Speech detection device for detecting a speech signal in an input signal

A speech detection device for detecting a speech signal in an input signal, including input means arranged to receive an input signal which can contain at least one of a speech signal and a tone signal, and including processing means for processing the input signal and for supplying at least a first and a second characteristic information of the input signal, and including threshold test means for checking whether the first and the second characteristic information exceed a threshold value and for supplying at least one item of detection information, and including evaluation means for evaluating the at least one item of detection information and for supplying speech detection information which is characteristic of the presence of a speech signal in the input signal, wherein the processing means include filter means adapted to extract a first frequency band signal from a first frequency band of the input signal as first characteristic information and to extract a second frequency band signal from a second frequency band of the input signal as second characteristic information, the first frequency band and the second frequency band not overlapping one another, and the threshold test means is adapted to supply first detection information when the first frequency band signal exceeds a first amplitude threshold value, and to supply second detection information when the second frequency band signal exceeds a second amplitude threshold value, when only the first detection information or only the second detection information occurs within a test time interval, the threshold value test means being adapted to supply third detection information when a third frequency band signal extracted, respectively, from a low frequency band of the first frequency band or the second frequency band of the input signal by the filter means exceeds a third amplitude threshold value, and to supply fourth detection information when a fourth frequency band signal extracted, respectively, from a high frequency band of the first frequency band FBI or the second frequency band of the input signal by the filter means exceeds a fourth amplitude threshold value, and the evaluation means are adapted to supply speech detection information when the first detection information and the second detection information within a test time interval appear at least for a threshold time interval and when the third detection information and the fourth detection information appear within the test time interval at least for the threshold time interval, and

• wherein there have been provided history memory means adapted to store at least detection information supplied by the threshold value detection means after a first test time interval, and the evaluation means are further adapted to evaluate at least the detection information of the first and of a second test time interval after the second test time interval which follows the first test time interval.