Real Time Voice Recognition based Smart Home

notownbuffΤεχνίτη Νοημοσύνη και Ρομποτική

17 Νοε 2013 (πριν από 3 χρόνια και 6 μήνες)

58 εμφανίσεις

Real
T
ime
V
oice
R
ecognition based
Smart H
ome
A
pplication


Abstract:


By the improvement of the technology, the applications used for simplifying
the daily life create an important work area day after day. In this study, an
equipment control component that could be used as intelligence home application
based on voice recogni
tion was improved. By this aim, database belonging to
intended control voices like turn on the lamp, close the curtain was created by pre
-
processing the steps of the voice samples taken from five people. For education
and test steps, the performance of the

algorithm was compared by using Euclidian
distance method, artificial neural network and dynamic time warping algorithm
separately in the voice recognition step. As for the recognition step, the proposed
are learned and detected using statistical models H
ybrid HMM’s

w
ere verified by
the implementation of the duties that correspond to pairing words as real time on
the control cards.

The proposed general framework is two
-
fold: the

offline analysis
of human behavior recordings and the

online detection of
learned human behavior
models. To evaluate

the proposed approach, several multimodal recordings
showing

different situations have been conducted
. The Offline analysis
showing
that

multimodality as well as multiperson observation generation are

benef
icial f
or
situation recognition
a hidden Markov

model is a stochastic process

where the
evolution is managed by with eight states
.




Existing system
:


1.

Dynamic Time Warping Algorithm

for voice recognition,

2.

Establishing the starting and ending point is problem if
the audio recording is
performed in ideal conditions.

3.

The comparison process in both cases must compensate for both the different
length of the sequences and non
-
linear nature of the sound
.

4.

Decreasing the number of numeric values can be performed by
removing
those ones between the extreme points.

5.

Signal and original signal begin to increase and rapidly approaches to peak at
the fundamental period
.


Proposed system
:

1.

High control component that could be used as intelligence home application
based on voi
ce recognition
.

2.

Voice recognition system is used to translate voice commands
complementing for the human
-
intuitive control of a personal service based
on new framework.


3.

It can be processed in an interaction distance, a role value for each entity.
Based on

role and sound detections, several situations are learned and
detected using hidden Markov models.

4.

The environment is used for noise detection. Based on the energy of the
audio streams, we determine
energy detector, a basic classifier
and a neural
net
trained to rec
ognize voiced segments like vowel
.

5.

System can detect the situation for an observation sequence, the probability
of this sequence
has
learned by HMM with the highest probability to
determine the situation label for the sequence.

Software Requi
rement:

Tool : Matlab 2012

Toolbox : Image Processing

Front End: GUI User Interface


Hardware Requirements


SYSTEM


: Pentium IV 2.4 GHz


HARD DISK


: 40 GB


RAM



: 2 GB