Exploiting spectro-temporal structures using NMF for DNN-based supervised speech separation.
Shuai NieShan LiangHao LiXueliang ZhangZhanlei YangWenju LiuLike DongPublished in: ICASSP (2016)
Keyphrases
- negative matrix factorization
- nonnegative matrix factorization
- speech recognition
- speech synthesis
- unsupervised learning
- feature selection
- recognition engine
- semi supervised
- supervised learning
- learning algorithm
- endpoint detection
- text to speech
- speech signal
- audio visual
- document clustering
- training process
- matrix factorization
- supervised classification
- multiple instance learning
- sparse representation
- image representation
- dimensionality reduction
- pairwise
- neural network