The Method And Technology Of Teaching System Based On Face Recognition And Speech Recognition

Nowadays,higher requirements are required by the safety and intelligence of laboratory teaching system.Because the existing laboratory system is so traditional in the way of login,that the existing laboratory system does not make full use of the biological features such as human faces,and lacks the interactive command function of speech recognition,which needs to be improved in the intelligence of teaching demonstration.Therefore,this paper studies the laboratory teaching system based on face recognition and speech recognition,focusing on the improvement of face recognition and speech recognition interaction.Compared with the traditional laboratory teaching system,the proposed laboratory teaching system based on face recognition and speech recognition has more comprehensive functions,more reasonable design,more intelligence and higher efficiency,which can fully improve the login security and flexibility of the laboratory teaching system.First of all,according to the actual needs of the laboratory teaching system,the resea rch background of the face recognition and speech recognition and the research significa nce of the laboratory teaching system are explored.The research status of the face recogn ition and speech recognition at home and abroad is analyzed,and the face recognition an d speech are analyzed.Identify and analyze the existing problems,then formulate related research contents of the laboratory teaching system based on face recognition and speech recognition,and determine the relevant technologies required by the system.Secondly,a face recognition model that fuses principal component analysis technolo gy is designed.The traditional face recognition steps are analyzed,then the lack of existi ng face recognition methods are summed up,based on Haar features,Adaboost is used t o complete face detection,based on Landmarks method,construct the regression tree afte r the end of the face alignment,fusion principal component analysis,the face images are re-constructed.The re-constructed images are to reduce the dimension of the face image,and the final face match is to achieve the face recognition task.Thirdly,the speech recognition model is designed in the laboratory teaching system.Voice signal preprocessing such as digitizing,pre-emphasis,windowing,frame segmentation,and endpoint detection is performed on the voice signal input by the microphone,an d the corresponding speech recognition feature is extracted based on the Mel frequency c ep-strum coefficient for the preprocessed speech signal.Based on the HMM-PNN hybrid model,acoustic model is constructed in speech recognition.The speech model is constru cted based on N-element statistical theory.The decoder in the speech recognition technol ogy is established based on the Viterbi algorithm to complete the entire speech recognitio n task.Finally,the proposed face recognition and speech recognition methods are verified.Through the data in GT database,the corresponding face recognition experiments are ver ified,the accuracy and time consumption of face recognition are estimated,and the exper imental situation of the proposed face recognition method is analyzed through compariso n with the existing face recognition methods.In terms of speech recognition,the recognit ion rate of the proposed speech recognition system model and the existing speech recogni tion system model are compared,and the anti-noise performance of the speech recognitio n system model is analyzed.