Abstract: A method and apparatus for video coding whereby a region of an image which includes a predetermined object such as a person's face in the foreground portion of the image is automatically determined. Specifically, the foreground portion of the image is identified, and one or more predetermined (geometric) shapes (e.g., ellipses) are compared with the shapes of objects found in the foreground portion of the image. The foreground portion of an image may be determined by performing a global motion estimation of the overall image to detect global image movement resulting, for example, from camera pan and zoom. That portion of the image whose movement is consistent with the estimated global motion may be identified as the background portion, with the remainder of the image identified as the foreground portion. The identified region of the image which includes the predetermined object and portions of the image which do not include the predetermined object may be coded with differing levels of coding accuracy (e.g.

Abstract: An acoustic signal receiving apparatus comprising a housing having an acoustically reflecting surface and a directional acoustic sensor unit having first-order gradient characteristics. The sensor unit is coupled to the housing with use of a retractable member having a retracted position and an extended position. When the member is extended, the sensor unit is positioned relative to the reflecting surface such that the acoustic interaction between the sensor unit and the reflecting surface causes the output of the sensor unit to have second-order gradient response characteristics. In accordance with one illustrative embodiment, in a notebook computer with a "flip-back" lid, the sensor element automatically extends to the desired position when the lid is opened, and automatically retracts to be flush with the housing surface when the lid is closed.

Abstract: A method and apparatus for low bit rate coding of color video signals in which the use of the codebook entry identified by the codebook index may be bypassed in the coding of chrominance information when it is determined that it is advantageous to do so. In particular and in accordance with an illustrative embodiment of the present invention, the method comprises the steps of generating a codebook index signal based on the luminance information of the block being coded; determining a coding error based on a comparison of the chrominance information contained in the given block and the chrominance vector represented by the generated codebook index signal; and generating a chrominance vector bypass signal based on whether the coding error exceeds a threshold. Specifically, the chrominance vector bypass signal is used to indicate whether the chrominance information of the given block has or has not been coded with use of the codebook.

Abstract: A highly efficient, low delay pitch parameter derivation and quantization permits overall delay which is a fraction of prior coding delays for equivalent speech quality at low bitrates. In distinguishing between pitch period information for voiced and non-voiced frames of input signals, non-voiced frames are assigned a non-zero "bias" value, while voiced frames have associated with them generated pitch information based on an analysis of signals in a present frame and comparison with signals relating to the pitch in a prior frame. Transitions from non-voiced to voiced input frames are efficiently accomplished using a non-uniform quantization method based on an analysis of a sequence of frames. Typical uses include low delay, low-bitrate coders such as Code Excited Linear Prediction (CELP).

Abstract: A method and apparatus for coding an image comprising (at least) two regions (e.g., a facial region and a background region) which are to be coded with different quantization error parameters (e.g., quantizer step sizes), wherein the respective quantization error parameters are determined with use of a feed-forward rate control technique. Initial quantization error parameters are assigned to each region, and the overall bit rate which would result for the coding of the given frame is computed based on these assigned quantization error parameters. The computed bit rate is compared to a given bit rate constraint which may, for example, represent a limitation of a transmission channel. Based on the result of this comparison, one or both of the assigned quantization error parameters are adjusted, thereby resulting in different quantization error parameters for the two regions.

Abstract: A method identifies and models the distortion in a document using a two-dimensional nonlinear filter. The two-dimensional nonlinear filter is then used to improve the performance of optical character recognition systems in recognizing characters in documents that have been degraded, as for example by repeated photocopy reproductions or by fax transmission.

Abstract: The invention relates to a method and apparatus for speech recognition, the speech to be recognized including one or more words. Recognition is based on an analysis of a first and a second utterance. In accordance with the invention, the first utterance is compared to one or more models of speech to determine a similarity metric for each such comparison. The model of speech which most closely matches the first utterance is determined based on the one or more similarity metrics. The similarity metric corresponding to the most closely matching model of speech is analyzed to determine whether the similarity metric satisfies a first recognition criterion. The second utterance is compared to one or more models of speech associated with the most closely matching model (which may include the most closely matching model) to determine a second utterance similarity metric for each such comparison.

Abstract: A speech recognition system may be trained with data that is independent from previous acoustics. This method of training is quicker and more cost effective than previous training methods. In training the system, after a vocabulary word is input into the system, a first set of phonemes representative of the vocabulary word is determined. Next, the first set of phonemes is compared with a second set of phonemes representative of a second vocabulary word. The first vocabulary word and the second vocabulary word are different. The comparison generates a confusability index. The confusability index for the second word is a measure of the likelihood that the second word will be mistaken as another vocabulary word, e.g., the first word, already in the system. This process may be repeated for each newly desired vocabulary word.

Abstract: A CELP speech decoder includes a first portion comprising an adaptive codebook and a second portion comprising a fixed codebook. The CS-ACELP decoder generates a speech excitation signal selectively based on output signals from said first and second portions when said decoder fails to receive reliably at least a portion of a current frame of compressed speech information. The decoder does this by classifying the speech signal to be generated as periodic (voiced) or non-periodic (unvoiced) and then generating an excitation signal based on this classification. If the speech signal is classified as periodic, the excitation signal is generated based on the output signal from the first portion and not on the output signal from the second portion. If the speech signal is classified as non-periodic, the excitation signal is generated based on the output signal from said second portion and not on the output signal from said first portion.

