Abstract: An improved method is disclosed for the encoding phase for Wavelet-based compression of digital data representing an object or image. Subband decomposition transforms the data into hierarchical tree data. Two terms are precomputed: the highest bit-plane in which any descendant of a node v has a non-zero bit (B1(v) and the highest bit-plane in which any indirect descendant (grandchild and beyond) becomes significant (B2(v). Node traversal is initiated but its extent is limited by a fixed bit-budget. The number of bits emitted by a node as a function of the bit-plane are counted through to the last bit-plane which exhausts the bit-budget. The user can predetermine the coefficient, bit-plane and specific bit at which the algorithm will halt. The tree structure is read only once, thus greatly accelereating the encoding and reducing the memory reference rate by a factor of 2 or more.

Abstract: An improved method is disclosed for the encoding phase for Wavelet-based compression of digital data representing an object or image. Subband decomposition transforms the data into hierarchical tree data. Two terms are precomputed: the highest bit-plane in which any descendant of a node v has a non-zero bit (B1(v) and the highest bit-plane in which any indirect descendant (grandchild and beyond) becomes significant (B2(v). Node traversal is initiated but its extent is limited by a fixed bit-budget. The number of bits emitted by a node as a function of the bit-plane are counted through to the last bit-plane which exhausts the bit-budget. The user can predetermine the coefficient, bit-plane and specific bit at which the algorithm will halt. The tree structure is read only once, thus greatly accelereating the encoding and reducing the memory reference rate by a factor of 2 or more.

Abstract: A multi-stage wavelet transform of an image signal is implementing using a first processing element to perform computations for a first stage of the transform, and a second processing element operating in a time-multiplexed manner to perform computations for subsequent stages of the transform. The first processing element includes first and second adder trees for implementing horizontal and vertical filtering operations, respectively, and a set of row buffers configured such that the total number of row buffers is only one more than the number of pixels required to generate a given vertically-filtered output. In a four-stage illustrative embodiment in which the first processing element processes image pixel data at a sample rate of fs, the multi-stage processing element receives inputs from the first stage processing element at a sample rate of fs/4, and generates coefficients for the second, third and fourth stages using sample rates of fs/16, fs/64 and fs/256, respectively.

Abstract: A system for, and method of, representing and resolving ambiguity in natural language text and a spoken dialogue system incorporating the system for representing and resolving ambiguity or the method. In one embodiment, the system for representing and resolving ambiguity includes: (1) a context tracker that places the natural language text in context to yield candidate attribute-value (AV) pairs and (2) a candidate scorer, associated with the context tracker, that adjusts a confidence associated with each candidate AV pair based on system intent.

Abstract: An interactive voice response unit which provides beneficial operation by including means to handle unconstrained input such as natural speech and to allow barge-in includes a prompter, a recognizer of speech signals, a meaningful phrase detector and classifier, and a turn-taking module, all under control of a dialog manager. In the course of listening to user input while outputting a voiced message, the voice response unit processes the received signal and ascertains whether it is receiving an utterance that is intended to interrupt the prompt, or merely noise or an utterance that is not meant to be used by the arrangement. The unit is sensitive to the speed and context of the speech provided by the user and is thus able to distinguish between a situation where a speaker is merely pausing and a situation where a speaker is done speaking.

Abstract: A multi-stage wavelet transform of an image signal is implementing using a first processing element to perform computations for a first stage of the transform, and a second processing element operating in a time-multiplexed manner to perform computations for subsequent stages of the transform. The first processing element includes first and second adder trees for implementing horizontal and vertical filtering operations, respectively, and a set of row buffers configured such that the total number of row buffers is only one more than the number of pixels required to generate a given vertically-filtered output. In a four-stage illustrative embodiment in which the first processing element processes image pixel data at a sample rate of fs, the multi-stage processing element receives inputs from the first stage processing element at a sample rate of fs/4, and generates coefficients for the second, third and fourth stages using sample rates of fs/16, fs/64 and fs/256, respectively.

Abstract: Data compression achieved by set partitioning of hierarchical trees is made more efficient by a family of related traversal schemes which act upon the tree structure and the significance values derived from the tree structure. The tree structure may be traversed in any of a number of sequences. In accordance with significance tests applied to the node data, bits are produced or not. The order in which the bits are produced may be freely chosen. Trees are traversed and bits are emitted which describe the relative magnitude of coefficients in the sub-trees of the current tree-node. The manner of traversal can either be known to the remote decoder or can be derived from the incoming data stream. The encoding scheme typically specifies all bits emitted by a given node v as a function of the traversal of its parents, the bit-plane b, the coefficient value c of v, the significances B1(v) and B2(v), and whether the parent node p still emits B2(p).

Abstract: An interactive voice response unit which provides beneficial operation by including means to handle unconstrained input such as natural speech and to allow barge-in includes a prompter, a recognizer of speech signals, a meaningful phrase detector and classifier, and a turn-taking module, all under control of a dialog manager. In the course of listening to user input while outputting a voiced message, the voice response unit processes the received signal and ascertains whether it is receiving an utterance that is intended to interrupt the prompt, or merely noise or an utterance that is not meant to be used by the arrangement. The unit is sensitive to the speed and context of the speech provided by the user and is thus able to distinguish between a situation where a speaker is merely pausing and a situation where a speaker is done speaking.

Abstract: An interactive voice response unit which provides beneficial operation by including means to handle unconstrained input such as natural speech and to allow barge-in includes a prompter, a recognizer of speech signals, a meaningful phrase detector and classifier, and a turn-taking module, all under control of a dialog manager. In the course of listening to user input while outputting a voiced message, the voice response unit processes the received signal and ascertains whether it is receiving an utterance that is intended to interrupt the prompt, or merely noise or an utterance that is not meant to be used by the arrangement. The unit is sensitive to the speed and context of the speech provided by the user and is thus able to distinguish between a situation where a speaker is merely pausing and a situation where a speaker is done speaking.

Abstract: A method and apparatus for enhanced speech recognition in a communication network in which a first input port is coupled to a first processor and receives a telephony signal from the communication network. A second input port, coupled to a second processor, receives the same telephony signal at substantially the same time as the first input port. Based on the telephony signal, the second processor generates recognized speech information. A control line coupled between the first and second processors lets the second processor send a command to the first processor, and the first processor changes state, such as by re-routing the telephony signal, based on the command. The second processor may also enter one of a plurality of states such that the state of the second processor corresponds to the state of the first processor at a given point during a telephone call. Instead of receiving the telephony signal, the second processor may receive speech data generated by the first processor.

Abstract: A signal processor for, e.g., a telephonic communication system interconnecting a near location and at least one far location includes an echo canceller for suppressing the re-transmission of received signals. The operation of the echo canceller is controlled according to whether the near location is transmitting or receiving, and whether the far location is transmitting or receiving. The signal processor includes means for detecting doubletalk, and for controlling the echo canceller in response to the detection of doubletalk, such that doubletalk is permitted. In one method of operation, a delay is applied to the inputs to the echo canceller from the near and far ends, such that each of these signals is used for determining the operating state of the echo canceller before being used to operate an adaptive filter of the echo canceller. In another method of operation, doubletalk detection is carried out, in part, by evaluating the performance of an auxiliary adaptive filter.