Study on modulation spectrum feature extraction of ship radiated noise based on auditory model

Under the noise interference environment, the auditory system of the human ear boasts strong robustness and reliability. Integrate the auditory model of the human ear into the ship radiated noise modulation spectrum extraction process can further improve the signal to noise ratio (SNR) and enhance the effect of the extracted modulation spectrum. This paper simulates the processing mechanism of the cochlear and the practical characteristics of ship radiated noise signals according to the integrated Gammatone auditory filter and the Meddis internal hair cell model, and it builds a Gamatone-Meddis auditory periphery model that is suitable for ship radiated noise signals. Then, the ship radiated noise signals are processed by the Gammatone-Meddis auditory model, the low-pass filtering (LPF), the Fast Fourier transform (FFT) and the multi-band integration in order to obtain an improved auditory spectrum extraction technique. The simulation results have verified that the Gammatone-Meddis model can extract modulation information well. After applying the method to a test of practical ship radiated noise signals, the author finds that the extraction method presented in this paper can obtain an obvious modulation spectrum at low SNR, and has a great advantage over the traditional modulation spectrum extraction methods.