Sinusoidal modeling is based on the decomposition of audio signals into a sum of sinusoidal components plus a noise residual part. It involves accurate sinusoid parameters estimation and, in particular, accurate frequency estimation. A broad category of methods uses the Fast Fourier Transform (FFT) as a starting point to compute frequency. All these methods present very similar forms of estimators, but the relations between them are not yet fully understood. This work proposes to take a deeper look into these relations. The first goal of this work is to present a clear review and description fo the classical FFT-based frequency estimator. A new estimator similar to the phase vocoder is presented. The second goal of this work is to identify the common hypotheses and the common steps of the process for this category of estimator. Lastly, experimental comparisons are given.
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 free for AES members and E-Library subscribers.