Progress report of Manipuri Text to Speech Synthesis System
May 9, 2014
1 Introduction
The work mandated for IIT Guwahati is to build monolingual (Manipuri) and bilingual
(Manipuri + English) TTS systems using both USS and HTS approach. The given task is
Progress report of Assamese Text to Speech Synthesis System
May 12, 2014
1 Introduction
The work mandated for IIT Guwahati is to build monolingual (Assamese) and bilingual
(Assamese + English) TTS systems using both USS and HTS approach. The given task is
Introduction to Linear Algebra
Matrix
representation
of size n X d
Symmetric Matrix
Diagonal Matrix
Matrix multiplication with a vector
gives a transformed vector
Inner Product
Norm of a
vector
Cosine of the angle
between 2 vectors
Cauchy Schwartz
Inequal
Parameter estimation:

2
Bayesian Estimation (Bayesian learning)
In Max Likelihood was supposed fix
In Bayesian Estimation, is a random variable
Computation of posterior probabilities P(i  x)
lies at the heart of Bayesian classification
Goal: comput
Pattern Recognition Algorithms
Bag of algorithms that can used to provide
some intelligence to a machine.
These algorithms have a solid probabilistic
framework.
Algorithms work on certain characteristics
defining a class refered as features.
What is a
Bayesian Decision Making
Discriminant functions, Normal
Distribution

2
2
Classifiers, Discriminant Functions
and Decision Surfaces
The multicategory case
Set of discriminant functions gi(x), i = 1, c
The classifier assigns a feature vector x to
clas
Bayesian Decision Making

2
2
Discriminant Functions for
the Normal Density
We saw that the minimum errorrate
classification can be achieved by the
discriminant function
gi(x) = ln p(x i) + ln P(i)
Case of multivariate normal
1
1
d
t
1
g i (x) = (x i
Learning paradigms
Supervised learning
A teacher provides a category label or cost for
each pattern in the training set
Unsupervised learning
The system forms clusters or natural groupings
of the input patterns
Unsupervised Learning
The system forms
Pattern Recognition and Machine
Learning
Dr Suresh Sundaram
[email protected]
Logistics
MWF : 12 12.55 PM
Venue: L1
Dont miss classes.
Losing out on classes may leave you out of
sync with the course, leading to low grades !
Pre requisites
Bayesian Decision Making

Introduction
2
Classification different from regression in the
sense that the output will be a discrete label
denoting the entity of the class.
We will study the decision making process, by
relying on probabilistic inferences.
Parameter Estimation

Introduction
2
Data availability in a Bayesian framework
We could design an optimal classifier if we knew:
P(i) (priors)
p(x  i) (classconditional densities)
Unfortunately, we rarely have this complete
information!
Design a c
Bayesian Decision Making

2
Risk minimization framework
Let cfw_1, 2, C be the set of C states
of nature (or categories)
Let cfw_1, 2, a be the set of possible
a actions
Let (i  j) be the loss incurred for taking
action i when the state of nature is j
3
Parameter Estimation

2
MLE for Bernoulli distribution
3
=0
Overfitting of MLE
4
With low number of samples, the MLE
suffers from overfitting .leading to poor
generalization to test data.
For Bernoulli distribution, suppose a coin
is actually unbiase
HandOut Chapter  7
Physics: Alternating Current
1.
PHASORS
Alternating Current
A time varying periodic current with constant amplitude which reverses its
direction at the end of every half cycle is referred to as alternating current.
In its sinusoidal
HandOut Chapter  2
Physics: Electrostatic Potential and Capacitance
Electrostatic Potential and Capacitance
APPROACH TO CHAPTER: ELECTROSTATIC POTENTIAL AND CAPACITANCE
The numerical problems based on electric potential are relatively easy. Numerical pr
HandOut Chapter  4
Physics: Moving Charges and Magnetism
1.
MAGNETIC FIELD AND LORENTZ FORCE
Magnetic Field
Magnetic field is the space around a magnet or the space around a conductor carrying current in which magnetic
influence can be experienced.
Forc
Solutions
of
Waves & Thermodynamics
Lesson 14th to 19th
By DC Pandey
14. Wave Motion
Introductory Exercise 14.1
1. A function, f can represent wave
equation, if it satisfy
2
2 f
2 f
=
v
t 2
x 2
For,
y = a sin wt,
2 y
=  w2 a sin wt =  w2 y
2
t
2 y
but,
HandOut Chapter  6
Physics: Electromagnetic Induction
1.
FARADAYS LAWS OF ELECTROMAGNETIC INDUCTION AND LENZS RULE
Faradays Laws of Electromagnetic Induction
(i) Whenever magnetic flux linked with a circuit changes, an induced emf is always produced in
HandOut Chapter  8
Physics: Electromagnetic Waves
ELECTROMAGNETIC WAVES
1.
DISPLACEMENT CURRENT
According to Amperes circuital law, the line integral of the magnetic field ( B ) around any closed path is equal to 0
times the net current across the area
INDIAN INSTITUTE OF TECHNOLOGY GUWAHATI DEPARTMENT OF
ELECTRONICS AND COMMUNICATIONS ENGINEERING
Exp. 4: Frequency Shift Keying (FSK) by carrier switching
GROUP #18
Group Members:
Tushar Garg
Suraj Saurabh
Twisha Prasad
09010250
09010249
09010251
Date of
EC 331: COMMUNICATION LABORATORY
Experiment 1
Amplitude Modulation
Manzil Zaheer 08010252
Tushar Sandhan 08010251
Navneet Kumar 08010253
To generate an AM signal through a switching modulator (shuntbridge diode modulator) and
demodulation of the AM signa
Sheet
1 of 7
DType flipflop (Toggle switch)
The Dtype flipflops are used in prescalar/divider circuits and frequency phase detectors.
Figure 1 shows how the flipflop (latch) can be made using 2input logic circuits and Figure 2
shows the input and ou
EC 331: COMMUNICATION LABORATORY
Experiment 3
FM Generation and
Demodulation
Group 16
Navneet Kumar 08010253
Tushar Sandhan 08010251
Manzil Zaheer 08010252
To implement an FM Modulator using IC555 configured as Voltage Controlled Ocsillator and its
Demodu
EC 331: COMMUNICATION LABORATORY
Experiment 3
Frequency Modulation
Group 16
Manzil Zaheer 08010252
Tushar Sandhan 08010251
Navneet Kumar 08010253
To implement an FM Modulator using IC555 configured as Voltage Controlled oscillator and to
demodulate the ge
EC 331: COMMUNICATION LABORATORY
Experiment 1
Amplitude Modulation
Group 16
Manzil Zaheer 08010252
Tushar Sandhan 08010251
Navneet Kumar 08010253
To generate an AM signal through a switching modulator (shuntbridge diode modulator) and
demodulation of the