System for conversation quality monitoring of call center conversation and a method thereof

• US 8,761,376 B2
• Filed: 01/16/2013
• Issued: 06/24/2014
• Est. Priority Date: 02/13/2012
• Status: Active Grant

• Current Assignee: TATA Consultancy Services Limited (Tata Sons Pvt Ltd.)
• Original Assignee: TATA Consultancy Services Limited (Tata Sons Pvt Ltd.)
• Inventors: Kopparapu, Sunil Kumar, Pande, Arun
• Primary Examiner(s): Asanbayev, Oleg
• Application Number:
  US13/742,829
• Publication Number:
  US 20130208881A1
• Time in Patent Office: 524 Days
• Field of Search: 379/32.01, 379/100.05, 379/114.01, 379/133, 379/265.06, 37926501-26514, 37926601-2661
• US Class Current: 379/265.06
• CPC Class Codes: H04M 2201/38 Displays
• CPC Class Codes: H04M 2203/401 Performance feedback
• CPC Class Codes: H04M 3/2281 Call monitoring, e.g. for law enforcement purposes; Call tracing; Detection or prevention of malicious calls
• CPC Class Codes: H04M 3/5175 Call or contact centers supervision arrangements

Abstract

The present invention is related to improving a conversation quality of calls in a call center setup/environment. The system of the present invention allows monitoring of several live ongoing audio conversations to alert supervisor regarding vulnerable/deviations in the call being handled by a call center agent. The alerts are displayed on the supervisor console by way of providing graphical visual display in order to seek attention of the supervisor. The system also facilitates the supervisor to listen to the live conversation on his desktop by clicking on the visual display.

Claims

1. A system for real time monitoring of a plurality of calls, each of the plurality of calls having an agent speech segment and a customer speech segment, the system comprising:

a supervisor console comprising a processor and a memory, and further comprising;

a segmentation module configured to disintegrate at least one of the plurality of calls into the agent speech segment and the customer speech segment;

an extraction module configured to extract a quality of the agent speech segment, wherein the quality of the agent speech segment comprises at least two of;

speaking rate, emotional state, loudness of an agent, and quality of speech, and wherein the extraction module analyzes the quality of the agent speech segment by comparing the quality of the agent speech segment against predefined quality parameters;

a display module comprising an actionable call window configured to receive and display in real time, a visual representation of the agent speech segment, wherein the visual representation of the agent speech segment is not a speech to text and closed captioning;

an alert module configured to classify the at least one of the plurality of calls as at least one vulnerable call based on a detection of a divergence from the predefined quality parameters, wherein the alert module notifies a supervisor through the actionable call window by displaying the divergence on the visual representation of the agent speech segment; and

a call switching module configured to transfer the at least one vulnerable call to a randomly selected idle agent and/or allow the supervisor to instantaneously connect to the at least one vulnerable call.

2. The system of claim 1, wherein the supervisor console comprises an actionable call progress window for each of the plurality of calls for monitoring performance of the randomly selected idle agent.

3. The system of claim 1, wherein the call switching module is further configured to provide an actionable bridging and connecting of the at least one call.

6. The system of claim 1, wherein the segmentation module extracts at least one speech feature from the at least one of the plurality of calls, wherein the at least one speech feature is used to disintegrate the at least one of the plurality of calls into the agent speech segment and the customer speech segment.

7. The system of claim 6, wherein the at least one speech feature includes at least one of:

a pitch, a variation in pitch, a signal energy, a Mel-scale frequency Cepstral coefficient (MFCC);

Linear predictor coefficients (LPC);

Cepstral coefficient-Linear predictor coefficients (CC-LPC); and

line spectral pairs (LSP).

4. A method for real time monitoring of a plurality of calls, each of the plurality of calls having an agent speech segment and a customer speech segment, the method comprising:

segmenting, by a segmentation module located in a supervisor console, at least one of the plurality of calls into the agent speech segment and the customer speech segment;

extracting, by an extraction module located in the supervisor console, a quality of the agent speech segment and analyzing the quality of the agent speech segment by comparing the quality of the agent speech segment against predefined quality parameters, wherein the quality of the agent speech segment comprises at least two of;

speaking rate, emotional state, loudness of an agent, and quality of speech;

rendering, in real-time, a visual representation of the agent speech segment on a display module located in the supervisor console, wherein the display module comprises an actionable call window, and wherein the visual representation of the agent speech segment is not a speech to text and closed captioning;

classifying, by an alert module located in the supervisor console, the at least one of the plurality of calls as a vulnerable call based on a detection of a divergence from the predefined quality parameters, wherein a supervisor is notified through the actionable call window by display of the divergence on the visual representation of the agent speech segment; and

transferring, by a switching module located in the supervisor console, the vulnerable call to a randomly selected idle agent and/or allowing the supervisor to instantaneously connect to the vulnerable call.

5. The method of claim 4, wherein the supervisor console comprises the actionable call window for each of the at least one call to monitor a performance of the randomly selected idle agent.