Technical Program

Paper Detail

Paper:	SLP-L8.2
Session:	Efficient Techniques for LVCSR
Time:	Thursday, May 18, 14:20 - 14:40
Presentation:	Lecture
Topic:	Speech and Spoken Language Processing: Resource constrained ASR for portable/mobile devices
Title:	Speaker-Independent Name Recognition using Improved Compensation and Acoustic Modeling Methods for Mobile Applications
Authors:	Kaisheng Yao, Lorin Netsch, Vishu Viswanathan, Texas Instruments, Inc., United States
Abstract:	Name recognition is an important application of automatic speech recognition in embedded devices. Since embedded devices are used in diverse environments, noise robustness is very important. Moreover, unlike normal computer-based speech recognition applications, embedded speech recognition must deal with problems arising from limited resources. Facing these challenges, we have developed environment compensation and acoustic modeling techniques that improve robustness and accuracy of a speaker-independent name recognition system in hands-free conditions. These techniques are efficient to implement and are effective for performance improvement. On a name recognition task, we observed more than 53% word error rate reduction, compared to a baseline system. These improvements were obtained with minimal increase of resources.