语音识别(Speech recognition)的核心内容是将语音转换成文字

https://www.bilibili.com/video/av16198207?from=search&seid=16570566229872205850

语音识别，又称为自动语音识别(Automatic Speech Recognition)、语音转文字(Speech to Text,STT)，是指让计算机自动将人类的语音内容转换为相应的文字。

语音识别(Speech recognition)的核心内容是将语音转换成文字。
Different from computer vision。Speech recognition only have one core task，which is converting human speech into normal language texts。

机器如何识别语音？
How could machine converting human speech into normal language texts？
语言由单词组成，单词由音素组成，我们将一段语音的声波按帧切开，用帧组成状态，用状态组成音素，再将音素合成单词，语音就变成了文字。与语音相关，仍属于人工智能研究范围内的任务还有不少。声纹识别，即识别说话这是谁；语音合成，即将文字信息转换成人类听得懂的语音。Siri、智能音箱、车载设备，都是语音识别看得见摸得着的应用。感觉效果不好？口音、距离、噪声都会影响识别结果。下次换个安静的环境试试。
下一期，我们将探讨自然语言处理,了解文字背后的秘密。
Language is built by words，words are built by phonemes. we cut a piece of speech by frames,then use frames to built status, use status to built phonemes.Finally ,we put phonemes together to built words, then speech is transformed to text. There are some of other tasks about speeches in the field of Artificial intelligence。Speaker recognition.which is to recognize who is the owner of the speech, Speech synthesis, which is converting normal language texts into human speech. Speech Recognition can be found in Siri, Amazon Echo and intelligent onboard terminals in our daily life.Feel bad?Accent,distance,noise will affect the recognition results. Next time you can try it in a quiet place.
We will talk about Natural Language Processing in the next video.Figure out how to make computers understand the meaning behind words.