The DWS pitch detection algorithm finds the best harmonic relation among the peaks in the amplitude spectrum of the acoustic signal. It is argued that, for high-pitch signals, there is relatively little information available in the spectrum compared to the amount of temporal information that can be obtained from the autocorrelation function, for instance. This paper describes how the DWS algorithm was applied to the autocorrelation function and how the results from the frequency and time domains were combined to obtain a more reliable pitch estimate.
|Journal||IPO Annual Progress Report|
|Publication status||Published - 1983|