| 摘要 |
The invention concerns a method and an apparatus for detecting vocal data flow in a packet data flow characterised by at least two measurable variables (X5Y). A distribution function (P{x 1 C}, P{y I C}) is provided of the values of each variable (X5Y) in a vocal data flow, and the values (x,y) of each variable (X,Y) are measured in order to obtain a sequence of measured values (x(k), y(k)) on a number (k) of blocks. Each measured value (x(k), y(k)) is therefore applied to the respective distributiion function (P{x 1 C}, P{yI C}) in order to generate a sequence of likelihood values Formula (I) from which respective average likelihood values (E [Bx],E[By]) are generated. The average values are processed in order to generate a reference likelihood value (B) which, when compared with a threshold likelihood value (Bmin) enables the presence of vocal data flow in a packet data flow to be detected. |
