White Papers

Large Margin Hidden Markov Models for Automatic Speech Recognition

Overview The paper studies the problem of parameter estimation in continuous density hidden Markov models (CD-HMMs) for Automatic Speech Recognition (ASR). As in support vector machines, the paper proposes a learning algorithm based on the goal of margin maximization. Unlike earlier work on max-margin Markov networks, the approach is specifically geared to the modeling of real-valued observations (such as acoustic feature vectors) using Gaussian mixture models. Unlike previous discriminative frameworks for ASR, such as maximum mutual information and minimum classification error, the framework leads to a convex optimization, without any spurious local minima. The objective function for large margin training of CD-HMMs is defined over a parameter space of positive semidefinite matrices.

By downloading you agree to our Terms and Conditions. These include information regarding use of your personal data.

Publisher: University of California
File Format: PDF
Date Published: Dec 13, 2008
Format: White Papers
Topics: Voice Recognition, Software Engineering

Similiar White Papers

MSDN Webcast: Building Production Speech Applications for Non-Speech Experts (Level 300)

The attendee of the webcast will learn how to build, debug, deploy and install production-quality speech interactive voi

Publisher: Microsoft | Tags: applications

Market-Leading Data-Modeling Tools: Research Report from the Burton Group

The Burton Group provides an in-depth research report on Market-Leading Data-Modeling Tools. According to their resear

Publisher: Sybase | Tags: data, it department

Improving Speech Recognition on a Mobile Robot Platform Through the Use of Top-Down Visual Queues

In many real-world environments, Automatic Speech Recognition (ASR) technologies fail to provide adequate performance fo

Publisher: University of Bremen | Tags: applications, robotics, speech recognition

The Potential Impact of Speech Recognition Technology on Workplace Productivity

Automatic Speech Recognition (ASR) software embedded in popular word processing programs shows great promise, offering t

Publisher: Bryant University | Tags: human computer interface, software, voice recognition

Building Domain Specific Languages for Voice Recognition Applications

This paper presents a method of implementing the voice recognition for the control of software applications. The solutio

Publisher: Soft Mentor | Tags: applications, software, software applications, voice recognition

University of California White Papers

Stateless Load Balancing Over Multiple MPLS Paths

Featured white papers

The Value of Location Intelligence in the Communications Industry

Pitney Bowes

Public Services are under pressure, the challenge is to do more with less. How do you improve citizen satisfaction, increase cost efficiencies and improve service delivery? The power of location intelligence is helping many local authorities...

Download now »
Best Practices for Translating Customer Satisfaction into Revenue

Citrix Online

Today's support organisations are focused on two top-level metrics: financial results and customer satisfaction. For most, it's easy to track financial performance, but customer satisfaction is akin to speaking a foreign language...

Download now »
HP print solutions and 3M

Hewlett-Packard (HP)

The objective for 3M was to optimize office printing infrastructure at 3M locations worldwide, reduce total cost and environmental footprint. Some of the business benefits acheived by switching to HP print solutions...

Download now »