Improvement of esophageal speech by adaptive line enhancement with bias model

Authors:
Nuntaporn Tuangpermsub;Panuthat Boonpranuk;Wutthichai Polwisate;Prakasit Kayasith
Affiliations:
King Mongkut's University of Technology Thonburi, Tungkru Area, Bangkok;King Mongkut's University of Technology Thonburi, Tungkru Area, Bangkok;King Mongkut's University of Technology Thonburi, Tungkru Area, Bangkok;National Electronics and Computer Technology Center, Thathumthani, Thailand
Venue:
Proceedings of the 2nd International Convention on Rehabilitation Engineering & Assistive Technology
Year:
2008

Citing 3
Cited 2

Adaptive filter theory (2nd ed.)

Adaptive filter theory (2nd ed.)
Enhancement of Esophageal Speech by Injection Noise Rejection1

ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97)-Volume 2 - Volume 2
Analytic behavior of the LMS adaptive line enhancer for sinusoidscorrupted by multiplicative and additive noise

IEEE Transactions on Signal Processing

Esophageal speeches modified by the speech enhancer program®

Proceedings of the 7th International Convention on Rehabilitation Engineering and Assistive Technology
Training software for esophageal speech developed on game-based PC

Proceedings of the 7th International Convention on Rehabilitation Engineering and Assistive Technology

Quantified Score

Hi-index	0.00

Visualization

Abstract

The main objective of this research is to develop a portable speech enhancement device that would sequentially estimate clean speech and re-synthesis a more intelligible speech output for esophageal speaker. To resolve a real time processing problem, we propose a noise reducing algorithm called Adaptive Line Enhancement (ALE) [3][7] with bias model. [8] The method simultaneously reduces noises both from environment (or external noises) and from his/her voice (or internal noises) such as breathing noise, esophageal utterance noise. Incorporating a weighting coefficient bias into a conventional ALE filter, the filter has been proved empirically to perform better than the conventional one. In addition, the method can be applied without neither a laborious pre-training process nor a priori time consuming noise-power estimation.