Abstract: A maximum likelihood (ML) linear regression (LR) solution to environment normalization is provided where the environment is modeled as a hidden (non-observable) variable. By application of an expectation maximization algorithm and extension of Baum-Welch forward and backward variables (Steps 23a-23d) a source normalization is achieved such that it is not necessary to label a database in terms of environment such as speaker identity, channel, microphone and noise type.

Abstract: A multimodal code-excited linear prediction (CELP) speech coder determines a pitch-lag-periodicity-independent peakiness measure from the input speech. If the measure is greater than a peakiness threshold the encoder classifies the speech in a first coding mode. In one embodiment only frames having an open-loop pitch prediction gain not greater than a threshold, a zero-crossing rate not less than a threshold, and a peakiness measure not greater than the peakiness threshold will be classified as unvoiced speech. Accordingly, the beginning or end of a voiced utterance will be properly coded as voiced speech and speech quality improved. In another embodiment, gain-match scaling matches coded speech energy to input speech energy. A target vector (the portion of input speech with any effects of previous signals removed) is approximated using the precomputed gain for excitation vectors while minimizing perceptually-weighted error.

Abstract: A communications device, such as a cellular telephone handset (10), and a method of operating the same to suppress noise in audio information such as speech, is presented. The handset (10) includes a digital signal processor (DSP) (30) having program memory (31) for controlling the DSP (30) to apply a hierarchical lapped transform to the input digital sequence. The hierarchical lapped transform decomposes the input sequence into coefficients representative of plurality of sub-bands corresponding to critical bands of the human ear. Each coefficient is modified by a noise suppression filter operator, based upon a ratio of an estimate of the noise power to an estimate of the signal power in the corresponding sub-band; clamping of changes in the noise power estimate over time, and use of a decaying signal envelope estimate, eliminate distortion in the processed signal. Musical noise is eliminated by using a minimum gain value in each sub-band.

Abstract: A camera has a user-operable input device, e.g., a writing pad, attached to its body. A user inputs information to be superimposed on a photographic negative onto the user-operable input device. A processor is connected to the user-operable input device. A display is also attached to the camera body and is connected to the processor. The processor transmits the information entered onto the user-operable input device to the display. When the negative is exposed, the display information is superimposed with the image of the subject being photographed. Thus, a user may enter a caption describing the subject directly onto the photograph.

Abstract: A method of forming an image scaling filter for converting a first number of input lines to a second number of output lines comprises the steps of determining an optimal frequency response without sharp cutoffs for a given scaling factor and for a determined number of taps per line 102. Filter coefficients are provided based on the determined filter taps per line and the determined optimal frequency response 103. The coefficients are grouped into sets 104 corresponding to the taps per output line and the coefficients are rescaled so the sum of the set equals one 105. The input lines are multiplied by the rescaled coefficients and summed 106 to achieve the output line values.

Abstract: A new method for quantization of the LPC coefficients in a speech coder includes an improved form of switched predictive multi-stage vector quantization. The switch predictive quantization includes at least a pair of codebook sets in a MSVQ quantizer and a first and second prediction matrix 24a and 24b with the first prediction matrix 1 used with codebook set 1 and prediction matrix 2 used with codebook set 2 and the encoder determines which prediction matrix/codebooks set produces the minimum quantization error at detector 35 and control 29 gates the indices with the minimum error out of the speech coder.

Abstract: A synthesis of percussion musical instruments sounds is provided using a microprocessor (17) that implements an all pole lattice filter and applying either a single impulse signal to the filter or N samples of an excitation signal sequence to the filter by a memory (19). The coefficients of the filter are determined by storing digital samples (501) of desired musical note from a desired percussion instrument, generating a Fourier transform to get a spectrum (502), picking the peaks of the spectrum (503) to select the most prominent components in the spectrum and determining wanted frequencies for decaying sine waves and for the frequencies finding the time envelope and estimating therefrom the pole radius.

Abstract: A method for establishing a correspondence between chemical use and the process where use occurs. The process is assigned a process identification number (PIN). After chemicals have been assigned into an inventory, the chemicals are assigned to the process and a report is generated reporting the chemical use corresponding to the process.

Abstract: A point-to-multipoint or two-way communications system is provided by a nodal transmitter located in a node with a plurality of nodal antennas radiating different polarization signals about the node. The system includes subscriber stations with directional antennas adapted to receive signals radiated from the nodal transmitter. The system may additionally include capability for transmitting and radiating subscriber signals to the nodal transmitter location for two-way communications.

Abstract: A method of quantizing harmonic amplitudes (FIG. 3), used in a speech encoder (10). The method compares variable dimension input vectors to fixed dimension codebook vectors, by first sampling each codebook vector so that it is converted to a vector having the same dimension as the input vector (FIG. 3, step 35). The resulting codebook vector is compared to the input vector (step 37). The difference (error) is weighted in favor of low frequency harmonics. Also, the weighting favors formant amplitudes so that they are quantized more accurately than formant nulls (FIG. 3, step 38; FIG. 5).

