Abstract: A system for detecting one or more keywords in continuous speech is disclosed wherein a speech processor extracts a plurality of analog speech parameters from the input continuous speech, an STV (Spectral Time Vector) generator circuit selectively converts the plurality of analog speech parameters into digitized spectral parameters, an asynchronous correlation circuit continuously correlates sequences of the digitized spectral parameters with stored reference mask templates representative of subelements (dyads or triads) of one or more desired keywords in order to produce continuous correlation data, and a decision function circuit is responsive to the correlation data for developing an occurrence decision output for each desired keyword that is detected.

Abstract: Apparatus for reducing (a) background noise in a signal channel, and (b) channel gain variations, useful with spectral analysis equipment. Two feed forward circuits are disclosed: one to enhance a signal in noise (deriving an estimated signal); the other for AGC (relative to a preselected level). Background noise is estimated as a bias level corresponding to a minimum signal level detected within a preselected interval, the duration of such interval corresponding to a spectral interval of interest. Such noise estimate is then subtracted from the noisy signal of interest to provide a compensatorily biased signal or estimated signal. The compensatorily biased or estimated signal is then subjected to compensatory gain control by employing the ratio of a reference level to a peak detected level occurring within the preselected time interval as a gain control signal.