Maximum Entropy, Auto Regression, Pole-Zero Modeling-On the Use of Modern Spectral Estimation in Audio Testing
The use of the discrete Fourier Transform (DFT or FFT) for audio testing has become widespread. As an estimator of the frequency response of a time domain signal, however, the FFT is not without its limitations. The most well known limitation is the time/frequency resolution tradeoff. This paper will discuss the basis for the inherent limitations of the FFT and the effect that they have on audio testing. A new approach to the determination of the spectrum of a time signal which is free of the time/frequency resolution tradeoff, namely "parametric spectral estimation" will then be introduced. This new approach will be shown to have superior resolution for very short time records, the most difficult audio testing situations. By combining the best advantages of the FFT (non-parametric spectral estimation) and parametric spectral estimation, a significantly improved acoustic measurement can be made in very small spaces.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.