Abstract: A method for retrieving video data from a video server, the video data having been stored on a plurality of disks based on a disk striping technique. In accordance with one illustrative embodiment, the method comprises the steps of retrieving a predetermined number of bits from the plurality of disks in the video server, and storing that predetermined number of bits in a buffer memory, wherein the number of bits retrieved and stored is based on the number of disks and on the capacity of the buffer memory. These steps, which together may illustratively constitute one round of the video retrieval process, may be repeated until the entire video has been retrieved and, for example, transmitted to the intended recipient(s) at a required transmission rate.

Abstract: A dynamic combined power control and forward error correction control (FEC) technique for mobile radio systems which illustratively decreases the power consumed by wireless transmitters and increases the number of simultaneous connections which may be supported thereby. Individual transmitter-receiver pairs may adaptively determine the minimal power and FEC required to satisfy specified quality-of-service (QOS) constraints. Specifically, in accordance with an illustrative embodiment of the present invention, a first portion of a signal is encoded with a first code to generate a first encoded signal portion. Then, the first encoded signal portion is transmitted with a first power level. Next, parameter data is received, wherein the parameter data is representative of one or more characteristics of a received signal portion having been received by the receiver, the received signal portion having been based on the transmitted first encoded signal portion.

Abstract: An improved long-term predictor (LTP) for use in analysis-by-synthesis coding systems, such as CELP is disclosed. The invention provides control of the periodicity of speech signals generated by the LTP. This control facilitates a reduction in perceptible noise/buzziness in reconstructed speech. An embodiment of the invention includes a conventional LTP in combination with a two-tap finite impulse response filter. The filter augments operation of the LTP by generating precursor signals of LTP output signals. These precursor signals are combined with the LTP output signals to form the output of the improved LTP.

Abstract: Synergy between operations performed by a speech-rate modification system and those operations performed in a speech coding system is exploited to provide a speech-rate modification system with reduced hardware requirements. The speech rate of an input signal is modified based on a signal representing a predetermined change in speech rate. The modified speech-rate signal is then filtered to generate a speech signal having increased short-term correlation. Modification of the input speech signal may be performed by inserting in the input speech signal a previous sequence of samples corresponding substantially to a pitch cycle. Alternatively, the input speech signal may be modified by removing from the input speech signal a sequence of samples corresponding substantially to a pitch cycle.

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. This synthesis differs for voiced and non-voiced speech. During erased frames, linear prediction filter coefficients are synthesized as a weighted extrapolation of a set of linear prediction filter coefficients determined during non-erased frames. The weighting factor is a number less than 1. This weighting accomplishes a bandwidth-expansion of peaks in the frequency response of a linear predictive filter. Computational complexity during erased frames is reduced through the elimination of certain computations needed during non-erased frames only.

Abstract: A speech compression system called "Transform Predictive Coding", or TPC, provides for encoding 7 kHz wideband speech (160 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: In a speech decoder which experiences frame erasure, the pitch delay associated with the first of consecutive erased frames is incremented. The incremented value is used as the pitch delay for the second of consecutive erased frames. Pitch delay associated with the first of consecutive erased frames may correspond to the last correctly received pitch delay information from a speech encoder (associated with a non-erased frame), or it may itself be the result of an increment added to a still previous value of pitch delay (associated with a still previous erased frame).

Abstract: The invention is used to shape noise in time domain and frequency domain coding schemes. The method advantageously uses a noise weighting filter based on a filterbank with variable gains. A method is presented for decoding an encoded signal based on received side information and on a masking matrix derived from the masking properties of speech. In particular, the encoded signal is separated into subband signal components, each of which is multiplied by a corresponding gain value based on the masking matrix. These multiplied subband signal components are then combined to produce a decoded signal.

Abstract: An adaptive postfilter is used on the decoding side of tandem codecs (coder/decoders). Post-filter parameters are adapted using a backward synthesis filter. 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: An image-processing system for perceptual coding of an image is disclosed. Coding is accomplished through an analysis of human visual sensitivity to noise in halftone images and an analysis of one or more signals representing the image to be coded. These analyses determine levels of noise. A first image is encoded so as to produce encoded values without introducing noise which exceeds a determined level of noise. Analysis of human visual sensitivity is carried out under a set of user determined conditions comprising viewing distance and lighting. The encoded image is communicated and decoded to produce representation of the first image for display. As part of displaying the representation of the first image, a halftoning process is carried out. The halftoning process may comprise the interpolation of data values of the representation of the first image as well as the addition of micro-dither to the representation.

Abstract: Codebook vectors may be considered critical if they give poor energy approximations and exhibit a particular shape with smaller components near the beginning and larger components toward the end of the vector. Standard deviation may be used to identify critical codevectors based on energy approximation error measured in decibels. A low-bit rate (typically 8 kbit/s or less), low-delay digital coder and decoder based on Code Excited Linear Prediction for speech and similar signals features backward adaptive adjustment for codebook gain and short-term synthesis filter parameters and forward adaptive adjustment of long-term (pitch) synthesis filter parameters. In addition, the coder makes use of an excitation codebook and the coding is based on a set of codebook vector energies for a set of codebook vectors in the codebook. The codebook energies are calculated by identifying a set of approximations for the non-critical codebook vector energies.