Abstract: A method to provide automatic content-based video indexing from object motion is described. Moving objects in video from a surveillance camera 11 detected in the video sequence using motion segmentation methods by motion segmentor 21. Objects are tracked through segmented data in an object tracker 22. A symbolic representation of the video is generated in the form of an annotated graphics describing the objects and their movement. A motion analyzer 23 analyzes results of object tracking and annotates the graph motion with indices describing several events. The graph is then indexed using a rule based classification scheme to identify events of interest such as appearance/disappearance, deposit/removal, entrance/exit, and motion/rest of objects. Clips of the video identified by spatio-temporal, event, and object-based queries are recalled to view the desired video.

Abstract: A system for dynamic adaptation of wireless communication between a Mobile Station (11) and a Base Station (13) wherein the transmitted frame from the Mobile Station includes a convolutionally coded portion containing a down-link measurement bit and a repetition code identifying the codec mode of the frame. The transmitted frame from the Base Station (13) includes a codec mode command signal for the Mobile Station (11) in the convolutionally encoded portion and the repetition code identify the codec mode of the down-link frame. The Base Station (13) includes means for analyzing the quality of the up-link frame and means from the received down-link measurement bit for determining the down-link quality.

Abstract: The system uses utterances recorded in low noise condition, such as a car engine off to optimally adapt speech acoustic models to transducer and speaker characteristics and uses speech pauses to adjust the adopted models to a changing background noise, such as when in a car with the engine running.

Abstract: An improvement in the synthesis of high-pitched voices and sounds is provided by downshifting the pitch 11 of the original input voice or sound before LPC analysis 13. This downshifting of the pitch is provided by upsampling 21, low pass filtering 22, downsampling 23, and performing time scale modification 24.

Abstract: A text-to-pronunciation system (11) includes a large training set of word pronunciations (19) and an extractor for extracting language specific information from the training set to produce pronunciations for words not in its training set. A learner (13) forms pronunciation guesses for words in the training set and for finding a transformation rule that improves the guesses. A rule applier (15) applies the transformation rule found to guesses. The learner (13) repeats the finding of another rule and the rule applier (15) applies the new rule to find the rules that improves the guesses the most.

Abstract: A system (100) for interfacing hardware and software components in a simulation system is disclosed. The present invention includes at least one cell (110) having a model access and control hub (140) and at least one block (142) within the cell (110) connected to the model access and control hub (140). The block (142) executes a command (196) sent by the model access and control hub (110). The present invention also includes an interprocess communication device (104) including shared memory (105) coupled to the model access and control hub (140). The shared memory (105) communicates the command (196) to the model access and control hub (140). The present invention also includes a debugger (106) coupled to the interprocess communication device (104) that issues the command (196) to the model access and control hub (140) via the shared memory (105), and receives information from the model access and control hub (140) via the shared memory (105).

Abstract: The recognition of hands-free speech in a car environment has to deal with variabilities from speaker, microphone channel and background noises. A two-stage model adaptation scheme is presented. The first stage adapts speaker-independent HMM seed model set to a speaker and microphone dependent model set. The second stage adapts speaker and microphone-dependent model set to a speaker, microphone, and noise dependent model set, which is then used for speech recognition. Both adaptations are based on maximum-likelihood linear regression (MLLR).

Abstract: Utterance-based mean removal in log-domain, or in any linear transformation of log-domain, e.g., cepstral domain, is known to improve substantially a recognizer's robustness to transducer difference, channel distortion, and speaker variation. Applicants teach a sequential determination of utterance log-spectral mean by a generalized maximum a posteriori estimation. The solution is generalized to a weighted sum of the prior mean and the mean estimated from available frames where the weights are a function of the number of available frames.

Abstract: An automatic recovery method for a Die Bonder Wafer Table in the event of loss of wafermap coordinate data is provided. If after moving to the first map die position and there is no die, the wafer table is moved back one die position in the direction of the track from the reference die to the first map die and then the closest coordinate from the map data of current bin in reverse direction is found and the table is moved to that position. If there is alignment fail or no die, the same step of moving back and finding the closest die coordinate from the map data in reverse direction and moving to that position is repeated. Otherwise, the next die coordinate from the map data of the current bin in reverse direction towards the first map die is found and the table is moved to this die coordinate position. The table is moved to the next die position in the forward direction if alignment fails or no die is encountered.

Abstract: A multi-channel adaptive filter system for use, for example, in echo cancellation for generating an echo estimate for each channel is described. The system includes a forward and backward prediction filter parameter generator that in response to the multi-channel inputs generates a single forward prediction filter vector signal {right arrow over (&ohgr;)}p−1(n) and corresponding single forward error F(n) signal and a single backward prediction filter vector signal {right arrow over (&ngr;)}p−1(n) and corresponding single backward error signal B(n). The system further includes an error generator responsive to errors in the estimated signal for multiplying the errors by a 1-&mgr; where &mgr; is a constant for generating error vector signals {right arrow over (e)}i.p-{right arrow over (e)}J,p(n) for each channel.

