Abstract: The present invention is concerned with a signal processing system having a learning function pursuant to the back-propagation learning rule by the neural network, in which the learning rate is dynamically changed as a function of input values to effect high-speed stable learning. The signal processing system of the present invention is so arranged that, by executing signal processing for the input signals by the recurrent network formed by units each corresponding to a neuron, the features of the sequential time series pattern such as voice signals fluctuating on the time axis can be extracted through learning the coupling state of the recurrent network.

Abstract: The present invention is concerned with a signal processing system having a learning function pursuant to the back-propagation learning rule by the neural network, in which the learning rate is dynamically changed as a function of input values to effect high-speed stable learning. The signal processing system of the present invention is so arranged that, by executing signal processing for the input signals by the recurrent network formed by units each corresponding to a neuron, the features of the sequential time series pattern such as voice signals fluctuating on the time axis can be extracted through learning the coupling state of the recurrent network. The present invention modifies the prior art weight change algorithm .DELTA.W.sub.ji(n+1) =.eta...delta..sub.pi.+.alpha..W.sub.ji(n) into .DELTA.W.sub.ji(n+1) =.eta...beta.(.alpha..sub.pj O.sub.pi)+.alpha..W.sub.ji(n) where .beta..sub.j =1/(.SIGMA..sub.i O.sub.pi.sup.2 +1) is used to normalize the learning constant.

Abstract: A speech recognition method and apparatus in which a plurality of time-sequential acoustic parameters are derived from a segmented input speech signal and are used to form a first trajectory that is time normalized. The time-normalized trajectory is sampled at predetermined lengths therealong and the sampling results used to form a new time-normalized trajectory with equally spaced points ("dots" on the graph), which is compared with a plurality of previously registered trajectories until a match is found between the new time-normalized trajectory and one of the registered trajectories, at which time an indication of the results of the matching is produced. A silence acoustic parameter can also be added to the time-sequential acoustic parameters so that the time-normalized trajectory can start and end from a point of silence.