Research On Voiceprint And Semantic Recognition Algorithm Of Intelligent Illumination Sound Control Signal

Intelligent lighting voice control has the advantages of high efficiency,energy saving,diversified control,low cost,etc,and has been rapidly developed in intelligent lighting systems.Aiming at the intermittent,non-uniform and non-stationary characteristics of the voice signal of intelligent lighting system,this paper adopts the theoretical methods of voiceprint and speech recognition to study the recognition algorithm of intelligent lighting voice signal,which provides a safe and comfortable technical solution for intelligent lighting and smart home applications.The main work of the thesis is as follows:1.Aiming at the intermittent and non-linearity of the intelligent lighting voice signal,the minimum embedded frame length of the speech signal time series is determined by the saturation embedding dimension calculation,andthe endpoint detection is performed on the framed speech signal by the short-time energy and the short-term average zero-crossing rate.Therefore,the goal of accurately and efficiently identifying the voice signal is achieved.2.Aiming at the identification function of the intelligent lighting voice signal,through the improvement of the voiceprint recognition algorithm of the GMM-UBM model,the speaker's identity determination of the voice signal is realized,and the voice lock password is added to the intelligent lighting system,which enhances the security of the intelligent lighting system.3.Aiming at the semantic recognition function of intelligent lighting voice signal,the voice recognition algorithm based on DTW method is used to search and match the strategy by vector cumulative distance method.The identification and understanding of the intelligent lighting voice signal is achieved through the best match with the voice template library.4.An adaptive real-time speech training sample library with multi-person,multiscene and multi-command stipulations is established to realize intelligent voice control of multiple semantics of multi-zone luminaires by defined identity persons,so that the system has wider applicability.In this paper,Matlab2017 A is used to simulate the intelligent lighting voice signal,and the influence of different feature extraction methods,different voiceprint recognition methods and different semantic recognition methods on system recognition performance under different parameters is compared and analyzed.The experi mental results show that the correct rate of voiceprint recognition can reach 100% in speaker recognition,the error acceptance rate in speaker confirmation is less than 2%,the semantic recognition correct rate is up to 99%,and the recognition speed is fast,and the experiment achieved good results.