Research on the interactive design of electric vehicle interior based on voice sensing and visual imagery

Abstract

With the complete function of modern automobiles, in-vehicle intelligent devices are becoming more and more complex and the requirements for human-computer interaction are also increasing. The research proposes a speech recognition method that combines multi-window estimation spectral subtraction and dynamic time warping to enhance the denoising ability and speech recognition ability of in-vehicle devices. It also proposes actions based on a Gaussian hybrid segmentation algorithm and a visual image functional space segmentation algorithm. The automatic identification method and the validity of the algorithm are verified. The results show that under different input signal-to-noise ratios, the denoising capability of the method is improved by 2.45% to 31.47% over the baseline method. And the accuracy of speech recognition in the vehicle environment is 92.3% to 98.7%. It is hoped that this research can make some contributions to the upgrading of voice and visual interaction within electric vehicles.