Abstract: Flow-induced noise affecting microphones used in such applications as cellular phones is reduced by providing a deflector, made of an acoustically transmissive, high flow resistance material, disposed at a certain distance from the microphone pressure sensing area. The distance between the pressure sensing are and the deflector may be determined experimentally in consideration of any space constraints imposed by the specific application, and for cellular phones in particular, is approximately as little as 2 or 3 mm from the outside surface of the deflector.

Abstract: A voice command control and verification system and method stores for each authorized user, one or a series of speech models of voice commands or phrases uttered by the authorized user. Each speech model has an associated action component which specifies the specific action that the authorized user desires in response to the issuance of the corresponding voice command. Each user has a means of asserting his or her claimed identity to the system, preferably without an overt action such as the entry of digits. When an identity is asserted, and a voice command is thereafter spoken by a person, the system first matches a model of the voice command against the stored models for the user having the claimed identity.

Abstract: A speech compression system called "Transform Predictive Coding", or TPC, provides for encoding 7 kHz wideband speech (16 kHz sampling) at a target bit-rate range of 16 to 32 kb/s (1 to 2 bits/sample). The system uses short-term and long-term prediction to remove the redundancy in speech. A prediction residual is transformed and coded in the frequency domain to take advantage of knowledge in human auditory perception. The TPC coder uses only open-loop quantization and therefore has a fairly low complexity. The speech quality of TPC is essentially transparent at 32 kb/s, very good at 24 kb/s, and acceptable at 16 kb/s.

Abstract: A global positioning satellite (GPS) receiver is used in an electronic lap counter which maintains a lap count and which increments the lap count upon "observing" via the GPS receiver that the user, having gone away from an initial "start" location (e.g., an end of a pool or the start line of a closed loop running track), has returned to that location. A user-supplied indication, such as the pressing of a push button, serves as an indication to the lap counter that the lap counter's then current location is to be used as the start location.

Abstract: A method and an apparatus for performing stochastic matching of a set of input test speech data with a corresponding set of training speech data. In particular, a set of input test speech feature information, having been generated from an input test speech utterance, is transformed so that the stochastic characteristics thereof more closely match the stochastic characteristics of a corresponding set of training speech feature information. The corresponding set of training speech data may, for example, comprise training data which was generated from a speaker having the claimed identity of the speaker of the input test speech utterance. Specifically, in accordance with the present invention, a first covariance matrix representative of stochastic characteristics of input test speech feature information is generated based on the input test speech feature information.

Abstract: A low-complexity method and apparatus for performing signal decomposition in a low bit-rate WI speech encoder. A time-ordered sequence of sets of time-domain parameters is generated based on samples of a speech signal to be coded, each set of time-domain parameters corresponding to a waveform characterizing the speech signal. A cross correlation is then performed between two or more of said sets of time-domain parameters to produce a set of signals which represents relatively high rates of evolution of characterizing waveform shape across the time-ordered sequence of sets. Finally, the speech signal is coded based on the produced set of signals. A set of signals which represents relatively low rates of evolution of characterizing waveform shape across the time-ordered sequence of sets may also be produced.

Abstract: A low-complexity method and apparatus for performing waveform interpolation in a low bit-rate WI speech decoder, wherein interpolation between received waveforms is performed with use of spline coefficients generated based thereupon. Specifically, two signals are received from a WI encoder, each comprising a set of frequency domain parameters representing a speech signal segment of a corresponding pitch period. Then, spline coefficients are generated from each of the received signals, wherein each set of spline coefficients comprises a spline representation of a time domain transformation of the corresponding set of frequency domain parameters. Finally, the decoder interpolates between the spline representations to generate interpolated time domain data which is used to synthesize a reconstructed speech signal. In certain embodiments of the present invention, the time scale of at least one of the spline representations is modified to enable the interpolation therebetween.

Abstract: A speech coding system providing reconstructed voiced speech with a smoothly evolving pitch-cycle waveform. A speech signal is represented by isolating and coding prototype waveforms. Each prototype waveform is an exemplary pitch-cycle of voiced speech. A coded prototype waveform is transmitted at regular intervals to a receiver which synthesizes (or reconstructs) an estimate of the original speech segment based on the prototypes. The estimate of the original speech signal is provided by a prototype interpolation process which provides a smooth time-evolution of pitch-cycle waveforms in the reconstructed speech. Illustratively, a frame of original speech is coded by first filtering the frame with a linear predictive filter. Next a pitch-cycle of the filtered original is identified and extracted as a prototype waveform. The prototype waveform is then represented as a set of Fourier series (frequency domain) coefficients.

