Title: Gaussian Mixture Model Weight Supervector Decomposition and Adaptation
Authors: Bahari, Mohamad Hasan
Dehak, Najim
Van hamme, Hugo
Issue Date: 5-Jun-2013
Series Title: Technical report KUL/ESAT/PSI/1302, KU Leuven, ESAT, Leuven, Belgium
Abstract: This report proposes a novel approach for Gaussian Mixture Model (GMM) weights decomposition and adaptation. This modeling suggests a new low-dimensional utterance representation method, which uses a simple factor analysis similar to that of the i-vector framework. The suggested approach is applied to the Robust Automatic Transcription of Speech (RATS) language identification evaluation corpus,
where the speech recordings are from highly degraded communication channels. In our experiments, after modeling each utterance using the proposed approach, a Deep Belief Networks (DBN) is utilized to recognize the language of utterances.The assessment results show that the proposed method improves conventional maximum likelihood weight adaptation. It is also shown that the absolute and relative improvement obtained by the score-level fusion of the i-vector framework and the proposed method are 5% and 17% respectively.
Description: Bahari M.H., Dehak N., Van hamme H., ''Gaussian mixture model weight supervector decomposition and adaptation'', Technical report KUL/ESAT/PSI/1302, KU Leuven, ESAT, June 2013, Leuven, Belgium.
Publication status: published
KU Leuven publication type: IR
Appears in Collections:ESAT - PSI, Processing Speech and Images

Files in This Item:
File Description Status SizeFormat
bare_jrnl_14.pdf Published 214KbAdobe PDFView/Open Request a copy
merge.pdf Published 270KbAdobe PDFView/Open

These files are only available to some KU Leuven Association staff members


All items in Lirias are protected by copyright, with all rights reserved.