A two-stage singing voice separation algorithm using spectro-temporal modulation features.
Frederick Z. YenMao-Chang HuangTai-Shih ChiPublished in: INTERSPEECH (2015)
Keyphrases
- detection algorithm
- cost function
- high accuracy
- np hard
- learning algorithm
- experimental evaluation
- computational complexity
- objective function
- k means
- estimation algorithm
- matching algorithm
- computational cost
- feature selection algorithms
- search space
- temporal information
- visual features
- expectation maximization
- particle swarm optimization
- similarity measure
- probabilistic model
- dynamic programming
- low level
- significant improvement
- worst case
- genetic algorithm
- times faster
- optimal solution