Abstract: Radio Frequency Identification (RFID) tags are used for automatically determining the connectivity or alignment between physical components, including, for example, connectivity of network cables and device ports, as well as alignment of components assembled by automated manufacturing systems. In one embodiment of the invention, accurate determinations of the physical three-dimensional locations of cables and equipment are employed to determine which cables are plugged into which device ports of which pieces of equipment. In another embodiment of the invention, multiple RFID tags are used to determine the appropriate alignment between components being assembled by an automated manufacturing system.

Abstract: A method and apparatus which reduces the start-up delay that may occur when switching programs in audio and/or video streaming applications while maintaining high quality steady-state performance thereof. A program source (e.g., an audio and/or video data stream) is encoded and transmitted as two or more separate bit streams (e.g., sequences of data packets), the transmission of one of these bit streams being delayed by a given amount of time relative to the transmission of the other bit stream(s). At the receiving end of the transmission channel, the two or more bit streams are buffered by receive buffers having different sizes (thereby resulting in different time delays when the contents thereof are decoded), wherein the time delay difference corresponds (inversely) to the relative delay times prior to transmission.

Abstract: A real-time passive acoustic source localization system for video camera steering advantageously determines the relative delay between the direct paths of two estimated channel impulse responses. The illustrative system employs an approach referred to herein as the “adaptive eigenvalue decomposition algorithm” (AEDA) to make such a determination, and then advantageously employs a “one-step least-squares algorithm” (OSLS) for purposes of acoustic source localization, providing the desired features of robustness, portability, and accuracy in a reverberant environment. The AEDA technique directly estimates the (direct path) impulse response from the sound source to each of a pair of microphones, and then uses these estimated impulse responses to determine the time delay of arrival (TDOA) between the two microphones by measuring the distance between the first peaks thereof (i.e., the first significant taps of the corresponding transfer functions).

Abstract: A method and apparatus for synthesizing speech from text whereby the speech may be generated in a manner so as to effectively convey a particular, selectable style. Repeated patterns of one or more prosodic features—such as, for example, pitch, amplitude, spectral tilt, and/or duration—occurring at characteristic locations in the synthesized speech, are advantageously used to convey a particular chosen style. For example, one or more of such feature patterns may be used to define a particular speaking style, and an illustrative text-to-speech system then makes use of such a defined style to adjust the specified parameter or parameters of the synthesized speech in a non-uniform manner (i.e., in accordance with the defined feature pattern or patterns).

Abstract: A method and apparatus for advantageously selecting video frames to be coded in order to improve the coding quality of a low bit-rate coder. In particular, temporal sub-sampling (i.e., selecting a set of frames to be coded from the complete incoming sequence of frames) is performed so that the frames which are to be coded are advantageously selected based upon a coding criterion, such as, for example, prediction gain (i.e., reduction in DFD variance). Specifically, in one illustrative embodiment, a larger number of frames are advantageously selected during periods of fast change, and correspondingly fewer frames are selected during other periods, while thereby keeping the overall apparent frame-rate fixed.

Abstract: A method and apparatus for performing real-time endpoint detection for use in automatic speech recognition. A filter is applied to the input speech signal and the filter output is then evaluated with use of a state transition diagram (i.e., a finite state machine). The filter is advantageously designed in light of several criteria in order to increase the accuracy and robustness of detection. The state transition diagram advantageously has three states. The endpoints which are detected may then be advantageously applied to the problem of energy normalization of the speech portion of the signal.

Abstract: A method and apparatus for performing double-talk detection in an acoustic echo canceller in which a detection statistic is advantageously computed based on an estimate of a cross-correlation between the far-end signal and the return signal which has been normalized with use of an estimate of a covariance matrix of the far-end signal. The estimate of the cross-correlation between the far-end signal and the return signal may be further normalized with use of either an estimate of a variance of the return signal or an estimate of a covariance matrix of the return signal. In certain illustrative embodiments of the invention, one or more of these quantities may be estimated based on signal samples sampled over a predetermined time window. And in another illustrative embodiment of the present invention, the coefficients of the adaptive filter employed in the acoustic echo canceller itself are advantageously used to compute the detection statistic.

Abstract: A method and apparatus for performing automatic speech recognition (ASR) in a distributed ASR system for use over a wireless channel takes advantage of probabilistic information concerning the likelihood that a given, portion of the data has been accurately decoded to a particular value. The probability of error in each feature in a transmitted feature set is employed to improve speech recognition performance under adverse channel conditions. Bit error probabilities for each of the bits which are used to encode a given ASR feature are used to compute the confidence level that the system may have in the decoded value of that feature. Features that have been corrupted with high probability are advantageously either not used or are weighted less in the acoustic distance computation performed by the speech recognizer.

Abstract: A method for providing packet loss recovery in a data packet-based network used for real-time multimedia communications. In accordance with a first illustrative embodiment of the present invention, the information payload associated with a given data packet k is identically copied and appended to data packet k+w (i.e., the information payload is repeated with a delay of w transmitted packets). More generally, the present invention provides a method of coding a sequence of data packets representing a contiguous stream of information, with each data packet comprising, a set of payload information representative of a segment of the stream of information corresponding thereto.

Abstract: A method and apparatus for extracting speech features from a speech signal in which the linear frequency spectrum data, as generated, for example, by a conventional frequency transform, is first converted to logarithmic frequency spectrum data having frequency data distributed on a substantially logarithmic (rather than linear) frequency scale. Then, a plurality of digital auditory filters is applied to the resultant logarithmic frequency spectrum data, each of these filters having a substantially similar shape, but centered at different points on the logarithmic frequency scale. Because each of the filters have a similar shape, the feature extraction approach of the present invention advantageously can be easily modified or tuned by adjusting each of the filters in a coordinated manner, with the adjustment of only a handful of filter parameters.

