2-Way Voice-Isolating Wireless Communication System

auditormineMobile - Wireless

Nov 24, 2013 (3 years and 10 months ago)

88 views

UNIVERSITY OF ILLINO
IS
-
URBANA CHAMPAIGN

2
-
Way Voice
-
Isolating Wireless
Communication System

ECE 445


Senior Design Laboratory


Alexander Joo, Frank Lam, Ryo Kondo

2
/
19
/2008





Page
2

of
13



I.

I
NTRODUCTION

1.

T
ITLE
:

2
-
WAY VOICE
-
ISOLATING WIRELESS C
OMMUNICATION SYST
EM

Our project’s goal is to allow easy wireless communication between two people in a noisy environment by
designing a headset system that can selectively isolate voice from a sum of voice and background noise,
and then wirelessly transmit the cleaned sign
al between headsets. We chose this project because we feel
that this is a common problem and because a solution like ours has not been implemented in this manner.

This is significantly different from noise
-
cancelling headsets as traditionally, these use de
structive
interference to remove noise from incoming signals; our goal is to use basic audio DSP to remove a known
noise pattern from an outgoing audio signal. If we can fit this into a small form factor, we feel that this can
become a marketable product.


2.

O
BJECTIVES

A.

B
ENEFITS TO
E
ND
U
SERS



Users

able to

communicate with each other in loud environments



User can communicate out of a loud environment to people elsewhere



Users can communicate wirelessly

with each other.

B.

F
EATURES
:



Have a wireless range of
~

50 fe
et (
10
0 feet

ideal)



Produce a significant gain in SNR (>20 dB


30 ideal)



Under

5 kg for prototype unit
/ Under 0.5 kg in potential production unit



2
-
way full
-
duplex communication



(near) Real
-
time communication (<100 ms delay)



Multiple non
-
interfering chan
nels (at least 10, preferred).

Page
3

of
13



II.

D
ESIGN

1.

B
LOCK DIAGRAM

DSP Chip
TI TMS
320
C
54
x
Transmitter
LINX TXM
-
900
-
HP
3
-
xxx
Voice
+
Background Noise
Background Noise
Speaker Side
Power Supply
1
Person
1
Speaker Side
Listener Side
Person
2
Speaker Side
Listener Side
Receiver
LINX RXM
-
900
-
HP
3
-
xxx
Digital to Analog Converter
Analog Devices AD
1955
Listener Side
Power Supply
2
Headphones

Page
4

of
13


2.

B
LOCK DESCRIPTIONS

Each person will carry the respective components for the speaking and listening. The Speaker Side
components take the sound, subtract the background noise
, and send the digital signal to the transmitter
where the signal is transmitted through the antenna. The listener will be carrying the Listener Side
components where the digital signal is transmitted from the antenna to the receiver, then processed in
the

DAC and converted to an analog signal. The analog audio signal is then outputted to the listener via
headphones. Both parties will be carrying components to speak and listen so that communication can go
both ways.

Speaker Side
:

1)

Power Supply 1:

The power s
upply will be a 5V voltage source powering the DSP chip and the
transmitter.

2)

DSP Chip:

The DSP Chip takes in two inputs. The first input is audio signal from a microphone that
takes in the voice from the speaker along with the background noise. The second
input is a second
audio signal from a second microphone that takes in only the background noise. The DSP chip then
converts both analog audio input signals into digital signals and then filters out the background noise
by subtracting the first audio signal

from the second signal. This digital signal is the filtered signal to
be sent to the listener. The background noise is subtracted out, so only the voice from the speaker is
left. After filtering, this digital signal is outputted to the transmitter.

3)

Transm
itter:

The transmitter takes the digital signal from the DSP chip and sends the signal to the
antenna to be transmitted wirelessly.

Listener Side
:

1)

Power Supply 2:

This power supply will be a 5V voltage source powering the receiver and the Digital
to Analog

Converter.

2)

Receiver:

The receiver takes the digital signal from the antenna and outputs the signal to the Digital
to Analog Converter.

3)

Digital to Analog Converter:

This device takes the digital signal from the receiver and converts it into
a coherent anal
og audio signal. It then outputs the analog signal to the headphones.

4)

Headphones:

The headphones take the analog signal from the Digital to Analog Converter and output
the signal for the person listening to hear.

3.

P
ERFORMANCE REQUIREME
NTS
(
GOALS
)

1.

Maximum r
ange of 100 feet

2.

