Automatic speech recognition (ASR) is increasingly being applied in various fields such as healthcare, Education and accessibility. However, ASR of mathematical expressions presents a unique challenge due to the symbolic, structured and hierarchical nature of math language. Accurate transcription of such expressions is essential for digital learning, documentation and computational processing. This paper presents a comparative study of four open-source ASR tools— Whisper, DeepSpeech, Vosk and PocketSphinx — focusing on their ability to recognize spoken mathematical expressions accurately. The study uses a dataset of several mathematical categories, such as algebra, trigonometry, calculus, and vectors, recorded by six speakers, to assess each tool’s performance based on Word error rate (WER) and processing time. According to the results, Vosk processes information the quickest, whereas Whisper performs better in terms of accuracy. The findings provide insights into tool selection for educational or assistive technology applications involving math speech transcriptions.