False tone detect suppression using multiple frame sweeping harmonic analysis

摘要

Detection of alerting or other tone signals in a composite signal are reliably detected with an improved false tone detect suppression technique which employs the use of a moving window analysis on a subset of frequencies detected within an underlying speech signal performed during each of a plurality of data frames. The composite signal is transformed during each data frame using a discrete Fourier transform comprising the so-called Goertzel Algorithm. All of the Goertzels relating to the tone detection are analyzed during each data frame, but only a subset of the Goertzels relating to the sweeping harmonic analysis of the underlying speech signal are analyzed during any particular data frame. Thus, rather than analyzing the underlying speech signal for the presence of any and all relevant frequencies during each data frame to prevent inadvertent tone detection, a piece wise sweeping harmonic analysis is performed over a plurality of data frames, providing a more efficient and less intensive moving sweeping harmonic analysis. Thus, in addition to the detection of the particular alerting or other tone signal, the underlying speech signal is analyzed over a moving plurality of data frames to provide a ‘canceling’ effect to the detection of a particular tone in the event that a harmonic frequency relating to the detected tone is present in the underlying speech signal. The result is a reliable, yet inexpensive tone detector.