Affiliation:
1. School of Journalism and Communication, Hunan University, Changsha, Hunan 410000, China
Abstract
When using DSP technology, technicians can easily and conveniently replace DSP audio processors, make second-party equipment, improve processor performance, reduce application costs, and receive and make music to meet the needs of different signals. Introduces knowledge of speech comprehension, which includes theoretical and cognitive processes, such as pre-speech characterization, final discovery, behavior, structure, and knowledge. By analyzing and comparing various common features, Mel frequency cepstrum coefficient is used to determine the physical features, and the results of traditional algorithms are studied, compared, and evaluated. Research, compare, and evaluate numbers in traditional algorithms. The simulation is based on the algorithms of all the connections used in the system and checks that the algorithm is correct. Design an experimental model, compare the advantages and disadvantages of algorithms and algorithms, consider the process requirements and changes of the whole algorithm, and select the algorithm accordingly algorithm. The experimental results show that the speed recognition of the TW algorithm is not much different from that of the HMM algorithm, and the efficiency of the DTW algorithm in the training models is lower than that of the HMM algorithm. Given the limited resources of the DSP platform in this system, we have chosen the DTW algorithm as the self-reporting system for this system. Finally, the performance of the system is checked and the test results are released. Observations of the experimental instrument showed that the system was effective in being able to recognize small words and discrepancies in speech.
Funder
Social Sciences Youth Fund of Hunan Province
Subject
Computer Networks and Communications,Information Systems
Cited by
2 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献