Abstract: Techniques and tools for encoding enhancement layer video with quantization that varies spatially and/or between color channels are presented, along with corresponding decoding techniques and tools. For example, an encoding tool determines whether quantization varies spatially over a picture, and the tool also determines whether quantization varies between color channels in the picture. The tool signals quantization parameters for macroblocks in the picture in an encoded bit stream. In some implementations, to signal the quantization parameters, the tool predicts the quantization parameters, and the quantization parameters are signaled with reference to the predicted quantization parameters. A decoding tool receives the encoded bit stream, predicts the quantization parameters, and uses the signaled information to determine the quantization parameters for the macroblocks of the enhancement layer video. The decoding tool performs inverse quantization that can vary spatially and/or between color channels.

Abstract: Techniques and tools are described for scalable video encoding and decoding. In some embodiments, an encoding tool encodes base layer video and outputs encoded base layer video in a base layer bit stream. The encoding tool encodes inter-layer residual video (representing differences between input video and reconstructed base layer video) using motion compensation relative to previously reconstructed inter-layer residual video. For the inter-layer residual video, the encoding tool outputs motion information and motion-compensated prediction residuals in an enhancement layer bit stream. A decoding tool receives the base layer bit stream and enhancement layer bit stream, reconstructs base layer video, reconstructs inter-layer residual video, and combines the reconstructed base layer video and reconstructed inter-layer residual video.

Abstract: A method is described for efficiently determining total end-to-end distortion of a pre-compressed data stream, such as video streams or other media streams, at the time of delivery over a lossy-network, and for providing adaptive error-resilient delivery schemes based on distortion estimates. The methods can be utilized with single or multilayer packet streams and are particularly well suited for video streams. By way of example, distortion estimates are performed by generating side-information at the time of data stream compression, wherein the side-information is used in conjunction with information about the network status to determine an estimated distortion for the group of packets when the data stream is transported over the network to a destination end. This estimation may be utilized within described resiliency techniques in which the error correction mechanism is selected in response to the estimated distortion, which may be additionally refined in reference to cost factors.

Abstract: Various new and non-obvious apparatus and methods for using frame caching to improve packet loss recovery are disclosed. One of the disclosed embodiments is a method for using periodical and synchronized frame caching within an encoder and its corresponding decoder. When the decoder discovers packet loss, it informs the encoder which then generates a frame based on one of the shared frames stored at both the encoder and the decoder. When the decoder receives this generated frame it can decode it using its locally cached frame.

Abstract: In certain embodiments, to eliminate DC leakage into surrounding AC values, scaling stage within a photo overlap transform operator is modified such that the off-diagonal elements of the associated scaling matrix have the values of 0. In certain embodiments, the on-diagonal scaling matrix are given the values (0.5, 2). In some embodiments, the scaling is performed using a combination of reversible modulo arithmetic and lifting steps. In yet other embodiments, amount of DC leakage is estimated at the encoder, and preprocessing occurs to mitigate amount of leakage, with the bitstream signaling that preprocessing has occurred. A decoder may then read the signal and use the information to mitigate DC leakage.

Abstract: Techniques and tools are described for scalable video encoding and decoding. In some embodiments, an encoding tool encodes base layer video and outputs encoded base layer video in a base layer bit stream. The encoding tool encodes inter-layer residual video (representing differences between input video and reconstructed base layer video) using motion compensation relative to previously reconstructed inter-layer residual video. For the inter-layer residual video, the encoding tool outputs motion information and motion-compensated prediction residuals in an enhancement layer bit stream. A decoding tool receives the base layer bit stream and enhancement layer bit stream, reconstructs base layer video, reconstructs inter-layer residual video, and combines the reconstructed base layer video and reconstructed inter-layer residual video.

Abstract: Techniques and tools for signaling and using image tiling information (such as syntax elements relating index tables and header size), signaling and using windowing information (such as techniques for using windowing parameters when rotating, cropping or flipping images), and signaling and using alpha channel information are described.

Abstract: Various new and non-obvious apparatus and methods for using frame caching to improve packet loss recovery are disclosed. One of the disclosed embodiments is a method for using periodical and synchronized frame caching within an encoder and its corresponding decoder. When the decoder discovers packet loss, it informs the encoder which then generates a frame based on one of the shared frames stored at both the encoder and the decoder. When the decoder receives this generated frame it can decode it using its locally cached frame.

Abstract: Techniques and tools for signaling and using image tiling information (such as syntax elements relating index tables and header size), signaling and using windowing information (such as techniques for using windowing parameters when rotating, cropping or flipping images), and signaling and using alpha channel information are described.

Abstract: Embodiments for implementing a speech recognition system that includes a speech classifier ensemble are disclosed. In accordance with one embodiment, the speech recognition system includes a classifier ensemble to convert feature vectors that represent a speech vector into log probability sets. The classifier ensemble includes a plurality of classifiers. The speech recognition system includes a decoder ensemble to transform the log probability sets into output symbol sequences. The speech recognition system further includes a query component to retrieve one or more speech utterances from a speech database using the output symbol sequences.

