Abstract: Systems and methods to implement customer contact service with real-time supervisor assistance. A supervisor may oversee multiple agents in a customer contact service. A service of a computing resource service provider may monitor a plurality of audio connections at a service of a computing resource service provider, generate transcripts for the plurality of audio data, analyze the transcripts using a set of natural language processing (NLP) techniques to generate metadata, tag the transcripts with categories based at least in prat on the metadata, generate information for at least a portion of the plurality of connections based on the transcripts, metadata, and categories, and provide the information to a supervisor of the agents.

Abstract: Systems and methods to analyze contacts data. Contacts data may be encoded as text (e.g., chat logs), audio (e.g., audio recordings), and various other modalities. A computing resource service provider may implement a service to obtain audio data from a client, transcribe the audio data, thereby generating text, execute one or more natural language processing techniques to generate metadata associated with the text, processing at least the metadata to generate an output, determine whether the output matches one or more categories, and provide the output to the client. Techniques described herein may be performed as an asynchronous workflow.

Abstract: Three dimensional [3D] image data and auxiliary graphical data are combined for rendering on a 3D display (30) by detecting depth values occurring in the 3D image data, and setting auxiliary depth values for the auxiliary graphical data (31) adaptively in dependence of the detected depth values. The 3D image data and the auxiliary graphical data at the auxiliary depth value are combined based on the depth values of the 3D image data. First an area of attention (32) in the 3D image data is detected. A depth pattern for the area of attention is determined, and the auxiliary depth values are set in dependence of the depth pattern.

Abstract: Systems and methods to implement customer contact service with real-time supervisor assistance. A supervisor may oversee multiple agents in a customer contact service. A service of a computing resource service provider may monitor a plurality of audio connections at a service of a computing resource service provider, generate transcripts for the plurality of audio data, analyze the transcripts using a set of natural language processing (NLP) techniques to generate metadata, tag the transcripts with categories based at least in prat on the metadata, generate information for at least a portion of the plurality of connections based on the transcripts, metadata, and categories, and provide the information to a supervisor of the agents.

Abstract: Systems and methods which may be implemented in the context of a customer contact service. A service of a computing resource service provider may obtain, at a first service of a computing resource service provider, audio source data from a client of the computing resource service provider, generate an output from the audio data, wherein the output encodes: a transcript of the audio data generated by a second service, wherein the transcript is partitioned by speaker, metadata generated by a third service based at least in part on the transcript, and, one or more categories triggered by the transcript, wherein a fourth service is used to determine whether the one or more categories match the transcript, and providing the output to the client.

Abstract: Systems and methods to analyze contacts data. Contacts data may be encoded as text (e.g., chat logs), audio (e.g., audio recordings), and various other modalities. A computing resource service provider may implement a service to obtain audio data from a client, transcribe the audio data, thereby generating text, execute one or more natural language processing techniques to generate metadata associated with the text, processing at least the metadata to generate an output, determine whether the output matches one or more categories, and provide the output to the client. Techniques described herein may be performed as an asynchronous workflow.

Abstract: Systems and methods to implement customer contact service with real-time agent assistance. A service of a computing resource service provider may establish a connection between an agent to obtain audio data of the agent and a customer, transcribe the audio data to generate at least a portion of a transcript, execute one or more natural language processing techniques to generate metadata associated with the transcript, determine, based at least in part on the metadata, whether one or more categories match the transcript, generate information by processing the transcript, the metadata, and the one or more categories, and provide, to the agent, a notification that encodes the information. Notifications may be provided in real-time, which may include suitable delay tolerances.

Abstract: Systems and methods to support searching and diagnostics capabilities in a customer contact service. Contacts data may be encoded as text (e.g., chat logs), audio (e.g., audio recordings), and various other modalities. A computing resource service provider may implement a service to receive, from a client, a request to perform a search with a specified set of parameters on a plurality of outputs associated with a plurality of contacts data, wherein the plurality of outputs are generated based at least in part by: transcripts generated based on audio data of the plurality of contacts data, metadata determined based the transcripts using one or more natural language processing (NLP) techniques, categories that are matched based at least in part on the transcripts, and search the plurality of outputs to obtain a search result determined based at least in part on the transcripts, metadata, and categories, and provide the search result to the client.

Abstract: Techniques for streaming real-time automated speech recognition (ASR) are described. A user can stream audio data to a frontend service of the ASR service. The frontend service can establish a bi-directional connection to an audio decoder host to perform ASR on the data stream. The audio decoder host may include a streaming ASR engine which can analyze chunks of the audio data stream using an acoustic model to divide the audio data into words, and a language model to identify sentences made of the words spoken in the audio file. The acoustic model can be trained using short audio sentence data (e.g., on the order of 30 seconds to a few minutes), enabling the transcription service to accurately transcribe short chunks of audio data. The results are then punctuated and normalized. The resulting transcript is then streamed back to the user over the bi-directional connection.

Abstract: Techniques for automated speech recognition (ASR) are described. A user can upload an audio file to a storage location. The user then provides the ASR service with a reference to the audio file. An ASR engine analyzes the audio file, using an acoustic model to divide the audio data into words, and a language model to identify the words spoken in the audio file. The acoustic model can be trained using audio sentence data, enabling the transcription service to accurately transcribe lengthy audio data. The results are punctuated and normalized, and the resulting transcript is returned to the user.

