Audio signal processing plays an important role in various applications such as speech recognition, music information retrieval, and speech synthesis. Among them, Mel spectrogram is a commonly used frequency domain feature representation method, which describes the sensitivity of the human auditory system to frequency. In this article, we will conduct performance tests on three commonly used audio processing libraries - audioflux, torchaudio, librosa, and essentia - to evaluate their efficiency in computing Mel spectrograms.