论文部分内容阅读
日本电器公司(NEC)近日宣布开发出一种能将语音瞬间转换成文字的新型CPU(中央处理器)。这种CPU可用在个人电脑上,但因手机芯片的处理能力较弱,因此,该CPU在手机上的应用尚需时日。根据电子语音设备识别语音的原理,该CPU首先将声音信号转换为电磁波信号,然后和声音原型相对照并搜索出最合适的单词。因此,使用这种CPU 将声音转换成文字时,应首先把电磁波信号分割开来,然后分段读取,之后再整段输出。设计时可将三个小型CPU组合起来并行处理,语音的识别率约为 85%。目前,NEC的技术人员还在试图进一步开发能将日语语音瞬间转换为英语语音或者文字的语音系统。
NEC recently announced the development of a new CPU (central processing unit) that converts speech instantly into text. This kind of CPU can be used in the personal computer, but because the processing ability of the mobile phone chip is weak, so, this CPU takes time on the mobile phone. According to the principle of recognizing a voice by an electronic voice device, the CPU first converts the voice signal into an electromagnetic wave signal, and then contrasts with the voice prototype and searches for the most suitable word. Therefore, using this CPU to convert sound into text, you should first separate the electromagnetic wave signal, and then read the segment, then the entire output. Three small CPUs can be designed to be combined and processed in parallel, with a speech recognition rate of about 85%. At present, NEC technicians are still trying to further develop a speech system that can instantly convert Japanese speech into English speech or text.