Abstract: Techniques for automated speech recognition (ASR) are described. A user can upload an audio file to a storage location. The user then provides the ASR service with a reference to the audio file. An ASR engine analyzes the audio file, using an acoustic model to divide the audio data into words, and a language model to identify the words spoken in the audio file. The acoustic model can be trained using audio sentence data, enabling the transcription service to accurately transcribe lengthy audio data. The results are punctuated and normalized, and the resulting transcript is returned to the user.

Abstract: Three dimensional [3D] image data and auxiliary graphical data are combined for rendering on a 3D display (30) by detecting depth values occurring in the 3D image data, and setting auxiliary depth values for the auxiliary graphical data (31) adaptively in dependence of the detected depth values. The 3D image data and the auxiliary graphical data at the auxiliary depth value are combined based on the depth values of the 3D image data. First an area of attention (32) in the 3D image data is detected. A depth pattern for the area of attention is determined, and the auxiliary depth values are set in dependence of the depth pattern.

Abstract: Three dimensional [3D] image data and auxiliary graphical data are combined for rendering on a 3D display by detecting depth values occurring in the 3D image data and setting auxiliary depth values for the auxiliary graphical data adaptively in dependence of the detected depth values. The 3D image data and the auxiliary graphical data at the auxiliary depth value are combined based on the depth values of the 3D image data. First an area of attention in the 3D image data is detected. A depth pattern for the area of attention is determined, and the auxiliary depth values are set in dependence of the depth pattern.

Abstract: A three dimensional display system comprises a display (207) presenting a plurality of views of a scene where each view corresponds to a viewing direction for the scene. A rendering unit (203) generates an image for each of the views. The rendering unit (203) is capable of generating differentiated images for neighboring views containing differentiated image data providing a three dimensional depth effect when neighboring views are viewed by different eyes of a viewer. A user processor (209) generates a user presence estimate in response to an attempted detection of a user in a viewing area for the display system. A 3D mode controller (213) modifies the three dimensional effect in response to the user presence estimate. For example, the display system may provide a two dimensional image when no user is present and a three dimensional image when a user is present. This may reduce discomfort experienced by viewers in suboptimal positions.

Abstract: Three dimensional [3D] image data and auxiliary graphical data are combined for rendering on a 3D display (30) by detecting depth values occurring in the 3D image data, and setting auxiliary depth values for the auxiliary graphical data (31) adaptively in dependence of the detected depth values. The 3D image data and the auxiliary graphical data at the auxiliary depth value are combined based on the depth values of the 3D image data. First an area of attention (32) in the 3D image data is detected. A depth pattern for the area of attention is determined, and the auxiliary depth values are set in dependence of the depth pattern.

Abstract: A method of generating an occlusion image property map for an occlusion viewing position for a three dimensional scene is provided. The occlusion image property map comprises at least some image property values that are occluded from the occlusion viewing position. The method utilises an algorithm which can generate an image property map for an image representing the scene as a function of a viewing position. The method generates (701, 703) image property map for different viewing positions by performing the algorithm for these positions. The occlusion image property map is generated (705) from the image property maps of different viewing positions. Specifically, the image property maps may in some examples be shifted to the occlusion viewing position and data of the occlusion image property map is then selected as a pixel from the shifted image property maps which does not correspond to the most forward pixel (unless all pixels have equal depth).

Abstract: A three dimensional display system comprises a display (207) presenting a plurality of views of a scene where each view corresponds to a viewing direction for the scene. A rendering unit (203) generates an image for each of the views. The rendering unit (203) is capable of generating differentiated images for neighbouring views containing differentiated image data providing a three dimensional depth effect when neighbouring views are viewed by different eyes of a viewer. A user processor (209) generates a user presence estimate in response to an attempted detection of a user in a viewing area for the display system. A 3D mode controller (213) modifies the three dimensional effect in response to the user presence estimate. For example, the display system may provide a two dimensional image when no user is present and a three dimensional image when a user is present. This may reduce discomfort experienced by viewers in suboptimal positions.

Abstract: An image processing system (250) and method (300) are disclosed for correcting a head pose in a video phone image, so that a frontal view is presented on a display. A disclosed head pose corrector (250) estimates the orientation of a head pose and adjusts the orientation of the head pose, if necessary, to present a frontal view. The orientation of the head pose is adjusted by generating a three dimensional model of the face surface and adjusting the orientation of the three dimensional face model to provide the desired frontal view. The head pose corrector (250) may be included in a video phone (100) to correct the head pose of transmitted or received images (or both) or may be included in a server on a network to automatically adjust the head shots of one or more participants to a video phone communication.

Abstract: The present invention relates to a method of communication (113) between a user (101) and a system (103) where it is detected whether the user looks at the system or somewhere else, and based thereon adjusting the communication.

Abstract: The invention describes a method of performing face recognition, which method comprises the steps of generating an average face model (MAV)—comprising a matrix of states representing regions of the face—from a number of distinct face images (I1, I2, . . . Ij) and training a reference face model (M1, M2, . . . , Mn) for each one of a number of known faces, where the reference face model (M1, M2, . . . , Mn) is based on the average face model (MAV). A test image (IT) is acquired for a face to be identified, and a best path through the average face model (MAv) is calculated, based on the test image (IT). A degree of similarity is evaluated for each reference face model (M1, M2, . . . , Mn) against the test image (IT) by applying the best path of the average face model (MAV) to each reference face model (M1, M2, . . . , Mn) to identify the reference face model (M1, M2, . . . , Mn) most similar to the test image (IT), which identified reference face mod el (M1, M2, . . .