Sei sulla pagina 1di 24

02

CONTENTS
PROJECT OVERVIEW
BLOCK DIAGRAM
VOICE RECOGNITION USING
MATLAB
CHALLENGES
FEATURES AND APPLICATIONS
ESTIMATED COST

4/19/15
05:27:16 AM

PROJECT STATUS
CONCLUSION

03

4/19/15
05:27:17 AM

PROJECT OVERVIEW
Voice recognition systems are
common today.
But we do not rely on it that
much.
It is because of the lack of
user -friendliness of those
systems.
Imagine, if you can speak to
your machines as if it were a
human.

04

PROJECT OVERVIEW
(continued)

4/19/15
05:27:17 AM

The name J.A.R.V.I.S stands


for JUST A RATHER VERY
INTELLIGENT SYSTEM
The name was adopted from
the movie-Iron Man.
The objective is to humanly
interact with a system that
runs some applications and
devices.
This project has wide range of

05

PROJECT OVERVIEW
(continued)

Atmega8

4/19/15
05:27:17 AM

06

PROJECT OVERVIEW
(continued)

4/19/15
05:27:18 AM

The user authentication is


done using a digital key
(figure is shown later).
A MicroController (ATmega8)
is interfaced with PC.
The Microcontroller fetches
the Temperature and Humidity
of the surroundings to the
Matlab.
A GUI displays the current

07

PROJECT OVERVIEW
(continued)

4/19/15
05:27:18 AM

After authentication, you can


start talking to your system.
It will speak back to you for
every actions it take.
This is done by creating two
databases.
One for the voice recognition
and the other for speech
response like in IVR.
The bigger the database, the

08

NOT

NOT

4/19/15
05:27:18 AM

AND

09

GENERAL BLOCK DIAGRAM

4/19/15
05:27:18 AM

10

VOICE RECOGNITION
USING MATLAB
The speech processing is done in
Matlab running PC.
The voice recognition is using MFCC
(Mel Frequency Cepstral Coeficient)
algorithm.
MFCC is a feature widely used in
automatic speech and speaker
recognition.

4/19/15
05:27:19 AM

11

VOICE RECOGNITION
USING MATLAB (continued)
It was introduced by Davis and
Mermelstein in the 1980's, and have
been state-of-the-art ever since.
The steps for implementation.

4/19/15
05:27:19 AM

Frame the signal into short frames.


For each frame calculate the periodogram
estimate of the power spectrum.
Apply the mel filterbank to the power
spectra, sum the energy in each filter.
Take the logarithm of all filterbank energies.

12

VOICE RECOGNITION
USING MATLAB (continued)

4/19/15
05:27:19 AM

Take the IFT of the log filterbank energies.


Keep 12-13 coefficients, discard the rest.
The coefficients will be a cluster for each
isolated signal having centroids.
The match is found by measuring the
apparent distance between these centroids
using a process called K-means.
Once the match is found, the Matlab
identifies the corresponding action to be
taken.

13

CHALLENGES

CHALLENGES
Unexpected output when
operated in different
environments.

SOLUTION
1.Introduction of proper noise
reduction techniques.
2.Enlargement of database.

Unexpected output even if


1.Write conditions for such case in
no input while the mic is on.
the Matlab code.
2.Ask for confirmation for some
important actions.
Slow response of the Matlab 1.Increase the ram memory.
running machine
2.Clean up the disk.
Isolation of words from
4/19/15
sentences.
05:27:19 AM

1.Developing an algorithm to detect


the silences between each words.

14

FEATURES
Normal voice operated
systems usually accepts only
the exact command word.
JARVIS allows the user to
communicate in a humanly
manner.

4/19/15
05:27:20 AM

Many applications and


devices can be interfaced to
the system and can be
controlled.

15

APPLICATIONS
JARVIS can act as an obedient
invisible Personal Assistant to
his Master.
User friendly.
Improved security than other
automation system.

4/19/15
05:27:20 AM

If interfaced with high quality


cameras and internet, it can
even run your office and

16

ESTIMATED COST
COMPONENTS

PRICE

ATMEGA8

200.00

LM 35
HUMIDITY SENSOR
GEAR MOTOR
IC L293D
RELAYS 5V- 2A

25.00
120.00
250.00
20.00
150.00

LED LAMP
BT HEADPHONES
2.1 SPEAKERS
4/19/15
05:27:20 AM TOTAL

170.00
1200.00
1300.00
3435.00

17

PROJECT STATUS
TASKS
The backbone of the
project was to
implement MFCC.
To control external
devices, ATmega8 is to
be interfaced.

4/19/15
05:27:20 AM

STATUS
COMPLETED AND TESTED

Microcontroller code
simulation completed.
Ready to be interfaced.

Enlargement of
databases.

To be done.

Adding additional
features like wireless
control of external
devices.

To be done.

18

4/19/15
05:27:20 AM

CONCLUSION
The project is very much alive
and the estimated completion
time is about 30 days.
The current position is about
45% of the total project in
which the most time
consuming tasks are
completed.
J.A.R.V.I.S is more than just a
voice recognition system but

19

4/19/15
05:27:21 AM

REFERENCE
1. Mondal, U.K; Mandal, J.K. Voice Recognition
Technique and Voice Processing, Recent
Trends in Information Technology (ICRTIT),
2011
2. Pankaj Pathak, Speech Recognition
Technology, International Journal on
Advanced Research Computer Science, 2012
3. Davis, S. Mermelstein, P. (1980)
Comparison of Parametric Representations
for Monosyllabic Word Recognition in
Continuously Spoken Sentences, in IEEE
Transactions on Acoustics, Speech, and
Signal Processing, Vol. 28 No. 4, pp. 357-366
4. X. Huang, A. Acero, and H. Hon. Spoken

23

FEASIBILITY CHART

COMPONENTS

PRICE DURABILITY FEASIBILITY

PIC 16F873

250.OO 36 MONTHS

LM 35
PIR SENSOR
GEAR MOTOR
IC L293D
RELAY 5V- 2A

25.OO
250.00
250.00
20.00
55.00

30.56%

48 MONTHS
60 MONTHS
60 MONTHS
60 MONTHS
18 MONTHS

94.79%
58.33%
58.33%
96.67%
69.44%

LED LAMP
170.00 60 MONTHS
BT HEADPHONES 1200.00 48 MONTHS
2.1 SPEAKERS
1300.00 60 MONTHS
4/19/15
05:27:22 AM TOTAL

71.67%
50.00%
56.67%
65.16%

0
24

FLOW CHART

NO
YE
S

YE
S
NO

NO

YE
S

4/19/15
05:27:22 AM

Potrebbero piacerti anche