Cyber-physical systems in the framework of audio song recognition and reliability engineering

Music Information Retrieval (MIR) is the interdisciplinary discipline of extracting information from music, and it is the topic of our research. The MIR system faces a significant issue in dealing with various genres of music. Music retrieval aims at helping end-users search and finds a desired piece of music from an extensive database. In other words, Music Information retrieval tries to make music information more accessible to listeners, musicians, and data scientists. The challenges and research problems that an audio recognition system faces in everyday use might come in a variety of forms. Significant aspects are: near-identical original audio, noise, and spectral or temporal distortion invariance, a minimal length of song track required for identification, retrieval speed, and processing load are all important factors. In order to overcome these problems and achieve our goal, a Short Time Power Spectral Density (ST-PSD) fingerprinting is proposed as an innovative, efficient, highly accurate, and exact fingerprinting approach. To maintain high accuracy and specificity on hard datasets, we propose matching features based on an efficient hamming distance search on a binary type fingerprint, followed by a verification step for match hypotheses. We gradually improve this system by adding additional components like the Mel frequency bank filter and progressive probability evaluation score. Besides, we introduce a new fingerprint generation method and we present the fundamentals for generating fingerprints and we show they are robust in the song recognition process. Then, we evaluate the performance of our proposed method using a scoring measure based on the accuracy classification of thousands of Songs. Our purpose is to communicate the effectiveness of the fingerprints generated with two proposed approaches; we will show that, even without any optimized searching algorithm, the accuracy obtained in recognizing pieces of songs is very good, thus making the apprapproachropose a good candidate to be used in an effective song recognition process. I will be discussing another area of research that was done as part of my period abroad at Duke University, USA, as part of an exchange program. The topic related to reliability engineering has been incorporated. The first part focuses on the reliability and interval reliability of the Phased Mission System (PMS) with repairable components and disconnected phases, using analytical modeling as a state space-oriented method using the Continuous-time Markov chain (CTMC). The second aspect focuses on non-repairable multi-state components PMS, in which we present a practical case study of a spacecraft satellite that was used to demonstrate only the (PMS-BDD) method proposed with the implementation of Sharpe tools based on (FT) configuration in order to demonstrate the system’s reliability/unreliability in this case.