Automatic speech recognition has grown widely with enormous applications in the area of human communication systems. The recent developments in automatic speech recognition even have the potential to help in applications like language translations, speech rehabilitation therapies, voice assistance, etc. This has made us overcome the limitations of data processing, analysis, and evaluation as humans. There has been much research conducted in the recent past which needs an analysis for different speech-related parameters as well as finding the accuracy for native (English) and non-native speakers. Achieving higher accuracy with such systems still has many challenges like the environmental noise, the domain-specific communication, native (English) and non-native speakers, pace of speech, accent and dialect, speech structure, etc. This work is an effort to learn the difference in the performance of two different models when applied for speech recognition of native (English) and non-native speakers. The study extends further to analyze the differences in political and technology-related domain-specific speech recognition. The experimental results show the difference in the performance of the models. The accuracy varied for both models.