Multi-party conversation analyzer & logger

• US 20110082874A1
• Filed: 09/20/2008
• Published: 04/07/2011
• Est. Priority Date: 09/20/2008
• Status: Active Grant

• Current Assignee: Securus Technologies Incorporated (Securus Special Investments LLC)
• Original Assignee: Charles Barrasso, George Roberts, Jay Gainsboro, Jonathan Klein
• Inventors: Klein, Jonathan, Barrasso, Charles, Gainsboro, Jay, Roberts, George
• Granted Patent: US 8,886,663 B2
• Time in Patent Office: Days
• US Class Current: 707/769
• CPC Class Codes: G06F 16/2428 Query predicate definition using graphical user interfaces, including menus and forms G06F16/2423 takes precedence
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: H04M 3/2218 Call detail recording
• CPC Class Codes: H04M 3/2281 Call monitoring, e.g. for law enforcement purposes; Call tracing; Detection or prevention of malicious calls

Abstract

The multi-party conversation analyzer of the present invention allows users to search a database of recorded phone calls to find calls which fit user-defined criteria for “suspicious calls”. Such criteria may include indications that a call included a 3-way call event, presence of an unauthorized voiced during the call, presence of the voice of an individual known to engage in identity theft, etc. A segment of speech within a call may be graphically selected and a search for calls with similar voices rapidly initiated. Searches across the database for specified voices are speeded by first searching for calls which contain speech from cohort speakers with similar voice characteristics.

Claims

1. A method allowing a user to automatically search a database containing data derived from a plurality of recorded phone calls, and display data pertaining to suspicious calls, said method comprising:

presenting to said user a graphical user interface comprising a menu of call selection criteria;

automatically searching said database for suspicious calls based on said criteria; and

automatically dynamically reconfiguring said graphical user interface to additionally display data pertaining to suspicious calls.

2. The method of claim 1, wherein said call selection criteria comprises calls where it is suspected that speech in a suspect speech segment is speech of an individual who has falsely identified himself or herself, and wherein said step of automatically searching said database comprises automatically scoring relative likelihoods that said suspect speech segment was spoken by each of a plurality of individuals for whom voice model data is available.

3. The method of claim 2, wherein said step of automatically searching said database comprises parsing voices of possible speakers into sets of cohort speakers and determining which set of cohort speakers said suspect speech segment fits, and said automatic scoring is done only for voices of individuals within said cohort speaker set.

4. A method allowing a user to automatically search for calls likely to contain speech from a given individual within a database containing data derived from a plurality of recorded phone calls, said method comprising:

displaying data representative of speech from said individual on a graphical user interface along a timeline;

manually selecting within said graphical user interface along said timeline a portion of speech from said individual;

deriving a speech model or voice print from audio data representing said portion of speech from said individual; and

automatically searching said database for conversation segments with speech characteristics close to those of said portion of speech from said individual.

5. The method of claim 4, further comprising the step of determining which of a plurality of sets of cohort speakers said portion of speech from said individual falls within, and wherein speech segments in said database are pre-identified as belonging to one of said plurality of groups of cohort speakers, and said step of automatically searching said database is limited to searching speech segments coming from persons within the same set of cohort speakers the speech segment from said individual falls in.

6. A method allowing a user to automatically search for calls likely to contain speech from a given individual within a database containing data derived from a plurality of recorded phone calls, said method comprising:

displaying data representative of speech from said individual on a graphical user interface along a timeline;

manually selecting within said graphical user interface along said timeline a portion of speech from said individual;

automatically doing a speech-to-phoneme conversion of said selected portion of speech to derive a phoneme sequence; and

automatically searching said database for conversation segments whose phonetic sequences represent a good statistical pattern match to said phoneme sequence.

7. A method facilitating automatic flagging of future calls containing the speech of a given individual, said method comprising:

displaying data representative of speech from said individual on a graphical user interface along a timeline;

manually selecting within said graphical user interface along said timeline a portion of speech from said individual;

deriving a speech model or voice print from audio data representing said portion of speech from said individual; and

monitoring future calls for voices whose voice model characteristics are a statistically close match to those of said individual; and

automatically phone calls on which such statistically close match is detected.

8. A method of searching a database containing audio data from a plurality of conversations in which an individual participated, said method comprising:

displaying data representative of speech from said individual on a graphical user interface along a timeline;

manually selecting within said graphical user interface along said timeline a portion of speech from said individual in which said individual evidences a particular emotional state;

automatically extracting first prosody data from said selected portion of speech; and

automatically searching said database for conversation segments where said individual was speaking, and prosodic data derived from such segments represent a good statistical match to said first prosody data.