With the speed at which modern technology is developing, efforts are being made to make the system easier to use by utilizing voice characteristics like speaker and gender identification. Speech recognition and automated interaction-based sound-reacting systems depend on gender as a core component. By identifying the voice’s gender, these systems’ computing demands are reduced for further processing. Using a learning technique makes it possible to determine the gender from voices. There are two ways that machine learning advances language: text processing and speech-based processing. Given that most KUI speakers are found in Odisha, where Odia is the official language, KUI is a tribal language with few resources. Most KUI-speaking Kondhs reside in South and Central Odisha's hilly and forested regions. Language processing over KUI thus becomes necessary for modernization. In this study, 2000 male and female KUI voice samples were taken as a dataset. Here, six different learning techniques: Random Forest (RF), Logistic Regression (LR), Naïve Bayes (NB), Convolutional Neural Network (CNN), Artificial Neural Network (ANN), and Deep Neural Network (DNN) are used for evaluating the accuracy. Additionally, the models are compared using performance measures such as F1-score, Recall, Precision, and AUC-Score.