ジラルジ アレサンドレ, シンガー ハラルド
Tied-Mixture Based SSS
HMnet Design
Abstract:This report describes a new approach to ML-SSS algorithm that uses tied-mixture representation of the output probability density function instead of single Gaussian during the splitting
phase of the SSS algorithm (Tied-Mixture SSS or TM-SSS algorithm). Due to this new representation we increase the recognition rate of the original ML-SSS algorithm by better choosing
the split state and the split itself. Implementation and results will be shown.