Abstract: A voice processing system includes a real-time voice server, which is arranged to process real-time voice processing tasks for clients of the system. A gateway processor is arranged to accept from a client a request to perform an off-line voice processing task, to convert the off-line voice processing task into an equivalent real-time voice processing task, to invoke the voice server to process the equivalent real-time voice processing task, and to output a result of the equivalent real-time voice processing task.

Abstract: Methods and systems are provided for remote tuning and debugging of an automatic speech recognition system. Trace files are generated on-site from input speech by efficient, lossless compression of MFCC data, which is merged with compressed pitch and voicing information and stored as trace files. The trace files are transferred to a remote site where human-intelligible speech is reconstructed and analyzed. Based on the analysis, parameters of the automatic speech recognition system are remotely adjusted.

Abstract: A method for improving spoken language includes accepting a speech input from by a speaker using a language, identifying the speaker with a predetermined speaker category and correcting an error in the speech input using an error model that is specific to the speaker category.

Abstract: A method for organizing a sequence of video frames includes selecting one of the frames in the sequence as an initial frame in a first portion of a segment of the sequence and adding further frames in the sequence, subsequent the initial frame, to the first portion, while a measure of similarity of each of the added frames to the frames already in the first portion is within a first predefined bound. One of the added frames in the first portion is selected to be a representative frame for the segment. A second portion of the segment is generated by adding still further frames in the sequence, subsequent to the last frame in the first portion, to the second portion, while the measure of similarity of the added frame to the representative frame is within a second predefined bound. The first and second portions together are determined to constitute the segment that is represented by the representative frame.

Abstract: Determining image block significance based on region contrast, including partitioning an image into multiple regions of at least one pixel each, pairing any region with an unpaired, adjoining region, calculating a similarity value for any region pair, merging paired regions of any pair into a single region, where the region pair to be merged has a predefined similarity value, thereby creating a new image partition of multiple regions, repeatedly performing the preceding steps, thereby obtaining a sequence of image partitions, selecting within the image an image sub-area of a predefined size and shape, identifying a partition in the sequence in which the image sub-area is covered by any region to the extent of a predefined coverage measure, and calculating a region contrast significance value of the image sub-area as a value which is proportional to the position of the identified partition in the sequence.

Abstract: Methods and systems are provided for remote tuning and debugging of an automatic speech recognition system. Trace files are generated on-site from input speech by efficient, lossless compression of MFCC data, which is merged with compressed pitch and voicing information and stored as trace files. The trace files are transferred to a remote site where human-intelligible speech is reconstructed and analyzed. Based on the analysis, parameters of the automatic speech recognition system are remotely adjusted.

Abstract: A voice processing system includes a real-time voice server, which is arranged to process real-time voice processing tasks for clients of the system. A gateway processor is arranged to accept from a client a request to perform an off-line voice processing task, to convert the off-line voice processing task into an equivalent real-time voice processing task, to invoke the voice server to process the equivalent real-time voice processing task, and to output a result of the equivalent real-time voice processing task.

Abstract: A method for querying an electronic dictionary using letters of an alphabet enunciated by a user includes accepting a speech input from the user. The speech input includes a sequence of spelled letters enunciated by the user that spell a query word. The speech input is analyzed to determine one or more sequences of the letters that approximate the sequence of spelled letters. The one or more sequences of the letters are post-processed so as to produce a plurality of recognized words approximating the query word. The electronic dictionary is queried with the plurality of recognized words so as to retrieve a respective plurality of dictionary entries. A list of results including the plurality of recognized words and the respective plurality of dictionary entries is presented to the user.

Abstract: A method for adapting a game session on a first device for continued play on a second device, the method including establishing a game session on a first device, scaling the functionality of the game within the game session, and continuing the game session with the scaled game functionality on a second device.

Abstract: A method for analyzing an image includes constructing a graph to represent an object appearing in the image, and generating a string of symbols corresponding to the graph. The string is processed so as to classify the object.

Abstract: A method for improving spoken language includes accepting a speech input from by a speaker using a language, identifying the speaker with a predetermined speaker category and correcting an error in the speech input using an error model that is specific to the speaker category.

