Comparing the frequency spectrum distributions calculated from several successive frames, the change of the frequency spectrum of speech frames between successive frames is larger than that of the ship-radiated noise. The aim of this work is to propose a novel speech detection algorithm in strong ship-radiated noise. As inaccurate sentence boundaries are a major cause in automatic speech recognition in strong noise background. Hence, based on that characteristic, a new feature repeating pattern of frequency spectrum trend (RPFST) was calculated based on spectrum entropy. Firstly, the speech is detected roughly with the precision of 1 s by calculating the feature RPFST. Then, the detection precision is up to 20 ms, the length of frames, by method of frame shifting. Finally, benchmarked on a large measured data set, the detection accuracy (92 %) is achieved. The experimental results show the feasibility of the algorithm to all kinds of speech and ship-radiated noise.

In this paper, the discrimination of speech and ship-radiated noise has been studied, for their interference with each other in the modern sea war, and a new feature abbreviated as SFSD (the Similarity of the Frequency Spectrum Distribution of two successive signal frames) is introduced. Based on the feature, a new two-stage approach is proposed, which firstly computes the SFSF and classifies the signal into speech or ship-radiated noise roughly, and then in the second stage, with the help of context smoothing, the signal are final discriminated in to speech or ship-radiated noise. The algorithm is benchmarked on a large measured data set, with correct recognition accuracy of 96% for ship-radiated noise and that of 92% for speech. Experimental results show that efficiency is exceptionally good.