Abstract: Techniques and tools for encoding and decoding data values that are hierarchically organized are presented. For example, an encoder encodes data as a set that has a hierarchy of subsets with set symbols. In the encoding, the encoder evaluates the data values of the set and selectively encodes a symbol combination code that indicates the set symbols of multiple subsets of the set. Then, for each of the multiple subsets considered as a new set, the encoder selectively repeats the evaluating, selective encoding and selective repetition for the new set. In corresponding decoding, a decoder decodes data encoded as a set that has a hierarchy of subsets with set symbols. In some implementations, the encoding and decoding are adaptive and use a symbol alphabet with nested elements.

Abstract: In certain embodiments, to eliminate DC leakage into surrounding AC values, scaling stage within a photo overlap transform operator is modified such that the off-diagonal elements of the associated scaling matrix have the values of 0. In certain embodiments, the on-diagonal scaling matrix are given the values (0.5, 2). In some embodiments, the scaling is performed using a combination of reversible modulo arithmetic and lifting steps. In yet other embodiments, amount of DC leakage is estimated at the encoder, and preprocessing occurs to mitigate amount of leakage, with the bitstream signaling that preprocessing has occurred. A decoder may then read the signal and use the information to mitigate DC leakage.

Abstract: An array of microphones placed on a mobile robot provides multiple channels of audio signals. A received set of audio signals is called an audio segment, which is divided into multiple frames. A phase analysis is performed on a frame of the signals from each pair of microphones. If both microphones are in an active state during the frame, a candidate angle is generated for each such pair of microphones. The result is a list of candidate angles for the frame. This list is processed to select a final candidate angle for the frame. The list of candidate angles is tracked over time to assist in the process of selecting the final candidate angle for an audio segment.

Abstract: Various new and non-obvious apparatus and methods for using frame caching to improve packet loss recovery are disclosed. One of the disclosed embodiments is a method for using periodical and synchronized frame caching within an encoder and its corresponding decoder. When the decoder discovers packet loss, it informs the encoder which then generates a frame based on one of the shared frames stored at both the encoder and the decoder. When the decoder receives this generated frame it can decode it using its locally cached frame.

Abstract: Techniques and tools are described for signaling hypothetical reference decoder parameters for video bitstreams, including signaling of buffer fullness. For example, a buffer size syntax element indicates a decoder buffer size, and a buffer fullness syntax element indicates a buffer fullness as a fraction of the decoder buffer size. As another example, buffer fullness is signaled in one or more entry point headers and other hypothetical reference decoder parameters are signaled in a sequence header.

Abstract: Techniques and tools for performing fading compensation in video processing applications are described. For example, during encoding, a video encoder performs fading compensation using fading parameters comprising a scaling parameter and a shifting parameter on one or more reference images. During decoding, a video decoder performs corresponding fading compensation on the one or more reference images.

Abstract: Techniques and tools for encoding and decoding a block of frequency coefficients are presented. An encoder selects a scan order from multiple available scan orders and then applies the selected scan order to a two-dimensional matrix of transform coefficients, grouping non-zero values of the frequency coefficients together in a one-dimensional string. The encoder entropy encodes the one-dimensional string of coefficient values according to a multi-level nested set representation. In decoding, a decoder entropy decodes the one-dimensional string of coefficient values from the multi-level nested set representation. The decoder selects the scan order from among multiple available scan orders and then reorders the coefficients back into a two-dimensional matrix using the selected scan order.

Abstract: Techniques and tools for performing fading compensation in video processing applications are described. For example, during encoding, a video encoder performs fading compensation using fading parameters comprising a scaling parameter and a shifting parameter on one or more reference images. During decoding, a video decoder performs corresponding fading compensation on the one or more reference images.

Abstract: Techniques and tools are described for compensating for rounding when estimating sample-domain distortion in the transform domain. For example, a video encoder estimates pixel-domain distortion in the transform domain for a block of transform coefficients after compensating for rounding in the DC coefficient of the block. In this way, the video encoder improves the accuracy of pixel-domain distortion estimation but retains the computational advantages of performing the estimation in the transform domain. Rounding compensation includes, for example, looking up an index (from a de-quantized transform coefficient) in a rounding offset table to determine a rounding offset, then adjusting the coefficient by the offset. Other techniques and tools described herein are directed to creating rounding offset tables and encoders that make encoding decisions after considering rounding effects that occur after an inverse frequency transform on de-quantized transform coefficient values.

Abstract: Techniques and tools for skip modes in encoding and decoding of inter-layer residual video are described. For example, an encoder encodes multiple macroblocks of a picture of inter-layer residual video. For a current macroblock that is skipped, the encoder selects a skip mode from among multiple available skip modes and uses the selected skip mode when encoding the current macroblock. The skip modes can include intra skip mode and predicted-motion skip mode. A corresponding decoder, for the current macroblock, selects and uses the skip mode for the current macroblock during decoding. As another example, an encoder encodes multiple channels of a picture of inter-layer residual video. For each channel, the encoder determines whether to skip the channel. The encoder signals channel skip information to indicate which channels are skipped. A corresponding decoder parses the channel skip information and determines on a channel-by-channel basis whether to skip the respective channels.