知识卡片
语音前后端会引入额外误差
语音识别先把声音转文字,核心对话系统处理,再由语音合成输出声音。识别错字会传到意图和槽位,合成语调又影响理解;听起来像人不证明任务能力提高,需分别测延迟、识别和输出可懂度。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
Giving your chatbot a voice Though we haven’t talked much about voice processing in this book, you may wonder if the NLP tools you’ve learned can help you build a voice assistant similar to Siri or Mycroft. To build a voice assistant, you can preprocess the inputs of your chatbot with existing voice recognition or speech-to-text (STT) software. You can then use speech generation, or
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第12章《对话引擎与会话设计》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-12.html)
- 结论依据:摘录中关于语音前后端会引入额外误差的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:Giving your chatbot a voice Though we haven’t talked much about voice processing in this book, you may wonder if the NLP tools you’ve learned can help you build a voice assistant similar to Siri or Mycroft. To build a voice assistant, you can preprocess the inputs of your chatbot with existing voice recognition or speech-to-text (STT) software. You can then use speech generation, or