多语言语音识别翻译系统

Translation System of Multilingual Automatic Speech Recognition

设计了一种多语言语音识别翻译系统,以实际需求为出发点,使用Python为基本开发语言,文本翻译可以选择有道和百度翻译两种引擎的翻译结果作对比。语音翻译主要在主程序中调用了Speech Recognition库和百度短语音识别API进行语音识别。视频中文字的提取功能主要使用了ffmpeg库和百度开放API接口进行了设计。经测试,文本翻译能够实现多达27种语言的互译,语音的识别翻译能够进行中英互译,视频文字提取能够对MP4视频文件中的文字进行提取并翻译,并且软件运行稳定。

A kind of translation system of multilingual automatic speech recognition is designed.According to the actual demand,Python is used as the basic development language.In text translation,comparison of the results of Youdao and Baidu is done.In speech-to-speech translation,Speech Recognition and Baidu short automatic speech recognition API is used to do speech recognition in direct procedure.In literal extraction in video function,the connector of ffmpeg and Baidu open API is designed.Through testing,intertranslation of 27 languages can be achieved in literal translation,Chinese-English translation can be achieved in phonetic recognition translation,and characters of MP4 video file can be extracted and translated in video characters extraction.The software operates smoothly.