Abstract: A method of enrolling phone-based speaker specific commands includes the first step of providing a set of (H) of speaker-independent phone-based Hidden Markov Models (HMMs), grammar (G) comprising a loop of phones with optional between word silence (BWS) and two utterances U1, and U2 of the command produced by the enrollment speaker and wherein the first frames of the first utterance contain only background noise. The processor generates a sequence of phone-like HMMs and the number of HMMs in that sequence as output. The second step performs model mean adjustment to suit enrollment microphone and speaker characteristics and performs segmentation. The third step generates an HMM for each segment except for silence for utterance U1. The fourth step re-estimates the HMM using both utterance U1 and U2.

Abstract: A memory management method is described for reducing the size of memory required in speech recognition searching. The searching involves parsing the input speech and building a dynamically changing search tree. The basic unit of the search network is a slot. The present invention describes ways of reducing the size of the slot and therefore the size of the required memory. The slot size is reduced by removing the time index, by the model_index and state_index being packed and by a coding for last_time field where one bit represents a slot is available for reuse and a second bit is for backtrace update.

Abstract: A method for N-best search for continuous speech recognition with limited storage space includes the steps of Viterbi pruning word level (same word, different time alignment, thus non-output differentiation) states and keeping the N-best sub-optimal paths for sentence level (output differentiation) states.

Abstract: Phonetic modeling includes the steps of forming triphone grammars (11) from phonetic data, training triphone models (13), clustering triphones (14) that are acoustically close together and mapping unclustered triphone grammars into a clustered model (16). The clustering process includes using a decision tree based on the acoustic likelihood and allows sub-model clusters in user-definable units.

Abstract: A system and method for providing a controllable virtual environment includes a computer (11) with processor and a display coupled to the processor to display 2-D or 3-D virtual environment objects. Speech grammars are stored as attributes of the virtual environment objects. Voice commands are recognized by a speech recognizer (19) and microphone (20) coupled to the processor whereby the voice commands are used to manipulate the virtual environment objects on the display. The system is further made role-dependent whereby the display of virtual environment objects and grammar is dependent on the role of the user.

Abstract: The invention provides methods and systems for power control in point-to-multipoint communications systems. Methods are described for both downstream and upstream control. Subscriber signals received at the base station 11 are level detected at receiver 77 and power adjust processing 79 and a control signal is sent to control output from the subscriber transmitter 65. The invention maximizes the signal-to-noise ratio and signal-to-interference ratio for subscribers (CPE) while minimizing transmitted power. The invention is ideal for use in local multipoint distribution service (LMDS) systems.

Abstract: A point-to-multipoint communication system wherein the base station broadcasts the same frequency band signals over multiple sectors of given node. The base station includes a plurality of channel receivers (ch. 1-ch. n) for each sector and a plurality of antennas 36a and 36b selectively coupled to said channel receivers. The subsector antennas have these radiation centers offset from each other to present different coverage over the node.

Abstract: A method of singing voice synthesis uses commercially-available MIDI-based music composition software as a user interface (13). The user specifies a musical score and lyrics; as well as other music control parameters. The control information is stored in a MIDI file (11). Based on the input to the MIDI file (11) the system selects synthesis model parameters from an inventory (15) of linguistic voice data units. The units are selected and concatenated in a linguistic processor (17). The units are smoothed in the processing and are modified according to the music control parameters in musical processor (19) to modify the pitch, duration, and spectral characteristics of the concatenated voice units as specified by the musical score. The output waveform is synthesized using a sinusoidal model 20.

Abstract: The system of the present invention having a first module (20) for storing information describing a plurality of data objects (150, 160, and 170), attributes of each of the plurality of data objects, and relationships therebetween; a second module (15) for accepting and processing user queries generating first results data, the first results data including the information describing the plurality of data objects stored in the first module; and a third module (85) responsive to the second module (15) for generating second results data using the first results data, the second results data including the information contained in the plurality of data objects.

Abstract: A system and method allow for associating a descriptive label with an anomaly on a manufactured object, such as a semiconductor wafer. The method includes placing the manufactured device on a moveable stage; capturing and preparing a digital-pixel-based representation of the image; symbolically decomposing the digital-pixel-based representation of the image to create a primitive-based representation of the image; analyzing the primitive-based representation of the image to detect and locate the anomaly; isolating primitives associated with the anomaly; comparing the isolated primitives associated with the anomaly with primitives in a knowledge base to locate a set of primitives in the knowledge base most like the isolated primitives associated with the anomaly; and assigning a label associated with the set of primitives in the knowledge base that was most similar to the isolated primitives associated with the anomaly.