Method and system for recording and indexing audio and video conference calls allowing topic-based notification and navigation of recordings
First Claim
1. A system for recording and indexing audio and video conference calls comprising, a configuration subsystem, at least one recording subsystem connected to the configuration subsystem for connecting an external audio or video conferencing system, at least one indexing subsystem connected to the at least one recording subsystem for automatically determining topic based intervals, at least one management subsystem connected to the at least one indexing subsystem for storing results of the indexing subsystem, at least one notification and navigation system connected to the at least one management subsystem for allowing a plurality of subscribers to query the at least one management subsystem and to submit requests to be notified of an occurrence of a topic based event, and at least one playback subsystem connected to the at least one recording subsystem for storing recorded conference calls in a variety of formats suitable for playback on a variety of devices.
1 Assignment
0 Petitions
Accused Products
Abstract
Recording and indexing audio and video conference calls allows topic-based notification, navigation and playback of one or many recordings. An external conferencing system with five internal modules, recording, indexing, management, notification, and playback allows a plurality of meeting participants to record a conference call, and a plurality of subscribers to access, search, and be notified of the occurrence of an event of interest. The system automatically creates contextual or topic based time intervals which are used to immediately notify a subscriber of an event occurrence. The subscriber can use the time stamps of the topic intervals to navigate and playback the recording. The system also indexes project management files and correlates recording intervals to project tasks, and subsequently correlates multiple recording intervals.
202 Citations
49 Claims
- 1. A system for recording and indexing audio and video conference calls comprising, a configuration subsystem, at least one recording subsystem connected to the configuration subsystem for connecting an external audio or video conferencing system, at least one indexing subsystem connected to the at least one recording subsystem for automatically determining topic based intervals, at least one management subsystem connected to the at least one indexing subsystem for storing results of the indexing subsystem, at least one notification and navigation system connected to the at least one management subsystem for allowing a plurality of subscribers to query the at least one management subsystem and to submit requests to be notified of an occurrence of a topic based event, and at least one playback subsystem connected to the at least one recording subsystem for storing recorded conference calls in a variety of formats suitable for playback on a variety of devices.
-
44. A method of automatically indexing audio or video conference calls for creating topic-based intervals comprising the steps of partitioning an audio input into one or more discrete segments, processing each of the segments using speech recognition, and saving recognized words and corresponding timestamping in a database, iterating over local sets of the recognized words, and looking up co-occurrence of the recognized words from a generic co-occurrence database, saving start and stop time of every interval where an entire set of recognized words have sufficiently high co-occurrence values, iterating over a set of saved intervals and creating a context-vector further comprising a vector representation of words in an interval within some defined N-space of words.
-
45. A method of automatically indexing project management files to create task context-vectors and task correlation values comprising the steps of iterating over each task in a project file and extracting task information including task name, description, comments, resource assignment, task dependencies and task schedule, and saving the task information in a task database, calculating a context-vector using the task name, the description and the comments for creating a vector representation of the task within an N-space of words, and saving the context-vector in a context-vector database, calculating a task-to-task correlation weighting based on the relationship between tasks, and saving the correlation weighting in a correlation weighting database.
- 46. A method of determining recording interval to task relationships comprising the steps of iterating over a set of intervals determined using the method of automatically indexing audio or video conference calls for creating topic-based intervals comprising the steps of partitioning an audio input into one or more discrete segments, processing each of the segments using speech recognition, and saving recognized words and corresponding timestamping in a database, iterating local sets of the recognized wards, and looking up co-occurrence of the recognized words from a generic co-occurrence database, saving start and stop time of every interval where an entire set of recognized words have sufficiently high co-occurrence values, iterating over a set of saved intervals and creating a context-vector further comprising a vector representation of words in an interval within some defined N-space of words, and for each interval further iterating over each task determined using the method of automatically indexing project management files to task context-vectors and task correlation values comprising the steps of iterating over each task in a project file and extracting task information including task name, description, comments, resource assignment, task dependencies and task schedule, and saving the task information in a task database, calculating a context-vector using the task name, the description and the comments for creating a vector representation of the task within an N-space of words, and saving the context-vector in a context-vector database, calculating a task-to-task correlation weighting based on the relationship between tasks, and saving the correlation weighting in a correlation weighting database, and determining interval task pairs, for each interval task pair, obtaining a dot product of the context-vectors from recording intervals and the context-vectors of the task, storing interval-task correlation values in a database.
Specification