Design And Implementation Of Speech Recognition In Multimedia Player

In the next ten years, speech recognition technology will be applied to the various fieldsof industry, home appliances, communications, automotive electronics, and consumerelectronics products. With the development of technology and people’s living standardsimprove, the multimedia player has got a great promotion on the market, applies speechrecognition technology to the multimedia player will make a great significance. The purposeof the paper is to design a multimedia player with voice recognition function, research andanalysis the speech recognition algorithm and suggests improvements. The algorithm sourcecode can be easily ported to other embedded devices, and it will make a great significance forthe research, improvement and application of follow-up speech recognition technology.Paper designed hardware and software of multimedia player. Master chip usingAllWinner’s F15, the hardware power management module uses AXP188which is a highlyintegrated power systems management chip. Designed and implemented the multimediaplayer GUI system, and developed music, movies and other multimedia applications.Design and implement the isolated word speech recognition system of non-specificcomputer instruction in a multimedia player. The papers studied and analysis variousalgorithms theory used in the speech recognition and with the comparative analysis selectedan algorithm suitable for embedded multimedia player application. Selected dual-thresholdmethod, which is based on short-time zero crossing rate and short-term energy, as speechendpoint detection algorithm; selected the MFCC as the characteristic parameters of the voice,in the feature extraction, used the efficient based-4FFT for spectral analysis; Used DTW toidentify results in the stage of voice template library matches. This paper improved thedual-threshold speech endpoint detection algorithm to prevent the phenomenon of instructionrecognition truncated. In order to reduce the false detection rate of the endpoint detectionaffects to the final recognition results, did some improvements to the DTW algorithm.This paper developed a speech recognition application, which is convenient forinteraction between user and multimedia-player. With system testing, the multimedia player isuser-friendly and has beautiful UI, can play1080p movies, music, pictures and othermultimedia files, has high rate of speech recognition and real-time features, and can do movieplay, music play, enter, exit and other operations through voice control, it verifies thefeasibility of the algorithm.