Time-frequency algorithm for octave analysis of nonstationary interior vehicle noises

Based on the discrete wavelet transform (DWT), a new time-frequency algorithm for octave analysis (OA) of nonstationary sounds, so-called DWT-OA method, is proposed in this paper. A DWT-based multirate filter bank combined by a set of low-pass and high-pass filters was developed for octave analysis of some measured nonstationary interior vehicle noises. To establish the multirate filter bank, the Mallat algorithm was applied, and some wavelet functions in common use, such as the Daubechies, Coiflets, Symlets, etc., were studied and discussed. The Daubechies wavelet with the filter length of seventy was determined and applied to the vehicle noises. Verifications suggest that the newly presented DWT-OA algorithm is accurate and feasible for octave analysis of the nonstationary vehicle noises. It may be extended to dealing with other types of sound signals in engineering.