Conference/Proceedings5th International Linux Audio Conference, LAC2007
Start date22.03.2007
End date25.03.2007
AddressBerlin, Germany
Author(s)Jan Weil, Kai Cl├╝ver, Thomas Sikora
TitleReal-Time Multiple-Description Coding of Speech Signals
AbstractWhen sending speech data over lossy networks like the internet, multiple-description (MD) coding is a means to improve the perceived quality by dividing the data into multiple descriptions which are then sent as separate packets. In doing so the speech signal can still be decoded even if only parts of these descriptions are received. The present paper describes the structure of a software which demonstrates the benefits of this coding scheme using a client-server architecture.
Key wordsMultiple-Description Coding, Speech Coding, Real-Time Coding