subject: Intelligent communication terminals and key technologies research - communication terminal, speech s [print this page] Intelligent communication terminals and key technologies research - communication terminal, speech s
Abstract Modern communications is toward digital, broadband, integrated, intelligent, and personal direction, this high-performance communications terminals, intelligent, multimedia, mobile technology such as miniaturization and raised higher requirements. This article discusses the communication terminal intelligent intelligent voice interface; speech synthesis, recognition and portal; key technologies such as adaptive networks. The name suggests, intelligent communication terminal is a communication terminal with smart features, which should be a fundamental characteristic of vision, hearing, speaking, understanding and self-learning, reasoning ability, performance personified in a man-machine interface and friendly interface, the application of speech recognition and synthesis, character recognition, natural language understanding and other key technologies of voice and language processing. 1 intelligent speech interface technology Intelligent interface, also known as human-machine interface, intelligent communication terminals in the establishment of an anthropomorphic human-computer interaction environment. Intelligent interface technology not only involves the addition of artificial intelligence and computer science, and signal processing, electronics, linguistics, cognitive science, and even modern psychology, neurophysiology, etc. are closely related. Currently, in the field of the main research topics are multimedia and virtual reality, speech recognition and synthesis, character recognition, computer vision and image recognition, natural language processing, automatic summarization and machine translation. Because the human voice is the most natural means of communication, information exchange of people, 60% is accomplished by voice communication, so the main body of the present communication terminal is also based on voice communication. 1.1 Speech Recognition Speech recognition is a smart communication terminal one of the most influential technology. Since the early speech recognition is a "template" to achieve, but this speech recognition based on template matching, can only meet a small vocabulary, specific people, the demand for non-continuous speech recognition, can not provide the people of the large vocabulary, non-specific people, the need for continuous speech recognition. The demand for intelligent communication of voice recognition technology, in turn, has put forward higher requirements. Telephone speech recognition in high-speed data channel and the popularity of portable data terminals, we are still the first use of traditional voice telephone channel and a general sense, the telephone terminal, which implies the need to identify through the telephone channel and resulted in a distortion of the voice signal. Call the user may be in a variety of noisy environments, background noise so voice recognition is to solve the problem. Telephone speech recognition is the most pressing market demand, has become the focus of research and development at home and abroad. robust speech recognition, all kinds of noise, especially generated by mobile phones coding and channel noise, is to identify the original speech signal a major disturbance, and the random occurrence of such interference. Therefore, the need for a can of environmental, channel noise, filtering or in the recognition process to correct techniques. In view of the user groups were the case, need to establish more general model of moderate or automatically adapt to different accents, the system has strong robustness. spoken telephone speech recognition, speech recognition technology when used to implement telephone inquiries, auto wiring, and special operations (travel information service, etc.) operation, there will be many cases incompatible with the grammar, resulting in redundant information, which for speech recognition bring big challenges. multi-lingual speech recognition Internet age is bound to involve multiple languages, such as voice recognition for the metropolitan telephone inquiry service, the user may use Mandarin, Cantonese, or English Query, which requires recognition system more language recognition capabilities. Distributed Speech Recognition distributed speech recognition is done on the client terminal voice signal parameter extraction work, but only the parameter sent to the server side for further identification. Not only solve the channel noise and the calculation of the current preparation of the terminal storage capacity, also has a narrow bandwidth, integrated and low cost, but the premise is the extraction of parameters must be standardized. With the personal mobile communication needs, communication terminal become smaller, a range of handheld devices entering people's lives, such as mobile phones, Shangwutong, personal assistants.
