Infinite Superimposed Discrete All-Pole Modeling for Multipitch Analysis of Wavelet Spectrograms
نویسندگان
چکیده
This paper presents a statistical multipich analyzer based on a source-filter model that decomposes a target music audio signal in terms of three major kinds of sound quantities: pitch (fundamental frequency: F0), timbre (spectral envelope), and intensity (amplitude). If the spectral envelope of an isolated sound is represented by an all-pole filter, linear predictive coding (LPC) can be used for filter estimation in the linear-frequency domain. The main problem of LPC is that although only the amplitudes of harmonic partials are reliable samples drawn from the spectral envelope, the whole spectrum is used for filter estimation. To solve this problem, we propose an infinite superimposed discrete all-pole (iSDAP) model that, given a music signal, can estimate an appropriate number of superimposed harmonic structures whose harmonic partials are drawn from a limited number of spectral envelopes. Our nonparametric Bayesian source-filter model is formulated in the logfrequency domain that better suits the frequency characteristics of human audition. Experimental results showed that the proposed model outperformed the counterpart model formulated in the linear frequency domain.
منابع مشابه
A Study of Mexican Free-Tailed Bat Chirp Syllables: Bayesian Functional Mixed Modeling of Nonstationary Time Series Data with Time-Dependent Spectra
We describe a new approach to analyze chirp syllables of free-tailed bats from two regions of Texas in which they are predominant: Austin and College Station. Our goal is to characterize any systematic regional differences in the mating chirps and assess whether individual bats have signature chirps. The data are analyzed by modeling spectrograms of the chirps as responses in a Bayesian functio...
متن کاملApplication of Radial Basis Neural Networks in Fault Diagnosis of Synchronous Generator
This paper presents the application of radial basis neural networks to the development of a novel method for the condition monitoring and fault diagnosis of synchronous generators. In the proposed scheme, flux linkage analysis is used to reach a decision. Probabilistic neural network (PNN) and discrete wavelet transform (DWT) are used in design of fault diagnosis system. PNN as main part of thi...
متن کاملA Performance Monitoring Approach to Fusing Enhanced Spectrogram Channels in Robust Speech Recognition
An implementation of a performance monitoring approach to feature channel integration in robust automatic speech recognition is presented. Motivated by psychophysical evidence in human speech perception, the approach combines multiple feature channels using a closed loop criterion relating to the overall performance of the system. The multiple feature channels correspond to an ensemble of recon...
متن کاملA Study of Mexican Free-Tailed Bat Chirp Syllables: Bayesian Functional Mixed Models for Nonstationary Acoustic Time Series.
We describe a new approach to analyze chirp syllables of free-tailed bats from two regions of Texas in which they are predominant: Austin and College Station. Our goal is to characterize any systematic regional differences in the mating chirps and assess whether individual bats have signature chirps. The data are analyzed by modeling spectrograms of the chirps as responses in a Bayesian functio...
متن کاملPole Assignment Of Linear Discrete-Time Periodic Systems In Specified Discs Through State Feedback
The problem of pole assignment, also known as an eigenvalue assignment, in linear discrete-time periodic systems in discs was solved by a novel method which employs elementary similarity operations. The former methods tried to assign the points inside the unit circle while preserving the stability of the discrete time periodic system. Nevertheless, now we can obtain the location of eigenvalues ...
متن کامل