This research presents, the development of Yoruba dialects classification Model for automatic speech recognition systems (ASRs) using K-Nearest Neighbor (K-NN). Research had revealed that ASRs perform better with correct dialects classification. Therefore, a non-parametric (i.e K-NN) model was developed and implemented on a Matlab 2021 platform to classify three (3) dialects (Ijebu, Ibadan and Ondo) from Ogun, Oyo and Ondo states respectively of Nigeria. The dialects were recorded at different environments, data sizes and at “opus file” format. They were later converted to “.wav” using the EZ CD Audio Converter Software. The Program4Pc Video Converter Pro was used to trim the converted audio waveforms to the same size and converted them to image signals suitable for model training, validation and testing. The results showed that the developed K-NN Classifier worked with an average performance accuracy of 91.11% and Recall {Sensitivity) of 86.67%. These results indicated that the model can be used to classify dialects of the same language hence, can help to improve the performance of robust ASR systems. However, for further improvement, better Classifiers that can handle large volumes of data should be employer.