Research On Pressure Buried Person Calling For Help Speech Signal Processing Algorithm

In recent years, the earthquakes happened in China more frequently than before. The current technology is still unable to accurately predict the occurrence of earthquakes, so the seismic damage is inevitable. Now that the earthquakes can’t be predicted, the death caused by the earthquakes hold be reduced as much as possible. Life detection equipment plays a great role in the previous earthquake relief process, in the sense of saving the buried people as early as possible.However, due to the speech signal of buried people detected by life detection equipment contains a lot of background noise, extraction and recognition of life signal will be affected. This paper focuses on short-time spectral estimation speech enhancement technology and speech enhancement technology integration. Based on the variable adaptive EASI, we propose an EASI algorithm speech signal processing method for short-time spectral estimation, which can separate hue and voice signals form the background noise effectively, so as to achieve a reinforcing effect on the speech signal of the call for help.This article mainly focuses on the research of signal processing method in the scenario of seismic rescuing, i.e. the separation of speech signal and some kinds of noise, so as to enhance speech signals calling for help. We simulate the separation of a call voice signal and background noise signal, and accomplish the design process of the whole algorithm. The main research contents are as follows:Firstly, the characteristics of voice and noise are analyzed:apart from analyzing emotional language features of the voice signal for help (useful signal), we also analyzed the various characteristics of the main sources of noise, such as large-scale rescue machine, on-site staff voices, recorded tank noise, restaurants noisy and staff noisy. Secondly, the research is about the enhancement of the voice of calling help; in this article, we simulate all kinds of noise in the earthquake disaster rescue areas, through simulating the mixture of voice signal and the background noise. It achieves enhanced effort to de-noise the mixed-signal by relative voice enhancement’s method. We also analyze various enhanced speech signal processing algorithm to evaluate the enhancement algorithm.Thirdly, call for help voice signal and background noise separation method is studied:Thought the use of blind source separation techniques, Equation variables adaptive ICA separation algorithm the hue and voice signals to be separated, its separation effect proved by experiments, and its separation algorithm evaluated.Fourthly, we propose a new hue and voice signal processing methods:the use of basing short-time spectrum estimated speech enhancement algorithm and the improved algorithm of EASI separation is effective combining to design a new speech enhancement algorithm, and this algorithm can make both voice of background noise and the vocal non background noise removed.This paper focuses on the implementation of key technologies based on short-time spectral estimation and EASI algorithm speech signal processing method, and its key technologies:one is voice enhancement technology in voice amplitude spectrum estimation based on short-term spectral estimation. The other is iteration of separating matrix and step-size adaptive principle in the ICA algorithm based on equation variables self-adaptive speech signal separation.In this paper, we propose a new speech signal processing method by researching key technologies of the speech signal processing method based on short-time spectral estimation speech enhancement algorithm and equation variables self-adaptive speech enhancement algorithm, and realize simulation.The chapters are organized as follows:First chapter research background, significance and both here and abroad research status. Relevant literature of both here and abroad were reviewed, the goal of this study, the content, methods, and expected research results. Second chapter analyzes the voice characteristics and background noise characteristics, analysis of useful signal and harmful interference noise signal of the rescue scene. Third chapter mainly research based on short-time spectral estimation enhancement algorithm and wavelet transform the hue and voice signal processing, to apply it in speech signal processing, and analog call for help voice signal will be simulated. Forth chapter mainly research separation of background noise in earthquake rescue scene call for help speech signal processing, through EASI algorithm to study the speech signal and the background noise separation, and through experiments to verify the effect of the separation. Fifth chapter proposing a new based on short-time spectral estimation and EASI algorithm speech signal processing method, and through the analog voice signal for help with background noise mixed-signal simulation experiments, and obtained good separation effect. In summary, showing the results of this research, and this paper summarizes the work and prospects for future research work.In conclusion, this paper combines short-time spectral estimation speech enhancement technology and equal variables self-adaptive speech signal separation algorithm to propose a new calling for help speech signal processing method. And through the method deals with the simulated speech signal contained noisy for help, successfully implement separation of the analog burying personnel voice signal for help and the background noise.