Research And Implementation On The Speech Recognition Simulation System Of Underwater Robots

In China underwater robot speech recognition technology to the level of research hasbeen largely synchronized with foreign and Chinese speech recognition technology also hasits own characteristics and advantages. The development of speech recognition technology totoday, especially in small and medium-sized vocabulary, speaker-independent speechrecognition system accuracy has been through more than98%higher recognition accuracy onthe independent speech recognition system. Due to the development of LSI technology, thesecomplex speech recognition system has been completely can be made into a dedicated chip tomass production. In recent years, the speech recognition system has to open up to new areas,that is, the spoken language system.Study and compare the speech signal feature extraction, and learn the speech recognitionalgorithm, according to the practical applications of underwater robots using DTW, of whichthe main part of the detailed research for the easy problems, the improved DTW algorithm.This paper discusses a method of speech signal processing and speech recognitionalgorithms, and focuses on the DTW algorithm, and makes some improvements to thealgorithm, a simple isolated word speech recognition system simulation on a PC. Speechsignals analysis and processing, including preprocessing of speech signal acquisition,analog-digital conversion, sampling and quantization, then pre-emphasis, framing andwindow treatment, finally the start and end points of each frame signal endpoint detection todetermine the signal. Feature extraction part of speech signal, respectively, LPCC featureparameters and MFCC feature parameters and its differential coefficient described in detailand give a specific extraction method.Finally, speech recognition simulation system for underwater robots, system simulation,the DTW algorithm used and the system implementation is described, at the same timeimproved its shortcomings.