Research On Key Technologies Based On Vehicular Speech Recognition

With the development of artificial intelligence,speech recognition technology has played an irreplaceable role in people’s lives.This thesis deeply studies the principle components of speech recognition system,including preprocessing,endpoint detection technology,extraction of feature parameters and identification methods.Two improved algorithms are proposed for endpoint detection technology in the case of low SNR,the specific research results include the following aspects:Improved endpoint detection method is based on EMD-Teager energy and sub-band spectral entropy.The algorithm first reconstructs the speech signal by using the middle and high frequency components of the IMF3-IMF5,which is decomposed by EMD.Then,the average teager energy is calculated and combined with the traditional sub-band spectral entropy method.Finally,a double-threshold endpoint detection algorithm is used to perform endpoint detection on noisy speech.An improved speech endpoint detection algorithm based on sub-band variance combined with bark wavelet packet and spectral subtraction.The algorithm is mainly based on the endpoint detection of the original variance method.First,the spectral subtraction is used to denoise the speech,and the bark wavelet packet is used to decompose the speech.Second,the variance of each sub-band and the average of the17 sub-band variances are calculated.Finally,a double-threshold endpoint detection algorithm is selected to perform endpoint detection.In the case of leopard noise,two improved speech endpoint detection algorithms are compared and analyzed.Experiments show that the improved speech endpoint detection algorithm based on sub-band variance combined with bark wavelet packet and spectral subtraction has better detection performance in low SNR leopard noise environment.A recognition system based on Hidden Markov Model(HMM)is established on the MATLAB software platform to identify tank voice commands.Finally,a human-computer interaction GUI interface is designed and implemented by the application requirements,and the performance of the system is improved.