Noise cancellation of 20 dB

3.

Under 2
kg

(not including power source)

4.


Power rating
within
FCC regulations for 900 MHz spectrum

(36 dBm)

Page
5

of
13



III.

V
ERIFICATION

1.

T
ESTING PROCEDURES

A.

R
ANGE

(
100

FEET
)

As we are using
digital
signals, w
e
expect
the signal

to be strong even at great distances
.
The spec
for our transmitter/receiver indicates up to a 1000 feet range. However, in order to test, the
range, we will conduct both indoor and outdoor tests to ensure that the digital signal remains
isolatable at 100
feet; and also test the maximum
usable
range.

The easiest way to do this is to
start at a close distance, and slowly walk the devices apart until the speaking
-
side signal can no
longer be processed by the listening
-
side.

B.

P
RODUCE A SIGNIFICANT

GAIN IN
SNR

(
>20

D
B

REDUCTION IN NOISE


30

D
B

IDEAL
)

To attain SNR, we would measure the absolute magnitude of noise from the noise
-
mic
rophone
and compare it to the noise component in the filtered signal.

To do so, we can use a known
quantity of noise (perhaps a sine
wave or an audio file)
, and measure the presence of it in the
output signal with the filter on and off. The logarithmic decease of the magnitude of this signal is
proportional to the noise reduction in decibels.

C.

U
NDER
10

KG FOR PROTOTYPE UNI
T
/

U
NDER
0.5

K
G IN POTENTIAL PRODU
CTION UNIT

This is a simple weight test; in our prototype unit, the combined weight of the headset and
accompanying hardware should not exceed this weight.

D.

2
-
WAY FULL
-
DUPLEX COMMUNICATION

Since each headset has a separate transmitter an
d receiver, we can have full
-
duplex
communications. We need to ensure that the frequencies do not overlap, and that a single pair
of frequencies can be uniquely selected without interfering with the other frequencies. To do
this, we can monitor each channe
l, and run a standard signal through one channel at a time. The
noise present in each channel should not exceed a minimum (experimentally defined) threshold.

E.

M
ULTIPLE NON
-
INTERFERING CHANNELS

(
AT LEAST
10,

PREFERRED
).

We need to ensure that the frequencies

do not overlap, and that a single pair of frequencies can
be uniquely selected without interfering with the other frequencies. To do this, we can monitor
each channel, and run a standard signal through one channel at a time. The noise present in each
chan
nel should not exceed a minimum (experimentally defined) threshold.

F.

(
NEAR
)

R
EAL
-
TIME COMMUNICATION
(<100

MS DELAY
)

The delay in the signal is predominantly due to the DSP processing steps

2.

T
OLERANCE
A
NALYSIS

The most important component of our system is the

ability to isolate voice from a noisy environment. We
shall refer to this as the
Voice
-
SNR with the voice component being the signal, and the background noise being
the “noise”. Initially, the noise might be
of
greater
magnitude
than the signal, so
must b
e able to perform an
Page
6

of
13


effective subtraction algorithm.
Sound magnitudes are measured in decibels, and the greatest
qualitative
measure is to

To achieve effective noise reduction, we
feel that
a 20 dB reduction is minimum requirement.
This would
effectively

reduce a “Loud Rock Concert” noise to the level of a subway train at 200’. This may facilitate much
easier conversation. Ideally, we would like to see 30 dB
, which would reduce the same “Loud Rock Concert”
noise to the level of “City Traffic (inside of a
car)”
.

To make a valid comparison, we feel that comparing the
proportions of the
magnitude

is best.

Environmental Noise

Weakest sound heard

0dB

Whisper Quiet Library

30dB

Normal conversation (3
-
5')

60
-
70dB

Telephone dial tone

80dB

City Traffic (inside

car)

85dB

Train whistle at 500', Truck Traffic

90dB

Subway train at 200'

95dB

Level at which sustained exposure may result
in hearing loss

90
-

95dB

Power mower at 3'

107dB

Snowmobile, Motorcycle

100dB

Power saw at 3'

110dB

Sandblasting, Loud Rock
Concert

115dB

Pain begins

125dB

Pneumatic riveter at 4'

125dB

Even short term exposure can cause
permanent damage
-

Loudest recommended
exposure
WITH

hearing protection

140dB

Jet engine at 100', Gun Blast

140dB

Death of hearing tissue

180dB

Loudest s
ound possible

194dB

Statistics for the Decibel (Loudness) Comparison Chart were taken from a study by Marshall Chasin , M.Sc., Aud(C),
FAAA, Centre for Human Performance & Health, Ontario, Canada. There were some conflicting readings and, in
many cases, a
uthors did not specify at what distance the readings were taken or what the musician was actually
playing. In general, when there were several readings, the higher one was chosen.

