Abstract
With a growing interest in advancing artificial intelligence, quantum computing has become a focus to overcome challenges presented by classical machine learning methods. The fields of sound mechanics and quantum mechanics share a fundamental connection as they both delve into the intriguing realm of wave phenomena. This interrelation holds immense potential for transforming the way we approach the analysis of audio signals. Audio signals are inherently complex as they are a summation of multiple sinusoidal components. Traditional computational methods have limitations in their ability to extract subtle features from audio signals, which hinders progress in areas such as speech recognition, music processing, audio compression and acoustic monitoring. Theoreticians have suggested that quantum algorithms may overcome classical limitations by modelling signals as quantum superpositions and entanglement. The quantum Fourier transform is hypothesized to have exponential advantages over classical methods for frequency decomposition, and when integrated into quantum neural networks, it could enable speed up and efficiency in audio processing and classification. To investigate this idea, a hybrid "Quantum Fourier Neural Network" model was developed and tested on benchmark datasets using circuit simulation techniques. This is designed on two separate audio datasets ESC-50 and GTZAN. The input data is pre-processed and transformed into quantum states. A quantum circuit is created whereby the QFT in applied. The frequency domain output from the QFT is used the input in the QNN model and evaluated. Preliminary results showed computational inefficiencies compared to classical methods for acoustic classification tasks. Although there are theoretical advantages to combining auditory science and quantum processing, where an accuracy of 70% is achieved by the QFNN. The study concluded that there is potential for the model to be advantages based on the results although the current quantum resources and simulated environments are inadequate to surpass traditional methods for audio signals complexity. Advancements in quantum hardware and algorithms will be crucial to realizing this method's potential in the long term. Ongoing research at the intersection of these fields may ultimately transform our understanding of signal processing.