Dolby Laboratories

About: Dolby Laboratories is a company organization based out in Amsterdam, Netherlands. It is known for research contribution in the topics: Audio signal & Audio signal flow. The organization has 956 authors who have published 1726 publications receiving 29456 citations.

Frame compatible depth map delivery formats for stereoscopic and auto-stereoscopic displays

Abstract: Stereoscopic video data and corresponding depth map data for stereoscopic and auto-stereoscopic displays are coded using a coded base layer and one or more coded enhancement layers. Given a 3D input picture and corresponding input depth map data, a side-by-side and a top-and-bottom picture are generated based on the input picture. Using an encoder, the side-by-side picture is coded to generate a coded base layer Using the encoder and a texture reference processing unit (RPU), the top-and-bottom picture is encoded to generate a first enhancement layer, wherein the first enhancement layer is coded based on the base layer stream, and using the encoder and a depth-map RPU, depth data for the side-by-side picture are encoded to generate a second enhancement layer, wherein the second enhancement layer is coded based on to the base layer. Alternative single, dual, and multi-layer depth map delivery systems are also presented.

Detecting Conference Call Performance Issue from Aberrant Behavior

Abstract: In a conference call having a plurality of participants interacting in a conference exchange of information in a digital transmission environment, the interaction being across a variable network transmission resource, a method of allocating the level of transmission resource, the methods including the steps of: (a) monitoring predetermined aspects of the participant's behavior during the conference call; (b) determining a divergence of participants behavior from normative values; (c) utilising any divergence as an indicator of aberrant operation of the participants; and (d) allocating the resource determinative on the divergence of participants behavior from normative values.

An Empirical Study of Conv-Tasnet

Abstract: Conv-TasNet is a recently proposed waveform-based deep neural network that achieves state-of-the-art performance in speech source separation. Its architecture consists of a learnable encoder/decoder and a separator that operates on top of this learned space. Various improvements have been proposed to Conv-TasNet. However, they mostly focus on the separator, leaving its encoder/decoder as a (shallow) linear operator. In this paper, we conduct an empirical study of Conv-TasNet and propose an enhancement to the encoder/decoder that is based on a (deep) non-linear variant of it. In addition, we experiment with the larger and more diverse LibriTTS dataset and investigate the generalization capabilities of the studied models when trained on a much larger dataset. We propose cross-dataset evaluation that includes assessing separations from the WSJ0-2mix, LibriTTS and VCTK databases. Our results show that enhancements to the encoder/decoder can improve average SI-SNR performance by more than 1 dB. Furthermore, we offer insights into the generalization capabilities of Conv-TasNet and the potential value of improvements to the encoder/decoder.

Spectral Synthesis for Image Capture Device Processing

Abstract: A substantially rectangular spectral representation is synthesized, which is adapted to produce image capture device sensor outputs if applied to an image capture device. The synthesized substantially rectangular spectral representation can be utilized in generating output color values of an output color space from image capture device sensor outputs, where the image capture device sensor outputs correspond to an image captured by an image capture device. The generated output color values correspond to colors perceived by the human visual system for the same image as that captured by the image capture device. Image capture device gamut is also determined.

Multi-Channel Audio Content Analysis Based Upmix Detection

Abstract: Forensic audio upmixer detection is described. Feature sets are extracted from an audio signal that has two or more individual channels. Based on the extracted feature sets, it is determined whether the audio signal was upmixed from audio content that has fewer channels.