Abstract: A speech coding system robust to frame erasure (or packet loss) is described. Illustrative embodiments are directed to a modified version of CCITT standard G.728. In the event of frame erasure, vectors of an excitation signal are synthesized based on previously stored excitation signal vectors generated during non-erased frames. Specifically, the decoder generates and stores samples of a first excitation signal in a memory, and then, in response to a signal indicating a frame erasure, the decoder synthesizes a second excitation signal based on the previously stored samples. In particular, the second excitation is synthesized by correlating a first subset of the stored samples with a second subset thereof, identifying a set of stored excitation signal samples based on the correlation, and synthesizing the second excitation signal based on the identified samples. Finally, the decoder then filters the second excitation signal to synthesize a signal reflecting human speech.

Abstract: Two methods are disclosed for storing multimedia data that reduces the amount of disk I/O required by the system and cache misses experienced by the system. The first method determines the future access of each data buffer in a cache memory. Once the future of the data buffer is determined, the data buffer with the maximum future is allocated to store new blocks of data. The method approximates an optimal method of data buffer allocation, by calculating the future of a data buffer, relative to clients that will access the data buffers. The second method orders the clients based on the increasing distance of each client from the previous client; clients release the buffers in this order into a LIFO queue; if a buffer is selected to load a new block of data, the buffer at the head of the LIFO queue is selected.

Abstract: A method and apparatus which allows the transmission of the perceptually important features of a speech-coding parameter at a low bit rate. The speech coding parameter may, for example, comprise the signal power of the speech. The parameter is processed on a block by block basis. The parameter value at the block boundaries is transmitted by conventional methods such as, for example, by means of differential quantization. The shape of the reconstructed parameter contour within block boundaries is based on a classification. The classification determines perceptually important features of the parameter contour within a block. The classification can be performed either at the transmitting end of the coder (using, for example, the original parameter contour with high time resolution and possibly other speech parameters as well) or at the receiving end of the coder (using, for example, the transmitted parameter values, and possibly other transmitted speech parameters as well).

Abstract: A method and apparatus for performing adaptive hybrid automatic repeat request (ARQ) coding in which the coding is modified based on the acknowledgements returned by the receiver. In accordance with one illustrative embodiment of the present invention, the state of the channel is implicitly determined by the transmitter based upon the frequency of acknowledgments (ACKs and NACKs) arriving from the receiver. For example, since a NACK implies a weak received signal strength, the code rate of the FEC is advantageously reduced in response to such an acknowledgement. On the other hand, the code rate of the FEC is advantageously increased in response to an ACK. In accordance with another illustrative embodiment of the present invention, the acknowledgement returned by the receiver is modified to convey the number of errors in the corresponding received data packet.

Abstract: A method and apparatus for estimating individual impulse responses for a stereophonic communication system, such as a teleconferencing system, which involves selectively reducing the correlation between the individual channel signals of the stereophonic system. Selective reduction of stereophonic source signal correlation advantageously results in the estimation of individual impulse responses of a receiving room of the stereophonic communication system. The selectively reduced-correlation source signals are provided to conventional adaptive filters and the receiving room loudspeakers. Automatic echo cancellation is performed in a conventional fashion, but on the selectively reduced-correlation source signals. Specifically, selective reduction of source signal correlation between two stereophonic channels of a teleconferencing system is achieved by introducing a small non-linearity into each channel in order to reduce the interchannel coherence.

Abstract: A method for retrieving video data which has been striped across a plurality of disks using a coarse-grained striping technique. Specifically, and in accordance with an illustrative embodiment of the present invention, the method comprises scheduling the retrieval of a video in response to an incoming request and based on the availability of bandwidth on the disks, and then rescheduling the retrieval of that video to occur at an earlier time, the rescheduling based on a change (i.e., an increase) in the availability of bandwidth on the disks which results from the retrieval of another video being completed. The scheduling and rescheduling may, for example, comprise assigning a disk to the video, where the method further comprises incrementing the disk assigned to the video as each round occurs and beginning the retrieval of the given video when the disk assigned to it is the disk on which the data for the given video begins.

