A Hands-on Comparison of DNNs for Dialog Separation Using Transfer Learning from Music Source Separation.
Martin StraussJouni PaulusMatteo TorcoliBernd EdlerPublished in: CoRR (2021)
Keyphrases
- transfer learning
- source separation
- cross domain
- labeled data
- audio features
- reinforcement learning
- blind source separation
- transfer knowledge
- independent component analysis
- semi supervised learning
- active learning
- activity recognition
- learning algorithm
- unlabeled data
- collaborative filtering
- machine learning
- machine learning algorithms
- denoising
- natural language
- single channel
- text categorization
- music information retrieval
- temporal structure
- image classification
- target domain
- knowledge discovery
- sound source
- support vector