Computer-based systems configured to monitor a communication session and generate a training session and methods of use thereof

Patent metadata

Field	Value
Publication number	US-12413671-B2
Application number	US-202318450886-A
Country	US
Kind code	B2
Filing date	Aug 16, 2023
Priority date	Aug 16, 2023
Publication date	Sep 9, 2025
Grant date	Sep 9, 2025

Abstract

In some embodiments, the present disclosure provides an exemplary method that may include steps of monitoring, a conversation script between a call center agent and a customer; utilizing, a speech-to-text deep machine learning model to transcribe the audio call to text; utilizing, a natural language processing deep machine learning model to map intent mappings of the audio call; utilizing, a similarity measurement model to determine a semantic similarity between predefined intent mappings and the intent mappings of the call audio text; determining, an error based on the semantic similarity in the intent mapping call audio text; determining a training session based on the error.

First claim

What is claimed:

1. A computer-implemented method comprising:
retrieving, by at least one processor, a predefined call script from a predefined call script library, the predefined call script having predefined intent mappings encoding predefined text representing predefined correct dialogue associated with a sample topic;
utilizing, by the at least one processor, a computer-based monitoring module to detect an error in a call script conversation based at least in part on the predefined call script;
wherein the computer-based monitoring module is configured to: utilize at least one speech-to-text deep machine learning model to transcribe a call script conversation comprising text representative of the call script conversation;
utilizing, by at least one processor, at least one natural language processing deep machine learning model to generate a plurality of call intent mappings associated from the script conversation, wherein the at least one natural language processing deep machine learning model comprises a plurality of parameters configured to encode the text of the call script into the plurality of call intent mappings to produce semantic encodings indicative of call script;
utilizing, by the at least one processor, at least one similarity measurement model to determine a semantic similarity between the predefined intent mappings and the plurality of call intent mappings based at least in part on at least one similarity measure;
determining, based on the semantic similarity, at least one error in at least one call intent mapping of the plurality of call intent mappings;
determining, by the at least one processor, a user training need based at least in part on the at least one error;
selecting, by the at least one processor, training data for a training call based at least in part on the at least one error and the script conversation;
selecting, by the at least one processor, a training call voice for the training call based at least in part on the training data;
initiating, by the at least one processor, the training call by calling a user and loading the training data in a user dashboard of a user computing device associated with the user;
utilizing, by the at least one processor, a call generation module to automatically generate caller speech for the training call based at least in part on the training data and the training call voice;
wherein the call generation module is configured to: receive user speech data representative of speech performed by a user during the training call in response to the generated caller speech;
utilizing, by the at least one processor, the at least one speech-to-text deep machine learning model to transcribe a user speech script representative of the user speech data;
utilizing, by the at least one processor, the at least one natural language processing deep machine learning model to generate at least one user speech intent mapping associated with the user speech data;
detecting, by the at least one processor, a new error in the user speech script based at least in part on the call script and the user speech script;
determining, by the at least one processor, a user training need based at least in part on, the at least one new error; and
determining, by the at least one processor, a training session initiation based at least in part on the user training need and the at least one new error.

2. The computer-implemented method of claim 1 , wherein the predefined call script library is generated by a deep machine learning algorithm.

3. The computer-implemented method of claim 1 , wherein the predefined call script library is at least in part generated by a person.

4. The computer-implemented method of claim 1 , wherein the computer-based monitoring module schedules a training session when the computer-based monitoring module detects an error.

5. The computer-implemented method of claim 1 , wherein the computer-based monitoring module monitors a user call time and schedules a training session when the computer-based monitoring module detects an error and the user call time on call exceeds a threshold.

6. The computer-implemented method of claim 1 , wherein the semantic similarity is determined from at least a lexical phrase of the call.

7. A computer-implemented method according to claim 1 , wherein a training session schedule of the user is determined by an administrator/manager.

8. A computer-implemented method according to claim 1 , wherein upon detecting an error the computer-based monitoring module notifies an administrator/manager of the user.

9. A system comprising:
a non-transient computer memory, storing software instructions; and
a least one processor of a first computing devices associated with a user;
wherein, then at least one processor executes the software instructions, the first computing device is programmed to:
retrieve, by at least one processor, a predefined call script from a predefined call script library, the predefined call script having predefined intent mappings encoding predefined text representing predefined correct dialogue associated with a sample topic;
utilize, by the at least one processor, a computer-based monitoring module to detect an error in a call script conversation based at least in part on the call script;

wherein the computer-based monitoring module is configured to: utilize at least one speech-to-text deep machine learning model to transcribe a call script comprising text representative of the call script of the conversation;
utilize, by the at least on processor, at least one natural language processing deep machine learning model to generate a plurality of call intent mappings associated from the conversation, wherein the at least one natural language processing deep machine learning model comprises a plurality of parameters configured to encode the text of the call script into the plurality of call intent mappings to produce semantic encodings indicative of call script;
utilize, by the at least one processor, at least one similarity measurement model to determine a semantic similarity between the predefined intent mappings and the plurality of call intent mappings based at least in part on at least one similarity measure;

determining, based on the semantic similarity, at least one error in at least one call intent mapping of the plurality of call intent mappings;
determine, by the at least one processor, a user training need based at least in part on the at least one error;
select, by the at least one processor, training data for a training call based at least in part on the at least one error and the topic;
select, by the at least one processor, a training call voice for the training call based at least in part on the training data;
initiate, by the at least one processor, the training call by calling the user and loading the training data in a user dashboard of a user computing device associated with the user;
utilize, by the at least one processor, a call generation module to automatically generate caller speech for the training call based at least in part on the training data and the training call voice;

wherein the call generation module is configured to: receive user speech data representative of speech performed by the user during the training call in response to the generated caller speech;
utilize, by the at least one processor, the at least one speech-to-text deep machine learning model to transcribe a user speech script representative of the user speech data;
utilize, by the at least one processor, the at least one natural language processing deep machine learning model to generate at least one user speech intent mapping associated with the user speech data;
detect, by the at least one processor, a new error in

Assignees

• Capital One Services Llc

Inventors

• Edwards Joshua
• Lin Alexander
• Rodriguez Mia
• Bonilla Guadalupe
• Can Aysu Ezen
• Mossoba Michael
• Qiu Feng
• Maiman Tyler
• Critzer Meredith L

Classifications

• G06F40/35

  Discourse or dialogue representation · CPC title

• G06F40/216

  using statistical methods · CPC title

• G10L15/22

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• H04M2203/403

  Agent or workforce training · CPC title

• H04M2201/40

  using speech recognition · CPC title