More Under the Environment of the Identity of the Noise Voice Identification Technology Research and Simulation
MA Zhong-hai · Jisuanji fangzhen · 2012
Study background based on sound identity accurately identify problems.When the individual voice confusion in the other voice background,individual unique speech characteristics due to the mixed with the characteristics,leading to their own characteristics occur degradation.By the characteristics of the traditional based on the speech recognition methods and,unable to overcome noise speech characteristics of the degradation of disturbance,cause recognition accuracy is not high.Was put forward based on the phonetic features of the mixed speech recognition probability decision-making method.Through the mix of voice independent effective feature detection,according to the characteristics of decision-making algorithm will phonetic features probability belonging to identify probability,through the calculation of ownership features complete identity recognition,and overcome speech characteristics due to the mix,causing the phonetic features of the result of degradation to identify problems.The experimental results show that this method can mix the phonetic features to effectively identify and improve the accuracy of the recognition.