Microsoft Technology Licensing, LLC

Other

Yiu-Ming Leung

Vincent Allio

Jisun Park

Warren Vincent Barkley

Pavel Curtis

Jiannan Zheng

Tim M Moore

Jay Herbison

Haoran Zheng

Li Zhang

David A Nichols

Rui Liang

Lei He

Xu Bin Liu

Danny Levin

Bo Yan

Pei Zhao

Sergei Osobov

Christopher Schindler

Yu Shao

Marcelo Farjalla

Danilo Perez Solano

Zhe Geng

BoRu Yan

Pronunciation issues for synthesized speech are automatically detected using human recordings as a reference within a Speech Recognition Assisted Evaluation (SRAE) framework including a Text-To-Speech flow and a Speech Recognition (SR) flow. A pronunciation issue detector evaluates results obtained at multiple levels of the TTS flow and the SR flow (e.g. phone, word, and signal level) by using the corresponding human recordings as the reference for the synthesized speech, and outputs possible pronunciation issues. A signal level may be used to determine similarities/differences between the recordings and the TTS output. A model level checker may provide results to the pronunciation issue detector to check the similarities of the TTS and the SR phone set including mapping relations. Results from a comparison of the SR output and the recordings may also be evaluation by the pronunciation issue detector. The pronunciation issue detector outputs a list that lists potential pronunciation issue candidates.

Speech recognition assisted evaluation on text-to-speech pronunciation issue detection

Optimized conferencing performance may be provided. First, a plurality of data streams respectively received from a plurality of conferencing users may be monitored. Then, for each of the plurality of conferencing users, a plurality of talk frequency conditions respectively corresponding to the plurality of conferencing users may be determined based upon the monitored plurality of data streams. The plurality of talk frequency conditions may comprise, for example, active-talker, infrequent talker, or listener-only. Next, a plurality of data packet size values respectively corresponding to the plurality of conferencing users may be determined based upon the determined plurality of talk frequency conditions. The plurality of data streams may then be mixed to create data. Next, the data may be transmitted to each of the plurality of conferencing users respectively using the determined plurality of data packet size values respectively corresponding to the plurality of conferencing users.

Optimizing conferencing performance

A system and method for mapping payload data to a spreadsheet list. The present invention comprises a system for mapping payload data to a spreadsheet list while supporting the implementation of nested tables, without the use of a schema. After receiving a spreadsheet list, the present invention generates the parent path of each field path defined in the spreadsheet list. The longest parent path may be used to generate a set of nodes from the payload data. If more than one unique parent path qualifies as the longest parent path, then such parent patents may be designated as branch bottom paths. All other parent paths may then be associated with an appropriate branch bottom path. During the binding process, each branch bottom path may be evaluated to generate result nodes, which may be used to obtain the values of the fields of the spreadsheet list.

System and method for schemaless data mapping with nested tables

A platform and end user interface are provided as part of a contextual communications framework for enabling developers to add context to conversations before they start or as they occur. The context may be consumed through a communication application user interface or through an application associated with a contextual document or file. The exchanged context and any updates to it may be stored following the conversation such that subsequent continuations of the conversation may be provided with the updated context.

Contextual conversation framework

Performance of endpoints, client devices and servers within a communication system, is determined by collecting call quality data from each endpoint by a quality monitoring server/application. Call quality data includes predefined metrics based on network and end device characteristics during each call. Calls include voice, video, and data exchanges. Collected metric values are then aggregated according to a formula for consistency and scaled based on factors such as traffic volume. Resulting performance values are used to order the endpoints such that those with degraded quality and prioritized based on factors like traffic volume can be attended to first.

Estimating endpoint performance in unified communication systems

Software applications are enabled with multimodal communication capabilities and features associated with enhanced communication systems through a communication application programming interface (API). Applications for document processing, scheduling, audio/video data processing, and/or business operations can exchange communication system data and parameters bi-directionally with components of an enhanced communication system, initiate communication sessions through the application's user interface, and preserve post-communication updates.

Communication interface for non-communication applications

User quality of experience in a real time communication system is determined by analyzing endpoint performance metric reports correlated with performance factors of each endpoint in an endpoint profile for a more complete view of an audio/video quality of experience. Reports may include endpoint profile reports, which may be sent once at the beginning and then again only if an endpoint attribute changes; session reports, which include measured metric information and are sent at the conclusion of each session; and session interval reports, which also include metric information, but are sent during the session based on a period or a predefined threshold.

Endpoint report aggregation in unified communication systems

Tools and techniques for determining conditions of conferences include calculating a global conference parameter that represents a rate at which several different conferences are processed. The machine may also calculate individual conference parameters representing a duration of time spent in processing the conferences considered individually.

Determining conditions of conferences

An XML payload is generated from an XML list irrespective of a schema associated with the XML list. The parent/child relationship paths are collected from the field entries in the XML list. An XML payload node tree is created reflective of the parent/child relationship of data in the paths of the XML list. The XML payload can then be generated from the XML payload node tree and exported as needed to target software applications or web pages.

Schemaless XML payload generation

Performance of endpoints, client devices and servers within a communication system, is monitored by collecting call quality data from each endpoint. The collected metric data is retrieved periodically from a database, categorized based on endpoint location and metric type, and compared to predefined thresholds for different endpoint types, metric categories, and call configurations. If an alert condition is met indicating a problem with quality of experience in the system, an alert is issued to an administrator.

Determining quality monitoring alerts in unified communication systems

Tools and techniques for application-specific heap management are described herein. The tools may provide machine-readable storage media containing machine-readable instructions for profiling an application to facilitate managing heap memory associated with the application, and for managing requests from the application to allocate or deallocate from the heap memory based on the profiling. The tools may also receive requests from the application to allocate buffers, and may determine whether an instance-level memory pool, which is associated with a portion of the application, contains enough free buffers to satisfy the request. Finally, the tools may receive requests from the application to deallocate buffers, and in response to the request, may deallocate the requested buffers into the instance-level memory pool. The tools may also determine whether the instance-level memory pool contains a number of free buffers that exceeds a threshold.

Application-specific heap management

A subsystem provides for availability checking on all necessary/desired objects prior to creating and coupling the objects. In one embodiment, the subsystem is knowledgeable of the objects it supports and each of the objects' related objects. The subsystem uses this information to determine the objects that are supported by both itself and a remote subsystem. The subsystem then uses this information to determine, prior to creating an instance of an object, whether the object and all of its related objects are supported by both the subsystem and the remote subsystem.

Method and system for availability checking on distributed objects

A subsystem provides for negotiating a common version for interfaces of objects supported by the subsystem and a remote subsystem. In one embodiment, the subsystem determines a list of interfaces, including a version range for each interface, of objects supported by the subsystem, and a list of interfaces, including a version range for reach interface, of objects supported by the remote subsystem. The subsystem then negotiates a list of a version of each interface supported by both the subsystem and the remote subsystem by matching the versions for each interface and selecting the highest version that is common to both the list of interfaces of objects supported by the subsystem and the list of interfaces of objects supported by the remote subsystem.

Growing community of inventors

Kirkland, WA, United States of America

Yiu-Ming Leung