Comparison and analysis of multimedia audio files with time-frequency method

Audio signal searching algorithms are often used to detect the required file in large databases, especially in various multimedia applications and smart devices. They generally rely on different audio fingerprinting techniques and have become very popular especially in the last two decades with the advancing technology in the digital world. In addition, similar algorithms inspired by this technique have been applied in different technological fields and have produced very successful results. The primary goal of these algorithms, which are developed based on different audio fingerprinting methods, is to perform the given task in the shortest possible time in a reliable and robust manner. In this study, using the traditional audio signal search methodology followed in the literature, an alternative audio fingerprint-based audio search program to the existing search algorithms has been created based on the spectral peaks method in order to determine the correct match in the database. After demonstrating the success of the program in correct matching, the effect of the window function used in the program on the performance of the program and the effects of some parameters in the spectrogram structure, which has an important role throughout the program, on the reliability and robustness of the program were tested. Thus, it is aimed to contribute to new studies based on the spectral peaks method by investigating the effect of the mentioned factors in the recognition of audio files analyzed using a similar methodology. As a result of the studies carried out, it has been observed that the audio file that was queried was detected in the database with a satisfactory reliability and robustness under all conditions.