Amazon Technologies, Inc.

Amarsingh Buckthasingh Winston

Lei Li

Yongjun Wu

Stefan Christian Richter

Nicholas James Benson

Marc Joliveau

Hai Wei

Mushegh Malkhasyan

Li Tong

Avisar Ten-Ami

Samuel James Van Pelt

Charles Benjamin Franklin Waggoner

Michael Flynn

James Marvin Freeman, Ii

Rahul Urgaonkar

Lars Christian Ulness

Girish Bansilal Bajaj

Andrew J Watts

Kenneth Thomas Burford

Deepthi Nandakumar

Yang Yang

Euan McLeod

Joshua Danovitz

Satheesh Ramalingam

Mohan Padmanabhan

Rakesh Tangirala

Manish Rao

Hargun Kaur

Umut Utkan

Adhvik Arun Shetty

Moping Dou

Khawaja Salman Shams

Abhishek Kumar

Gaurav Saxena

Andrew Stuart Glass

Charles Benjamin Waggoner

Matthew Brinkley

Mairo Pedrini

Srikanth Kiran Kotagiri

Kyle Bradley Koceski

Matthew Scharr

David Wayne Bourgoyne

Christopher Steven Mark Pringle

Michael Robert Starr

Techniques for combining dynamic metadata for multiple view (multiview) video displaying are described. According to some examples, a computer-implemented method includes receiving, by a client device, a first video comprising a first set of dynamic metadata including a corresponding first color value histogram for a plurality of frames of the first video, and a second video comprising a second set of dynamic metadata including a corresponding second color value histogram for a plurality of frames of the second video; combining the first video and the second video into a combined video; generating a first combined color value histogram for a first frame of the combined video based on a first color value histogram of the first set of dynamic metadata for a first frame of the first video and a second color value histogram of the second set of dynamic metadata for a first frame of the second video, and a second different combined color value histogram for a second frame of the combined video based on a third color value histogram of the first set of dynamic metadata for a second frame of the first video and a fourth color value histogram of the second set of dynamic metadata for a second frame of the second video; performing, by a tone mapper of the client device, a tone mapping of the combined video based on the first combined color value histogram and the second different combined color value histogram to generate an output video with multiple views; and sending the output video to a display or storage.

Computer-implemented methods for combining dynamic metadata for multiple view (multiview) video displaying

Techniques are described for fulfilling an utterance request for an item represented within a video rendered at a client device. In some implementations, a user account associated with the request is identified, enabling a video stream transmitted in association with the user account at the time that the request was uttered to be identified. In one technique, a timestamp associated with the request is used to identify the relevant portion of the video stream. The item represented within the portion of the video stream can be identified using various techniques and/or information such as image recognition, metadata within the video, subtitles, closed captions, and/or a database mapping between the item and a video content item transmitted in the video stream.

Utterance request of items as seen within video

Systems, methods, and computer-readable media are disclosed for optimizing network performance on a computer device to improve quality of experience by determining which network settings on the computing device to adjust. A clustering algorithm may identify various classes of networks and a classification algorithm may determine a network class specific to a network on a computing device. The effects of certain network settings for that networks class may be determined and the network setting and/or settings that optimizes the network performance may be promoted. The system may periodically analyze network data to recalculate the appropriate networks class and may determine different network settings based on the recalculation, facilitating mid-session improvements to the quality of experience.

Machine learning-based playback optimization using network-wide heuristics

Techniques are described for optimizing video decoder operations.

Video decoder memory optimization

Systems and methods for optimizing network settings

Techniques are described for adjusting parameter settings for bitrate selection algorithms for different segments of a population of devices streaming content. Streaming sessions are identified according to session characteristics. Within each segment of sessions, control parameter settings are sent to devices corresponding to a subset of each segment. Test parameter settings are sent to devices corresponding to another subset of each segment. If the test parameter settings result in better playback performance relative to the control parameter settings, the test parameter settings become the new control parameter settings, and new test parameter settings are generated.

Adjusting parameter settings for bitrate selection algorithms

Techniques are described for providing metadata corresponding to primary content of live media streams that include secondary content inserted into the streams.

Providing metadata for live media streams

Techniques are described for server-side fragment insertion and delivery of media content. After receiving a request to play back media content, a manifest is generated that includes fragments associated with the requested media content and fragments associated with different media content. A client device requests fragments from a media server according to the manifest, and the media server provides the correct fragment in response to the request.

Server-side fragment insertion and delivery

Techniques are described for optimizing event-adaptive live video encoding profiles.

Content delivery of live streams with event-adaptive encoding

Techniques are described for encoding information in manifest data in a way that enables identification of a version of media content. Such a capability may be used to identify dynamic content inserted into the media content for the purpose of tracking the effect of the dynamic content on user generated feedback. This capability might also be used to identify a user or device associated with a particular copy of the media content for anti-piracy purposes.

Encoding identifiers into customized manifest data

Encoding of a video file includes determining a plurality of scenes associated with a video file, and determining at least one group of pictures (GOP). Starting sequentially from a beginning frame of the video file, the system identifies a first GOP having a first encoding error characteristic. The system changes a bitrate allocation setting from a first setting to a second setting based on the encoding error characteristic. The system identifies a second frame having a second encoding error characteristic, and changes a second bitrate allocation setting from the second setting to a third setting based on the second encoding error characteristic. The system generates an encoded video file that includes an encoded plurality of scenes.

Adaptive group of pictures (GOP) encoding

Techniques are described for optimizing streaming video encoding profiles.

Optimizing streaming video encoding profiles

Techniques are described for adjusting parameter settings for bitrate selection algorithms for devices streaming media content. Control parameter settings are selected for playback of first media content. Test parameter settings are selected for playback of second media content. If the test parameter settings result in better playback performance relative to the control parameter settings, the test parameter settings become the new control parameter settings.

Techniques for delivering live streaming content are described by which a streaming heuristic is selected based on the delay between the live and client playheads. For live content for which the delay is short, a low latency streaming heuristic is selected which is designed to handle live content streams for which only a few seconds of content can be buffered. For live content for which the delay is longer, a higher latency streaming heuristic is selected which is designed to handle streams for which the delay is sufficient to build a buffer of content fragments.

Heuristics for streaming live content

Growing community of inventors

Seattle, WA, United States of America

Amarsingh Buckthasingh Winston