Chart obtained

from <

http://www.gcaudio.com/resources/howtos/loudness.html
>
.

IV.

C
OST AND
S
CHEDULE

1.

C
OST
A
NALYSIS

A.

L
ABOR

Names


Hourly
Wage

Hours

Multiplier

Total

Page
7

of
13


Joo, Alexander

$50.00

100

2.5

$12,500.00

Kondo, Ryo

$50.00

100

2.5

$12,500.00

Lam, Frank

$50.00

100

2.5

$12,500.00





Total Labor Cost

$37,500.00


B.

P
ARTS

Item

Descriptio
n

Cost/unit

Quantity

Cost

Texas Instrument

TMS320C54x DSPs

Digital signal processing
chip

$20.00

2

$40.00

Microchip
Technology

PIC16F877A

Programmable intelligent

computer chip

$10.00

2

$20.00

Linx

TXM
-
900
-
HP3
-
xxx

Radio frequency transmitter

$30.00

2

$60.00

Linx

RXM
-
900
-
HP3
-
xxx

Radio frequency receiver

$45.00

2

$90.00

Analog Devices

AD1955

Digital
-
analog converter

$7.00

2

$14.00

Microphone

Lapel microphone

$25.00

2

$50.00

Microphone

Omnidirectional
microphone

$25.00

2

$50.00

Earphones

Mono ear bud

ear phones

$15.00

2

$30.00

Other devices

Antenna, resistors etc.

$10.00

1

$10.00





Total Parts Cost

$364.00


C.

T
OTAL

Labor

$37,500.00

Parts

$364.00

Total

$37,864.00


2.

S
CHEDULE

Week

Task

Task

Leader

2/4

Project Pro
po
sal
-
Intro,

Project Proposal
-
Diagr
ams

Project Proposal
-
Cost & Schedule

Joo

Lam

Kondo

Page
8

of
13


2/11

Finalize specifications

Research hardware options

Research Labview

Begin Labview simulation

Sign up for design review

Begin preparation for design review

Lam

Lam

Joo/Kondo

Joo/Kondo

Lam

Kondo

2/18

C
omplete circuit diagram

Order hardware

Continue Labview simulation

Finish preparation for design review


Design Review

Lam

Lam

Joo/Kondo

Joo/Kondo/Lam

2/25

Complete Labview simulation

Verify hardware is operable

Begin hardware setup

Joo/Kondo

Lam

Lam

3/3

Begin software development for DSP

Test hardware set up

Joo/Kondo

Lam

3/10

Complete software development

Complete dsp circuit set up

Complete wireless circuit set up

Joo/Kondo

Kondo

Lam

3/17

Confirm modules work separately

(Spring Break)

Joo/Kondo/Lam

3/24

Finalize modules


Demo I

Joo/Kondo/Lam

3/31

Begin mating dsp and wireless
modules

Begin finalizing write up

Joo/Lam

Kondo

4/7

Cont. mating dsp and wireless
modules

Continue finalizing write up

Sign up for demo and presentation

Joo/Lam

Kondo

Lam

4/
14

Test complete set up

Continue write up

Joo/Lam

Kondo

4/21

Finalize complete set up

Finish write up


Demo II

Joo/Lam

Kondo

4/28

Final write up

Kondo


Page
9

of
13



Design Review


Block diagram

Page
10

of
13




Page
11

of
13



Page
12

of
13



The electrical schematic above is used to test the LINX HP
-
3

transmitter and receiver. The LINX HP
-
3 modules are
able to receive input voltages between 2.8
-
13 V, so a 5V DC source was selected based on convenience and
connected to the Vcc transmitter and receiver pins. The MODE is grounded on both modules, as this
selects
parallel input. The CS0, CS1, and CS2 are all grounded to select channel 1. The resulting output is shown on the
above waveform. Channel 1 of the oscilloscope waveform is the square wave from the function generator inputted
to the transmitter. This

will eventually be replaced by the audio signal to be transmitted. Channel 2 is the output
from the receiver after the square wave was transmitted wirelessly. As you can see, there is a roughly 15 μs delay
from the transmitted signal.


Page
13

of
13