1. Vector: Embedding, Latent Representation, Latent Code
2. Binary Classifier 評(píng)估 Encoder
3. Feature Disentangle 特征拆解
3.1 聲音變聲
3.2 IN & AdaIN
IN = Instance Normalization (remove global information)
AdaIN = Adaptive Instance Normalization (only influence global information)
4. Discrete Representation
Binary vector (參數(shù)較少,還可以識(shí)別沒(méi)有見(jiàn)到的樣本)
參考文獻(xiàn)
Machine Learning (2019,Spring)
Voice Conversion