ostendorf96fromhmmstosegmentmodels - 360 IEEE TRANSACTIONS...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
360 IEEE TRANSACTIONS ON SPEECH AND AUDIO PROCESSING, VOL 4, NO 5, SEPTEMBER 1996 ’s nified View sti ecognition Mari Ostendorf, Member, IEEE, Vassilios V. Digalakis, and Owen A. Kimball, Member, IEEE Abstract--In recent years, many alternative models have been proposed to address some of the shortcomings of the hidden Markov model (NMM), which is currently the most popular approach to speech recognition. In particular, a variety of models that could be broadly classified as segment models have been described for representing a variable-length sequence of obser- vation vectors in speech recognition applications. Since there are many aspects in common between these approaches, including the general recognition and training problems, it is useful to consider them in a unified framework. Thus, the goal of this paper will be to describe a general stochastic model that encompasses most of the models proposed in the literature, pointing out similari- ties of the models in terms of correlation and parameter tying assumptions, and drawing analogies between segment models and WMM’s. In addition, we summarize experimental results assessing different modeling assumptions and point out remaining open questions. I. INTRODUCTION 0 date, the most successful speech recognition sys- tems have been based on the hidden Markov model (HMM) [l], [2], and the use of HMM’s for acoustic modeling dominates the continuous speech recognition field. Although HMM’s will continue to play a role in most recognition systems for a long time to come, many alternative models have been proposed in recent years to address some of the shortcomings of HMM’s. These new higher order models tend to require more computation than HMM’s but with the increase in computational power and the broad use of progressive search techniques, they are viable and of interest for current systems. Unfortunately, the research on new models has tended to proceed in isolated pockets, and the proliferation of terms used to describe different modeling assumptions has made it difficult to appreciate the common themes across the various proposals. The goal of this paper is thus to bring together a variety of work under a common framework in order to make it easier for different researchers to benefit from the successes of others in developing robust estimation techniques and making appropriate assumptions about variable dependence and parameter tying. Manuscript received June 20, 1995; revised February 22, 1996. This work was funded by ARPA and ONR under grant number ONR-N00014-92-J-1778, with additional support for M. Ostendorf provided by ATR. The associate editor coordinating the review of this paper and approving it for publication was Dr. Douglas D. O’Shaughnessy. M. Ostendorf is with the Electrical, Computer, and Systems Engineering Department, Boston University, 44 Cummington St., Boston, MA 02215 USA.
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 05/08/2010 for the course CS 6.345 taught by Professor Glass during the Spring '10 term at MIT.

Page1 / 19

ostendorf96fromhmmstosegmentmodels - 360 IEEE TRANSACTIONS...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online