| 摘要 |
A means (V6) is provided for updating at least one normalisation parameter as a function of a value preceding this parameter and of the verification score for each vocal segment test whenever the normalised score is at least equal to a second threshold value equal to or greater than the first threshold value. A means (A1-A3) is provided for generating parameters associated with an acceptance vocal model relative to a vocal segment spoken by the authorised speaker and for generating parameters associated with a refusal vocal model during a training stage. A means (V1-V3) is provided for normalising a verification score for the speaker using normalisation parameters, depending on the degree of resemblance between a vocal segment to be tested and the acceptance and refusal models. A means (V4) is provided for comparing the normalised verification score with a first threshold value so that the speaker of the vocal segment to be tested is only granted access if the normalised score is equal to or greater than this threshold value. |
