摘要 |
PROBLEM TO BE SOLVED: To enable an apparatus, whereby two parties located at different places make a speech mutially, to excellently perform a call speech under various noise environments. SOLUTION: The voice speech apparatus selects an average signal strength for a prescribed time of voice signals received from a transmission input terminal as an estimate noise level and classifies the right and wrong of a speech environment to a plurality of predetermined stages in response to the strength of the estimate noise level. The voice speech apparatus classifies a speech state into three speech states: an under low noise speech state wherein the noise level is low and processing of loss insertion amount estimate, echo cancellation and voice signal detection can respectively be attained; an under noise speech state wherein the noise level is high but echo cancellation can still be attained; and an under high noise speech state wherein the noise level is high and each processing above is no longer available. The voice speech apparatus executes loss insertion adaptable to the respective states after executing the classifications above. COPYRIGHT: (C)2008,JPO&INPIT
|