Abstract: A system and method for detecting anomalies on a manufactured device includes or utilizes a moveable stage for holding and positioning the device, a camera for capturing an image of the device on the stage, a digitizer coupled to the camera for producing a digital-pixel-based representation of the image, and a computer having a processor and memory. The computer is coupled to the digitizer for receiving the digital-pixel-based representation from the digitizer and coupled to the stage for selectively moving the stage to align the device. The computer is programmed to be operable to symbolically decompose a digital-pixel-based representation of an image to create a primitive-based representation of the image. The image may be aligned by the computer with respect to the rotation of geometric objects in the image by developing a histogram of angles and lengths and matching them to determine a rotational shift.

Abstract: This application describes the Interactive PP Assembly Language Editor (IPPALE), a software tool intended to help programmers write parallel PP instructions quickly and effectively. IPPALE consists of an editor that (on command from the user) extracts the current assembly language instruction, assembles it, and displays a graphical representation of how the instruction uses the resources of the processor. This allows the programmer to see immediately whether the instruction is legal, and also whether there are idle resources that could potentially still be used. The result is that programmers can experiment with a whole set of parallel instructions without ever invoking the PP assembler. We expect this to lead to faster learning, reduced programmer frustration, and improved overall productivity.

Abstract: A method of using a processor-based integrated circuit (IC) tester (12) to automatically invoke diagnostic testing. A diagnostic schedule (30) is user-defined (FIG. 3), and checked by the tester (12) at various times, such as before and after the tester 12 tests an IC lot (FIGS. 6, 7, 8). The schedule (30) has various scheduling parameters that permit diagnostic testing to be scheduled to occur at certain times, at the expiration of intervals, or upon certain events. The schedule (30) also permits different levels of diagnostic testing, each level with its own schedule. The tester (12) is programmed to run an appropriate level of diagnostic testing if more than one level is due to occur.

Abstract: Method and system of determining an out-of-vocabulary score for speaker-independent recognition of user-defined phrases comprises enrolling a user-defined phrase with a set of speaker-independent (SI) recognition models using an enrollment grammar. An enrollment grammar score of the spoken phrase may be determined by comparing features of the spoken phrase to the SI recognition models using the enrollment grammar. The enrollment grammar score may be penalized to generate an out-of-vocabulary score.

Abstract: Object detection for scene change analysis is performed by a statistical test applied to data extracted from two images taken from the same scene from identical viewpoints. It is assumed that a single change region corresponding to an object that is present in one image but absence in the other is given. In the case of TV data, the test consists of measuring the coincidence of edge pixels in each image with the boundary of the change region. In the case of IR data, the tests consist of measuring the pixel intensity variance within the change region in each image.

Abstract: The method uses a three-dimensional face model and a technique called eigenface decomposition to analyze the video at one end. The facial feature locations and eigenface coding of the face image are sent to a decoder. The decoder synthesizes the face image at the receiving end. Eigenface decoding is used to texture map a three-dimensional model warped by detected feature locations.

Abstract: A method for determining the position of a die on a wafer table after a wafer (20) has been cut includes stepping the wafer table (22) a series of one die lengths diagonally and after each step compare the die pattern to a reference die to determine street widths between dies. The widths are then averaged. This averaged street width is added to the die length to at the computer (26) to determine the jump distance between good dies. The distance from the reference die is continuously updated and this is used with the average street width to compute the next die position to jump. If the jump to the next die is unsuccessful, jump back to mid-point and if that is unsuccessful, reiterate mid-point jumps until successful.

Abstract: An apparatus and method for automatic knowledge-based object or anomaly classification is provided by capturing a pixel map of an image and from that generating high level descriptors of the object or anomaly such as size, shape, color and sharpness. These descriptors are compared with sets of descriptors in a knowledge-base to classify the object or anomaly.

Abstract: A synthesizer may synthesize speech by receiving an adaptive codebook excitation signal and an adaptive codebook gain. The adaptive codebook excitation signal may be scaled using the adaptive codebook gain to generate a scaled adaptive codebook excitation signal. A fixed excitation signal and a fixed excitation gain may also be received. The fixed excitation signal may be scaled using the fixed excitation gain to generate a scaled fixed excitation signal. The scaled adaptive codebook excitation signal and the scaled fixed excitation signal may be combined to generate the excitation signal having a first word length. An overall gain signal of the excitation signal may also be received. A scaled excitation signal may then be generated by scaling the excitation signal using the overall gain signal. The scaled excitation signal may have a second word length greater than the first word length.

Abstract: The method and system of adapting speech recognition models to a speaker environment may comprise receiving a spoken password (52) and getting a set of speaker independent (SI) speech recognition models (54). A mapping sequence may be determined for the spoken password (56). Using the mapping sequence, a speaker ID may be identified (58). A transform may be determined (66) between the SI speech recognition models and the spoken password using the mapping sequence. Speaker adapted (SA) speech recognition models may be generated (68) by applying the transform to SI speech recognition models. A speech input may be recognized (70) by applying the SA speech recognition models.