Signal processing applications use sinusoidal modelling for speech synthesis,speech coding, and audio coding. Estimation of the model parameters involvesnon-linear optimisation methods, which can be very costly for real-timeapplications. We propose a low-complexity iterative method that starts frominitial frequency estimates and converges rapidly. We show that for N sinusoidsin a frame of length L, the proposed method has a complexity of O(LN), which issignificantly less than the matching pursuits method. Furthermore, the proposedmethod is shown to be more accurate than the matching pursuits andtime-frequency reassignment methods in our experiments.
展开▼