 |
ESCA - NATO/RSG 10 Workshop on
|
 |
|
ESCA - NATO/RSG 10 Workshop on
|
|
The paper reviews the current state of speech coding technology with reference to applications involving the storage of speech and its transmission in computer networks. It examines the different constraints placed on algorithms for storage, and explores the design flexibility which this allows. An example algorithm is used to illustrate these points. The paper also compares the advantages and disadvantages of speech coding with speech synthesis and text to speech for message storage. Text to speech has the advantage of the lowest overall bit rate, less than 200 b/s, and can provide an unlimited vocabulary allowing messages to be constructed for unforeseen dialogue but has the lowest speech quality. Speech synthesis requires the off-line construction of a predetermined set of messages, often involving the skilled editing of the speech parameters, but offers higher quality at bit rates around 2 kb/s. Speech coding gives the highest quality and does not require off-line editing. The falling cost of storage media will result in it becoming the preferred technique for applications where the dialogue can be predicted in advance.
Bibliographic reference. Carey, M. J. / Parris, C. J. / Wong, D. Y. (1993): "Speech coding for storage and transmission", In AST-1993, 27-30.