Abstract: A method and apparatus for performing joint stereo coding of multi-channel audio signals using intensity stereo coding techniques. In particular, predictive filtering techniques are applied to the spectral coefficient data, thereby preserving the time structure of the output signal of each channel, while maintaining the benefit of the high bit rate savings offered by intensity stereo coding. In one illustrative embodiment of the invention, the input signal is decomposed into spectral coefficients by a high-resolution filterbank/transform; the time-dependent masking threshold of the signal is estimated using a perceptual model; a filter performing linear prediction in frequency is applied at the filterbank outputs for each channel; intensity stereo coding techniques are applied for coding both residual signals into one carrier signal; the spectral values of the carrier signal are quantized and coded according to the precision corresponding to the masking threshold estimate; and all relevant information (i.e.

Abstract: A method for automatically converting a decision tree into one or more weighted finite-state transducers. Specifically, the method in accordance with an illustrative embodiment of the present invention processes one or more terminal (i.e., leaf) nodes of a given decision tree to generate one or more corresponding weighted rewrite rules. Then, these weighted rewrite rules are processed to generate weighted finite-state transducers corresponding to the one or more terminal nodes of the decision tree. In this manner, decision trees may be advantageously compiled into weighted finite-state transducers, and these transducers may then be used directly in various speech and natural language processing systems. The weighted rewrite rules employed herein comprise an extension of conventional rewrite rules, familiar to those skilled in the art.

Abstract: A method and apparatus for recognition of objects such as faces in images using signal compression techniques (i.e., coding techniques) in which a portion of the image which includes the object to be recognized (e.g., the face) is coded, and the resultant coded data is matched against previously coded and stored training data which makes up a known object database. A given object in an input image signal is matched to one of a plurality of known objects stored in a database, wherein the stored representation of each of the known objects comprises a codebook generated based on training image signals comprising the known object. A first illustrative embodiment comprises the steps of decomposing the given object into blocks; performing a plurality of encodings of the given object, each encoding comprising coding the object with use of one of the codebooks; determining a coding error for each encoding; and matching the given object to one of the known objects based on the coding errors.

Abstract: A key-phrase detection and verification method that can be advantageously used to realize understanding of flexible (i.e., unconstrained) speech. A "multiple pass" procedure is applied to a spoken utterance comprising a sequence of words (i.e., a "sentence"). First, a plurality of key-phrases are detected (i.e., recognized) based on a set of phrase sub-grammars which may, for example, be specific to the state of the dialogue. These key-phrases are then verified by assigning confidence measures thereto and comparing these confidence measures to a threshold, resulting in a set of verified key-phrase candidates. Next, the verified key-phrase candidates are connected into sentence hypotheses based upon the confidence measures and predetermined (e.g., task-specific) semantic information. And, finally, one or more of these sentence hypotheses are verified to produce a verified sentence hypothesis and, from that, a resultant understanding of the spoken utterance.

Abstract: A speech compression system called "Transform Predictive Coding", or TPC, provides for encoding 7 kHz wideband speech (16 kHz sampling) at a target bit-rate range of 16 to 32 kb/s (1 to 2 bits/sample). The system uses short-term and long-term prediction to remove the redundancy in speech. A prediction residual is transformed and coded in the frequency domain to take advantage of knowledge in human auditory perception. The TPC coder uses only open-loop quantization and therefore has a fairly low complexety.

Abstract: A method and apparatus for the perceptual coding of audio signals in which perceptual noise shaping is achieved in the time domain by performing a (linear) prediction (i.e., filtering) in the frequency domain. As a result, the temporal spread of quantization noise is reduced. Specifically, according to one illustrative embodiment of the present invention, a method comprises decomposition of the audio signal into a plurality of spectral component signals; generating a prediction signal representative of a prediction of one of said spectral component signals, said prediction based on one or more other ones of said spectral component signals; comparing the prediction signal with said one of said spectral component signals to generate a prediction error signal; coding said one of said spectral component signals based on the prediction error signal to generate a coded spectral component signal; and generating the encoded signal based on the coded spectral component signal.