DWT) with three different wavelets such as Symlet 4, Daubechies 4, and Haar wavelets. The time and frequency domain features have been extracted from the result of the DWT which uses three different wavelets. The features have been tested with a proposed Convolutional Neural Network (CNN) model. To the best of our knowledge, this CNN architecture hasn't been used before. The results with different metrics and confusion matrix for each trial are given in the results section. The highest and the lowest accuracy rates have been achieved with the Symlet 4 wavelet and Haar wavelet, respectively. The performance ranking of the reported wavelets is Symlet 4, Daubechies, and Haar with accuracy rates of 91.56%, 90.66%, and 90.02%, respectively.