Abstract: Determining image block significance based on region contrast, including partitioning an image into multiple regions of at least one pixel each, pairing any region with an unpaired, adjoining region, calculating a similarity value for any region pair, merging paired regions of any pair into a single region, where the region pair to be merged has a predefined similarity value, thereby creating a new image partition of multiple regions, repeatedly performing the preceding steps, thereby obtaining a sequence of image partitions, selecting within the image an image sub-area of a predefined size and shape, identifying a partition in the sequence in which the image sub-area is covered by any region to the extent of a predefined coverage measure, and calculating a region contrast significance value of the image sub-area as a value which is proportional to the position of the identified partition in the sequence.

Abstract: A method for using a Java-based servlet running on a HTTP server to stream media content to a client. The method allows a servlet to optionally modify a media file, and to specify the byte range within the file to stream to the client.

Abstract: A method for capture of computer screens in a sequence of frames. A first set of one or more windows appearing in a first frame in the sequence is identified, each window in the set having respective first-frame window characteristics and window contents. A description of the first set of windows is encoded, indicative of the appearance of the computer screen in the first frame. In a second frame in the sequence, a second set of one or more windows is identified, having respective second-frame window characteristics and window contents, the second set including one or more windows corresponding respectively to one or more of the windows in the first set. One or more transformations are determined, which are applied to the first-frame window characteristics of the windows in the first set to generate the second-frame window characteristics of the corresponding windows in the second set.

Abstract: A method for analyzing an image includes constructing a graph to represent an object appearing in the image, and generating a string of symbols corresponding to the graph. The string is processed so as to classify the object.

Abstract: Device for generating multiple quality level bit-rates in a video encoder having a motion estimator providing a predicted block for each predefined block based upon estimating the motion between the predefined block of the current image and the corresponding block in the previous image, a transformer for transforming a prediction error resulting from the difference between the predicted block and the predefined block into the frequency domain, and a quantizer for quantizing the coefficients of the prediction error and providing the quantized coefficients to a video multiplex coding unit.

Abstract: A method of comparing a current block in a current frame to a plurality of background blocks in a preceding frame. Each block includes a matrix of data values arranged in a given number of columns. The method uses a processor which has a plurality of computational registers, each register capable of receiving a number of the data values at least equal to the given number. The data values are loaded from a memory beginning at an address in the memory evenly divisible by the given number. The method includes storing the preceding frame in the memory beginning at an address that is divisible by the given number with a predetermined remainder, loading at least some of the data values of the current block into one or more of the plurality of registers, loading at least some of the data values of one or more of the background blocks into another one or more of the plurality of registers, and comparing the background blocks to the current block using the registers.

Abstract: In a method and system for downloading graphic images on the Network, at least one high-resolution graphic image file of a reference image is stored at a network server. At least part of the reference image is compressed and downloaded at reduced resolution from the network server to a client connected to the network server. Upon suitable decompression, a corresponding low-resolution image is displayed and size data uniquely specifying a portion of the displayed image is uploaded from the client to the network server. The specified portion of the high-resolution graphic image file is extracted from the network server and downloaded to the client where it is displayed. Such an approach obviates the need to transfer high-resolution data intensive graphic files to the client and further allows the heavy image processing to be performed on the server much faster than is achieved in prior approaches where processing is performed locally at the client.

Abstract: Conditional replenishment device in a video encoder having a motion estimator (22) providing a predicted block for each predefined block based upon estimating the motion between the predefined block of the current image and the corresponding block in the previous image, transformer (16) for transforming a prediction error resulting from the difference between the predicted block and the predefined block into the frequency domain, and quantizer (20) for quantizing the coefficients of the prediction error and providing the quantized coefficients to a video multiplex coding unit (30). Such a conditional replenishment device includes a change detector for producing a segmentation map in which each pixel of the predefined block is marked as moving or stationary, a motion determinator for determining whether there is a group of moving pixels the number of which exceeds a predetermined threshold defining a moving object in the image, and a coder for encoding only blocks containing the group of moving pixels.