The following information was submitted:
Transactions: WSEAS TRANSACTIONS ON SIGNAL PROCESSING
Transactions ID Number: 32-826
Full Name: Ali Eslamzadeh
Position: Please, select:
Age: ON
Sex: Male
Address: No.6 Kordestan Highway, West 27th street, 2nd 20meter street,Tehran,Iran
Country: IRAN
Tel:
Tel prefix:
Fax:
E-mail address: ali.eslamzade@gmail.com
Other E-mails:
Title of the Paper: Enhanced Neural Vector Quantizers for Bit Rate Reduction of LPC-10 Speech Coder
Authors as they appear in the Paper: ALI ESLAMZADEH
Email addresses of all the authors: ali.eslamzade@gmail.com
Number of paper pages: 10
Abstract: Abstract:-The goal of speech coding is to represent speech in digital form with as few bits as possible while maintaining the intelligibility and quality required for the particular application. This work is interested to use the ANNs to accomplish the goal of speech coding system. In this way, low bit rate at high quality speech signal is an important target in the voice telecommunications system. Linear predictive coder (LPC) parameters are widely used in various speech coding methods. On the other hand, line spectrum pairs (LSP), as alternate representation of LPC presents some useful properties for quantization and transmission. Artificial neural networks have been widely used as a useful means to improve quality of speech coders and reduce computations complexity of them. In this paper we have tried to reduce bit rate of the standard LPC-10 coder by applying two modifications in its algorithm: 1) Using LSP parameters instead of LPC parameters, 2) Employing fuz!
zy ARTMAP and a modified version of KSOFM as neural vector quantizers. Empirical results show that the bit rate of coder have been reduced to 1.9kbps and the quality of synthesized speech have been improved 0.13 and 0.26 respectively in terms of MOS scale. The execution time of algorithm, when using mentioned neural vector quantizer models, have been reduced significantly as compared to the standard LPC-10 speech coder.
Keywords: Key-words: -Speech Coding, LPC-10, Neural Vector Quantizer, KSOFM, Fuzzy ARTMAP, MOS
EXTENSION of the file: .doc
Special (Invited) Session:
Organizer of the Session:
How Did you learn about congress:
IP ADDRESS: 78.158.164.210
