Abstract: A method in a communication terminal includes receiving a CDMA signal that was transmitted over a communication channel by a base station at a given transmission power, and received at the communication terminal with noise of a given noise level. The CDMA signal includes at least a pilot channel that was spread using a pilot spreading sequence. The received CDMA signal is de-spread with the pilot spreading sequence, and a pilot amplitude of the pilot channel is calculated based on the de-spread CDMA signal. Based on the received CDMA signal and the pilot amplitude, the transmission power of the base station and the noise level of the noise are estimated.

Abstract: A method for communication includes receiving at a receiver a signal from a transmitter embodying data encoded with an error correction code. The signal is processed in order to extract a sequence of samples in a complex signal space. Scalar values are extracted from the samples and the scalar values are processed so as to define one or more clusters of scalar data points. Gain and noise of the signal are estimated responsively to the defined clusters. Bit value metrics for the signal are computed based on the samples and the estimated gain and noise of the signal. The error correction code is decoded using the bit value metrics.

Abstract: A method for communication includes receiving signals at a receiver from one or more sources, including a target signal transmitted by a given transmitter. A channel response is estimated from the given transmitter to the receiver, and a filter response is computed by taking a sum of an autocorrelation of the received signals with an adaptive noise factor, and applying the sum to the estimated channel response. The filter response is applied to the received signals in order to recover the target signal.

Abstract: A method for communication includes receiving at a receiver a signal from a transmitter embodying data encoded with an error correction code. The signal is processed in order to extract a sequence of samples in a complex signal space. Scalar values are extracted from the samples and the scalar values are processed so as to define one or more clusters of scalar data points. Gain and noise of the signal are estimated responsively to the defined clusters. Bit value metrics for the signal are computed based on the samples and the estimated gain and noise of the signal. The error correction code is decoded using the bit value metrics.

Abstract: A method for communication includes receiving signals at a receiver from one or more sources, including a target signal transmitted by a given transmitter. A channel response is estimated from the given transmitter to the receiver, and a filter response is computed by taking a sum of an autocorrelation of the received signals with an adaptive noise factor, and applying the sum to the estimated channel response. The filter response is applied to the received signals in order to recover the target signal.

Abstract: A method for communication includes receiving signals at a receiver from multiple sources, including a target signal transmitted by a given transmitter, and estimating a channel response from the given transmitter to the receiver. A filter response is computed by taking a sum including a first autocorrelation of the received signals with a second autocorrelation of the channel response, and applying the sum to the estimated channel response. The filter response is applied to the received signals in order to recover the target signal.

Abstract: In an embodiment, a method includes performing a redundancy check to determine if a baseline bit sequence is compliant. When the baseline bit sequence is not compliant, the method additionally includes performing an iterative process until a compliant, candidate bit sequence is identified. The iterative process includes identifying one or more existing branches within a conceptual tree diagram, calculating scores for potential paths branching from the one or more existing branches, and performing a subsequent redundancy check on a next candidate bit sequence, which corresponds to a potential path that has a next lowest score, to determine if the next candidate bit sequence is compliant.

Abstract: Techniques and structures are provided for use in configuring an MMSE equalizer in a multiple antenna system in a computationally efficient manner.

Abstract: A nonlinear response function of a loudspeaker is determined by an iterative process during which the nonlinear response function and a linear response associated with an echo and microphone are alternately revised.

Abstract: In an embodiment, a method includes performing a redundancy check to determine if a baseline bit sequence is compliant. When the baseline bit sequence is not compliant, the method additionally includes performing an iterative process until a compliant, candidate bit sequence is identified. The iterative process includes identifying one or more existing branches within a conceptual tree diagram, calculating scores for potential paths branching from the one or more existing branches, and performing a subsequent redundancy check on a next candidate bit sequence, which corresponds to a potential path that has a next lowest score, to determine if the next candidate bit sequence is compliant.

Abstract: Briefly, a method and apparatus to generate a pronunciation network of a written word is provided. The generation of the pronunciation network may be done by receiving at least one pronunciation string of the written word from a phoneme string generator able to generate the pronunciation network of the written word. The pronunciation network may include a node list of phonemes combined from different pronunciation strings of the written word. A speech recognition apparatus based on the pronunciation network is also provided.

Abstract: A method and a system for corrective training of speech models includes changing a weight of a date sample whenever a data sample is incorrectly associated with a classifier and retraining each classifier with the weights.

Abstract: An HMM-based text-to-phoneme parser uses probability information within a probability database to generate one or more phoneme strings for a written input word. Techniques for training the text-to-phoneme parser are provided.

Abstract: A nonlinear response function of a loudspeaker is determined by an iterative process during which the nonlinear response function and a linear response associated with an echo and microphone are alternately revised.

Abstract: Briefly, in accordance with one embodiment of the invention, a recognition system may modify speech models using noise in an input signal that is received prior to a speech sample in the input sample.