Abstract: A method and apparatus for coding and decoding a sequence of data packets with use of a novel class of forward error correcting codes having coding rates greater than 1/2 which nonetheless provide relatively high levels of channel protection against burst erasures with a relatively low decoding delay. In accordance with certain illustrative encoder embodiments of the present invention, the source information contained in each of a plurality of packets to be coded is similarly divided into a plurality of (similar) corresponding portions, and “checksums” are computed over multiple data packets, each such checksum being based on different (i.e., non-corresponding) portions of at least two of the multiple packets. These “checksums” are then advantageously appended to various subsequent data packets to be coded.

Abstract: A method and apparatus for performing text-to-speech conversion in a client/server environment partitions an otherwise conventional text-to-speech conversion algorithm into two portions: a first “text analysis” portion, which generates from an original input text an intermediate representation thereof and a second “speech synthesis” portion, which synthesizes speech waveforms from the intermediate representation generated by the first portion (i.e., the text analysis portion) The text analysis portion of the algorithm is executed exclusively on a server while the speech synthesis portion is executed exclusively on a client which may be associated therewith. The client may comprise a hand-held device such as, for example, a cell phone, and the intermediate representation of the input text advantageously comprises at least a sequence of phonemes representative of the input text.

Abstract: A speaker verification method and apparatus which advantageously minimizes the constraints on the customer and simplifies the system architecture by using a speaker dependent, rather than a speaker independent, background model, thereby obtaining many of the advantages of using a background model in a speaker verification process without many of the disadvantages thereof. In particular, no training data (e.g. speech) from anyone other than the customer is required, no speaker independent models need to be produced, no a priori knowledge of acoustic rules are required, and, no multi-lingual phone models, dictionaries, or letter-to-sound rules are needed. Nonetheless, in accordance with an illustrative embodiment of the present invention, the customer is free to select any password phrase in any language.

Abstract: A method (and apparatus) for coding an audio signal, the method comprising the steps of partitioning the audio signal into a sequence of successive frames; calculating one or more noise thresholds for each of a plurality of frames in the sequence, each noise threshold for a particular one of the frames corresponding to a different perceptual coding quality for the particular frame; estimating a bit demand for each of a corresponding one or more perceptual coding qualities for each frame, wherein each estimated bit demand comprises a number of bits which would be used to code a given frame at the corresponding perceptual coding quality; selecting one of the perceptual coding qualities for the coding of a particular frame based upon the estimated bit demand for the perceptual coding quality for the particular frame, and further based on one or more bit demands estimated for one or more other frames; and coding the particular frame based on the noise threshold corresponding to the selected perceptual coding qual

Abstract: An automated method for the development of fixed-point algorithms which have been initially implemented as floating-point code which advantageously hides most of the tedious operations that need to be performed across the various stages of such a conversion procedure inside the definitions of a set of C++ classes. With the aid of these C++ class definitions, the fixed-point design process (i. e., the conversion from floating-point code to equivalent fixed-point code) is substantially simplified. Specifically, in accordance with the preferred embodiment of the present invention, a programmer need only to include and/or exclude certain previously defined header files, and to change the variable declarations within the floating-point code, in order to simulate the source codes across various stages of the conversion process.

Abstract: A customized method or algorithm for holding an interactive dialogue session between a (human) user and a machine (hereinafter referred to simply as a “dialogue”) is generated, such that the resulting dialogue advantageously responds to the user's requests and wherein the system's capability (i.e., the dialogue) is automatically modified thereafter based on dynamically changing external databases. Specifically, a computer system acts as a Dialogue Generator agent by creating such a customized dialogue consisting of services that are organized and presented in a form that is a combination of the user's expectations and the system's capabilities.

Abstract: Multiple, yet plausible, pronunciations of a proper name are generated based on one or more potential language origins of the name, and based further on the context in which the name is being spoken—namely, on characteristics of the population of potential speakers. Conventional techniques may be employed to identify likely candidates for the language origin of the name, and the characteristics of the speaker population on which the generation of the pronunciations is further based may comprise, for example, the national origin of the speakers, the purpose of the speech, the geographical location of the speakers, or the general level of sophistication of the speaker population.

Abstract: A generalized analysis-by-synthesis method and apparatus are disclosed. A plurality of trial original signals are generated based on an original signal for coding. The trial original signals are constrained to be perceptually similar to the original signal. Trial original signals are coded to produce one or more parameters representative thereof. Estimates of the trial original signals are synthesized from these parameters. Errors between the trial original signals and the synthesized estimates are determined. A coded representation of the original signal is determined which comprises parameters of the trial original signal having an associated error which satisfies an error evaluation process. Trial original signals may be generated by application of time-warps or time-shifts to the original signal. Coding of a trial original signal may be performed with conventional analysis-by-synthesis coding such as code-excited linear prediction coding (CELP).

Abstract: A tunable perceptual weighting filter is used in tandem codecs (coder/decoders). Specific filter parameters are advantageously tuned to provide improved performance in tandeming contexts. The parameters used are 10th order LPC (Linear Predictive Coding) predictor coefficients. The system employed uses Low-Delay Code Excited Linear Predictive codecs (LD-CELP).

Abstract: A clock pulse generator which has a signal controlled oscillator for producing output clock pulses at a repetition rate determined by the value of a control signal. Control means is operative in a calibration cycle to set the control signal to a low or high value and record the clock pulses counted in a period of predetermined duration, to set the control signal to a high or low value and record the clock pulses counted in a period of said predetermined duration, and to calculate rate of change data representing the rate of change of recorded clock pulses with reference to change in the value of the control signal.