Azure-speech-to-subtitle

Speech-to-Subtitle (word-level timestamp) code snippet based on MS Azure Speech.

基于微软 Azure Speech 的语音转字幕（字词级时间戳）的代码片段。

How my algorithm differs from RequestWordLevelTimestamps()?

我的算法与使用 RequestWordLevelTimestamps() 有何不同？

The DetailedSpeechRecognitionResult.Words property of the detailed output message obtained by RequestWordLevelTimestamps() is LexicalForm and contains profanity.

RequestWordLevelTimestamps() 获取的详细输出信息的 DetailedSpeechRecognitionResult.Words 属性内容是 LexicalForm，并且包含亵渎性语言。

Another difference is that the final result of my algorithm is that words are consecutive with no tiny time gaps between them, i.e. Word.Begin = PreviousWord.End.

另一个差异是，我的算法最终的结果是：单词之间是连续的，没有微小的时间空隙，即 Word.Begin = PreviousWord.End。

Code snippets in my gist:

代码片段在我的 gist:

https://gist.github.com/CodingOctocat/3d80252fd59d3895cae7ab370afd442e

Demo(Not included in gist/repo):

Demo(不包含在 gist/仓库):