NON-NEGATIVE DURATIONAL HMM

Hamar, Jarle Bauck and Doddipatla, Rama Sanand and Svendsen, Torbjorn and Sreenivas, Thippur (2013) NON-NEGATIVE DURATIONAL HMM. In: 23rd IEEE International Workshop on Machine Learning for Signal Processing (MLSP), SEP 22-25, 2013, Southampton, ENGLAND.

PDF IEEE_Int_Wor_MLSP_2013.pdf - Published Version Restricted to Registered users only Download (259kB) | Request a copy

Abstract

Non-negative HMM (N-HMM) has been proposed in the literature as a combination of NMF (non-negative matrix factorisation) and HMM, to model a mixture of non-stationary signals using latent variables. The original formulation of N-HMM does not generalise to unseen data and hence limits its usage in automatic speech recognition (ASR). We propose modifications to the N-HMM formulation to generalise for unseen data and thereby making it suitable for ASR. The modified model is referred to as Non-negative durational HMM(NdHMM). We derive the EM algorithm for estimating the NdHMM parameters and show that the proposed model requires less number of parameters than conventional HMM.

Item Type:	Conference Proceedings
Series.:	IEEE International Workshop on Machine Learning for Signal Processing
Publisher:	IEEE
Additional Information:	Copy right for this article belongs to the 23rd IEEE International Workshop on Machine Learning for Signal Processing (MLSP), Southampton, ENGLAND, SEP 22-25, 2013
Keywords:	Non-negative matrix factorization;Hidden Markov model;Non-negative HMM; N-HMM;Non-negative durational HMM;NdHMM;Automatic speech recognition;ASR
Department/Centre:	Division of Electrical Sciences > Electrical Communication Engineering
Date Deposited:	24 Aug 2016 10:36
Last Modified:	24 Aug 2016 10:36
URI:	http://eprints.iisc.ac.in/id/eprint/54414

Actions (login required)

View Item