speech analyser for call centers FYP 3July 2017 - PowerPoint PPT Presentation

About This Presentation
Title:

speech analyser for call centers FYP 3July 2017

Description:

speech analyser for call centers FYP 3July 2017 (Lutaaya Shafiq, Lusoma Joseph, Samira Zein, Wasike Timothy- Team CS16-07) – PowerPoint PPT presentation

Number of Views:125

less

Transcript and Presenter's Notes

Title: speech analyser for call centers FYP 3July 2017


1
Speech Analysis System Speech analyzer for Call
Centers
  • CSC 17- 06

GROUP MEMBERSHIP SAMIRA ZEIN
14/U/16012/PS WASIKE TIMOTHY 14/U/15867/PS LUTAAY
A SHAFIQ 14/U8759/PS LUSOMA JOSEPH 14/U/8747/PS
https//www.speechanalyser.com
2
Speech analyzer for Call Centers by CS17-06
System Developers.
SAMIRA ZEIN 14/U/16012/PS WASIKE
TIMOTHY 14/U/8747/PS LUTAAYA SHAFIQ 14/U8759/PS LU
SOMA JOSEPH 14/U/15867/PS
CS17-06
Supervisor Assoc. Prof., Dr ENGINEER
BAINOMUGISHA, PhD
3
Speech analyzer for Call Centers by CS17-06
Overview of Our Presentation
CS17-06

4
Project Introduction
Speech analyzer for Call Centers by CS17-06
1
5
Speech analyzer for Call Centers by CS17-06
Background of the study
1
Many businesses focus on customer relationship
management systems and techniques to manage the
relationships between them and their customers
and Call Centers are a widely used technique
to follow up on managing a companys interaction
with current and potential customers as well
as customer satisfaction. A call center uses
data analysis about the customers history with a
company to improve business relationships with
their customers, specifically focusing on
customer retention and driving sales growth to
the company.
6
Business decision making and business performance
evaluation have proved to be a major problem
facing large organizations with big data
specifically customer call center data. Call
centers, put more emphasis on solving issues that
come from calls but none on the number of calls
received concerning a particular issue so that
they can get long term solutions to the problem
and this leaves the information about the trend
of issues involved in the customer calls are not
being catered for.
Speech analyzer for Call Centers by CS17-06
Problem Statement
7
This project is intended to create an effective
speech analysis system for call centers that
analyzesand produces business insights from
recorded customer calls so as to support
organizationsdecision making process.
Speech analyzer for Call Centers by CS17-06
Objective of the Project
8
Supervised Learning. - Functions are inferred a
labeled training dataUnsupervised Learning - A
function is inferred to describe hidden structure
from Un labeled data
Speech analyzer for Call Centers by CS17-06
EXISTING APPROACHES
9
The Call surf. - It analyses speech for call
centers but it uses French as its language
model.customer relation management
systems-Zoho crm, Prosper works
Speech analyzer for Call Centers by CS17-06
EXISTING SYSTEMS
10
-Use foreign language models-Very complex to
learn and use
Speech analyzer for Call Centers by CS17-06
CHALLENGES FACE BY THESE SYSTEMS
11
The speech analyzer for call center systems is
aimed at solving the above issues by
enablingtapping into the content of
conversations
Speech analyzer for Call Centers by CS17-06
THE PROPOSED SYSTEM
12
Speech analyzer for Call Centers by CS17-06
STEPS REQUIRED TO ARCHIEVE THE PROJECT OBJECTIVE
Phonetic Indexing Converting the conversation
into a string of phonemes, the basic units of
speech, will allow the identification of a
predefined list of words or allow free text
searches. this approach is perfect for assessing
calls as they happen
Speech to Text This approach converts entire
audio exchanges into standard text, facilitating
deeper data mining. By independently identifying
patterns, across many recordings and channels,
The Speech analyzer will help to find root causes
and trending topics in customer interactions?
13
Speech analyzer for Call Centers by CS17-06
2
3
Methodology System analysis/design




14
Speech analyzer for Call Centers by CS17-06
  • Data Collection Techniques
  • First, at least 10 or more hours of speech need
    to be collected for human transcription, to
    provide in-context example on how all the
    keywords are spoken on call centers
  • The two analysis results
  • - Qualitative
  • -Quantitative

Methodology
2.Interviews
15
Data Analysis and Design
  • User Requirements
  • Voice based email system will be a phone
    application which is-
  • Easy to learn.
  • It should be able to understand and adopt to
    users pronunciation.
  • It will recognize words in English.
  • Functional Requirements
  • It accepts input from the user.
  • Synchronizes with the email.
  • Processes data and transfers it in a 2-way for
    example voice-to-text and text-to-voice

Systems Analysis
  • Non-Functional Requirements
  • Security requirements.
  • Performance.
  • Hardware and software requirements.

