Research On Voiceprint Identification Technology Of Voice Carried By Laser Under Weak Channel

Voiceprint recognition,also known as speaker recognition,is a biometric identification technology that has emerged in recent years.As an important part of the artificial intelligence era,voiceprint recognition has been recognized by the world for its unique convenience,economy,and accuracy.However,channel noise interference will be introduced in the speech acquisition process and affects the accuracy of voiceprint recognition.At present,there is an emerging long-distance voice collection method,that is,using a laser as a carrier of a voice signal.After a transmission channel,a voice signal is separated from a receiver and processed in a subsequent manner.The voice signal collected in this manner is affected by noise interference of the laser transmission channel.This article studies voiceprint recognition technology for collecting voice under the most common kind of laser transmission channel—the weak turbulence channel.The voiceprint feature used in this paper is the Mel frequency cepstrum coefficient(MFCC).The voiceprint recognition system based on the artificial neural network-vector quantization(ANN-VQ)model is used as a baseline system,and then the voiceprint recognition system based on the total variability-identity vector(TV-Ivec)model is constructed as an optimization system for comparison.Through the experimental verification on the collection of speech data collected under different length of weak channel,the recognition performance of TV-Ivec system is better than the baseline system.Therefore,the TV-Ivec system is more suitable for voiceprint recognition applications of sound carried by laser in a weak channel environment.