系统语音听写语法改进 [英] System Speech DictationGrammar Improvements

问题描述

我现在一直在尝试System.Speech DictationGrammar，发现我试图从音频质量的最差组合中获得结果.即从具有许多扬声器的电话质量波形中全文转录而没有 识别引擎的语法构建/训练.结果令人失望，我有什么办法可以提高赔率?

I have now been experimenting with the System.Speech DictationGrammar and discovered that I am trying to get results out of the worst combination of audio qualities. Namely full text transcription from telephone quality wavs with numerous speakers and no grammar building/training of the recognition engine. The results have been disappointing, is there anything I can do to improve the odds?

-增加音频采样率+降噪会有所帮助吗?
-我可以使用现有的录音来训练程序吗?
-在这种情况下是否要求全文转录?

- will increasing audio sample rate + noise reduction help?
- can I train the program using the existing recordings?
- is asking for full text transcription asking too much in this scenario?

推荐答案

嗨 TruObsidian ，
我正在尝试让对此主题有更多经验的人们参与.请耐心等待，待进一步处理后，我们会尽快为您提供更多信息.
感谢您的理解！

Hi TruObsidian ,
I am trying to involve people who have more experience on this topic. Please be patient, we will back to give you more information soon when get further process on it.
Thanks for your understanding!

关于

这篇关于系统语音听写语法改进的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！