Underdetermined Sound Source Separation Using Power Spectrum Density Estimated by Combination of Directivity Gain.
Yusuke HiokaKen'ichi FuruyaKazunori KobayashiKenta NiwaYoichi HanedaPublished in: IEEE Trans. Speech Audio Process. (2013)
Keyphrases
- power spectrum
- source separation
- mixing matrix
- blind source separation
- frequency domain
- natural images
- density function
- independent component analysis
- denoising
- single channel
- sound source
- source signals
- audio features
- multiscale
- higher order
- non stationary
- mean shift
- machine learning
- temporal structure
- blind separation