16
Speech analyzer for Call Centers by CS17-06
2
3
System analysis/design




Click on Audio File Icon Below to Listen to More
information
17
Speech analyzer for Call Centers by CS17-06
2
System Design



  • Data Flow Diagrams
  • Represents how data moves through system
  • Use Case Diagrams
  • Represent interaction between users and the system

18
Speech Analyzer for Call Centers SACC How the
System WORKS
The Speech Analysis System for Call Centers has
been developed to provide accurate information
about calls by providing a summarized and
categorized output of received calls in form of
a simple bar, line graph ,bar graph, pie chart,
bubble chart or table. The system
analyzes and produces business insights from
recorded customer calls that aid in the
organizations performance evaluation and the
decision making process as well.
19
SACC Dash Board The dashboard interface
automatically generates customer calls recent
analysis. This interface provides a brief
information overview to the user.
20
SACC Analysis Page and Sentiment Analysis
Pane This interface previews the dynamic
analysis made by the speech analysis system. By
default the analysis page is set to display
information for the current week. Sentiment
Analysis (on the right) deals with the
Positivity and Negativity in the Analyzed data
21
SACC Analysis page Bar Graph This interface is
a representation of information in terms of bars,
it can also be reflect using different chart as
seen from the previous diagram
22
Analysis Page Line Graph The analysis page also
allows users to also display an in form of a line
graph. Different points on a line represents the
occurrence or frequency of the issue in each month
23
Analysis Page Months The above interface
provides users with information about call center
issues per month in a full year.
24
Analysis page Menu option-after right clicking
the mouse When the users right click the menu. A
menu list appears having, 1. Back this option
is used to load the previous page 2. Forward
The option move to the previous current page 3.
Print Print information in the PDF format 4.
Open in external Browser and Open frame
25
Settings window
This interface requires the system administrator
to provide the storage location of the call
center audio storage. He must also provide the
location for backing up the analyzed data.
26
QR CODE takes user to www.speechanalyser.com
27
Download link https//www.speechanalyser.com/d
ownload DOWNLOAD SYSTEM SETUP page When the
user Clicks on the DOWNLOAD link on the MENU bar,
the system redirects them to the download page
which is only accessible to Logged in users who
must have paid membership on the system before
they can access the Speech analysis system Setup
Installer, and other resources like timely
updates.
28
Login Page. LOGIN If a user is a member, that
is they have already registered and have paid
membership (via PayPal), they are able to log in
with their username and password, and are able to
access the DOWNLOAD page with the download link
to the System Setup Installer of which they can
now download and install the software on their
local computer / server.
29
User Registration Page NEW USER REGISTRATION
Otherwise, if the user is not registered with the
speech analysis website, they have to opt for
this registration page and then fill in a desired
login username, password, email, names and other
details and is then redirected to proceed to the
checkout page and the payment for the
subscription is done. A password is emailed to
the email that the user submits. Thereafter, they
can now login to the system.
30
SYSTEM IMPLEMENTATION
  • System will be developed using
  • Python Language
  • Why Python?
  • Portability
  • Easy to learn
  • Has very good data mining modules

31
SYSTEM IMPLEMENTATION
  • System will be developed using
  • Modules like PyAudioAnalysis
  • Anaconda Features
  • Integrated Development Environments(IDEs)
  • PyCharm

32
Derive Customer Based strategic insights
Architectural design for speech analyzer for call
centers
produces
Speech Analysis System
Customer calls the call center
Call Center System
Server provides access to recorded data
The call is recorded and saved on the server
33
Architectural design for speech analyzer for call
centers
34
Algorithms for Transforming Speech to
Structured Data
35
Speech analyzer for Call Centers by CS17-06
Speech Recognition Techniques
3



  • Simple Pattern Matching
  • Pattern and Feature Analysis
  • Language Modelling and Feature Analysis
  • Algorithm will convert Text to Speech and then
    Make comparisons with in-built category Features

36
Turning text to SPEECH
  • The transcribed words must be entered in
    pronunciation dictionary showing the commonest
    sequences of sounds people make as they say each
    word.
  • If there is no existing pronunciation dictionary
    available, one needs to be created by listing all
    of the words in the desired language and how they
    are all constructed from the phonemes theirs
    speakers use.

37
SYSTEM TESTING AND VALIDATION
  • Testing individual components of the system.
  • Testing the system as a whole

38
FINAL OUTPUT OF OUR PRODUCT
  • The Analysed will be output in a simplified and
    understandable way.
  • Consider a telecom company like Airtel or MTN or
    any other
  • This Data Will be viewed Inform of

39
BAR GRAPH
40
PIE CHART
41
LINE GRAPH
42
THE END
For more information contact us
www.speechanalyser.com
TEAM CONTACTS Assoc. Prof. Dr Engineer
Bainomugisha ebainomugisha_at_cis.mak.ac.ug LUSOMA
JOSEPH 256705781367 josephlusoma_at_gmail.com LUT
AAYA SHAFIQ 256702772721 lutayashafiq_at_gmail.co
m SAMIRA ZEIN 256705891024
samirazein20_at_gmail.com WASIKE TIMOTHY 25677343931
2 wasiketimothylambert_at_gmail.com See more
https//www.speechanalyser.com/team
Write a Comment
User Comments (0)
About PowerShow.com