Font Size: a A A

Research On Modern Speech Recognition Technology And Application In The Telecommunications Customized Ringing Tone Service

Posted on:2015-09-27Degree:MasterType:Thesis
Country:ChinaCandidate:Y LuoFull Text:PDF
GTID:2308330482953084Subject:Electronics and Communications Engineering
Abstract/Summary:PDF Full Text Request
Automatic researched and Recognition (ASR) is a kind of technology of converting speech signals into text information which could be recognized by computer programs, so as to identify the intention of speakers by machine automaticly. The final goal of ASR is to realize that the machine can interact with people by natural language. In the 1950s, American Telephone and Telegraph Company (AT&T) built the first ASR system-Audry system, it can only identify ten English Numbers. In the late 1980s, the development of the digital information technology, computer software and hardware technology, recognition brings a new technical foundation for the research and development of ASR. Speech recognition technology was first used in the field of business, and quickly spread to every fields of social production and life, itself has also been constantly updating of technical requirements. ASR technology were first used in the field of business, and quickly spread to every fields of social life.Firstly, there is a brief introduction of development status of voice recognition technology in the first chapter. Secondly, from the basic model aspects of speech recognition technology, introduces the endpoint detection, pretreatment, module generation, matching module and module management principles and other aspects of pattern recognition and digital signal processing. Through the development process of analysis and research of speech recognition technology, respectively, through HMM (Hidden Markov Model) and DTW (dynamic time warping Fa) Design Chinese language voice recognition system. HMM has a strong time-series modeling capabilities, the characteristic parameters of time for training, each corresponding to its own voice hidden Markov model, voice and hidden Markov models corresponding match, complete the identification process. HMM is usually applied in large vocabulary needed to identify the case. Time DTW able to handle the length of the speech signal characteristic parameters of different issues, with a recognition speed, low cost and effective system of small vocabulary treatment effect. By the end of each section are Matlab simulation, and provide a script. Finally, a detailed description of the entire system CRBT speech recognition platform. RBT sound through the use of voice recognition platform moving CRBT IVR system architecture design techniques, combined with key technologies such as speech recognition and speech synthesis voice in the field, using an application system voice xml standard language tools developed. User by entering a specific number into the server, artist name or song form elected by voice command or button transmission to the system, you can get your own custom ring tones. By using a dedicated language interface to facilitate the operation of users, mainly using VXML technology to achieve the call flow. RBT RBT speech recognition system also features voice recognition platform scene and script code.
Keywords/Search Tags:Feature selection, speech recognition, Hidden Markov Model Customized ringing tone platform
PDF Full Text Request
Related items