术语记录

ML

2022-02-23 10:13:25
目标检测：

• backbone：representation
• head：目标的坐标（框），类别（识别）。
• postprocess：从多个框中选框。

Speech

1. embedding 嵌入
2. speaker embedding 声纹编码：是一种representation，提取表示speaker的语音特征。
3. speaker diarization 人声分离
   speaker diarization 用深度学习怎么做？ - 鲤鲤的回答 - 知乎
4. speech production model：Speech production is the process by which thoughts are translated into speech. This includes the selection of words, the organization of relevant grammatical forms, and then the articulation想法、思想等的）语言表达 of the resulting sounds by the motor system using the vocal apparatus装置.
5. vocal cord 声带
6. vocal tract 声道
7. vibration 振动
8. periodic 周期的
9. glottal pulse 脉冲波
10. Formant Frequency: Resonance frequency (regions of emphasis of speech spectra) of the vocal tract is called the Formant Frequency 声道的共振频率（言语谱的重点区域）称为共振峰频率