| Hi-index | 0.00 |
Sinusoidal models have long been used in the processing of audio signals. Many of these models are usually used to select significant frequencies in the amplitude spectrum for analysis and synthesis purposes. Therefore, when using these models, it is important to estimate the true frequency of the multiple sinusoids present in the signal. In this paper, the performance of frequency estimation using the phase vocoder is investigated. The performance of the frequency estimation under varying noise power and different window shapes and sizes is analyzed. The effect of the presence of multiple sinusoids in the signal on the frequency estimation is also discussed. In particular, the frequency bias resulting from the presence of multiple sinusoids, the selection of optimum window size and the impact of the frequency of the sinusoids on the estimates are comprehensively investigated