Abstract
Solutions for two important problems for the deployment of noise-robust large vocabulary automatic speech recognizers using the missing data paradigm are presented. irst problem is the generation of missing data masks. We propose and evaluate a method based on vector quantization and harmonicity that successfully exploits the characteristics of speech while requiring only weak assumptions on the noise. A second problem that is addressed is computational efficiency. We advocate the usage of PROSPECT features and the L-cluster-Mbest method for Gaussian selection. In total, a speed up of a factor of about 6 can be achieved with these methods.
Keywords: Large vocabulary continuous speech recognition, missing feature, vector quantization, harmonicity.
Related Journals
Related Books
Computational Intelligence For Data Analysis
Machine Learning and Its Application: A Quick Guide for Beginners
Applied Machine Learning and Multi-Criteria Decision-Making in Healthcare
A First Course in Artificial Intelligence
Artificial Intelligence: Models, Algorithms and Applications
Advanced Computing Techniques: Implementation, Informatics and Emerging Technologies
Handbook of Mobile Application Development: A Guide to Selecting the Right Engineering and Quality Features
Ad Hoc and Sensor Wireless Networks: Architectures, Algorithms and Protocols
Applications of Modern High Performance Networks
Arduino Meets Matlab: Interfacing, Programs and Simulink