<p>In today’s world, automatic gender recognition using voice is one of the important fields of speech processing. The pitch parameter is considered significant for classifying the masculine and feminine genders as women have a greater pitch value than men. The challenges arise when the pitch of females is low or the pitch of males is high. Another challenge in the gender recognition is the language-based gender recognition. So, the proposed work solves these pitch issues and results in a high-performance language-independent gender recognition system. This article proposes a generic and language-independent voice-based gender recognition system with two different approaches called a hybrid decision tree with fuzzy rules and CNN. The first approach extracts the features from the pre-processed data and designs the HeatMap to find the correlations among features. It applies four selected features to construct the decision tree of depth four and then to derive the fuzzy rules using a decision tree. These features were chosen because they play a significant role in determining the gender. The second approach configures the hidden layers without pre-processing and applies two activation functions using the hidden and output layers. It constructs the HeatMap with the samples and recognizes the person’s gender as male or female. Both models are implemented on a large dataset of voice signals. The dataset was collected from Kaggle and its size is 3168 rows × 20 columns. The system was implemented with Python and achieved 93% and 98% accuracies from the first and second approaches, respectively. The second approach outperforms the first approach in accuracy, while the first approach is more time-efficient than the second one.</p>