close

Вход

Забыли?

вход по аккаунту

?

SSPS.2017.8071649

код для вставкиСкачать
2017 IEEE 3rd International Conference on Sensing, Signal Processing and Security (ICSSS)
Snappy Surrounding Alert for Android
1
1
•
Mr.R.Arunkumar, 2Ms.TharaniVimal
B.Tech- Information Technology , St. Peter’s college of Engineering and Technology,Chennai-600054
2
Assistant Professor , St. Peter’s college of Engineering and Technology,Chennai-600054
Email:1akrainarox@gmail.com.2tharanimadhan@yahoo.com
Abstract—Snappy surrounding alert is an android application
for headphone users who receive voice alert message about
their surroundings while they are listening to something using
their headphone. It alerts the users automatically by capturing
the situation in his surroundings. If the user wants to be
alerted about the surrounding when he uses the headphone he
should first enable the Snappy surrounding alert app and the
GPS . After enabling, the application starts to work and alerts
the user under three different scenarios. First is a location
based alert, where the GPS continuously collects the current
latitude and longitude of the user and alerts him through his
headphone speaker with a voice message when he is in careful
zone such as railway station, highways, railway track. Second
is a high pitch alert, which is triggered when the headphones
microphone receives a high pitch sound such as vehicle horn
from the surrounding. Third is a calling alert that will be
raised when someone from nearest surrounding calls the
mobile user by his name or other general context such as sir,
excuse me, madam and so on. The alert message for all the
scenarios are played in the format of voice. Thus this
application can continuously monitor the surroundings and
warn the mobile headphone users. It reduces many accidents
and introduces a system which solves the problem for
headphone users.
Keywords: GPS , Speech To Text, Text To Speech, Speech To Speech.
user has reached a dangerous zone like railway, road or when
someone calls us from a certain distance.
This application provides a useful tool to the mobile
headphone users when they are crossing the roads and
railways in their daily life. Initially, the headphone
connectivity is checked. Then the user’s mobile device sends
GPS coordinates as input to the application which is passed to
the database for checking the location. Secondly, a headphone
user receives voice message when a high pitch noise is picked
up from the surrounding. The mobile will also vibrate to
double alert the user. Thirdly, the user is alerted when
someone calls him by name or general context. These three
scenarios are named as,
• Location based alert
• High pitch based alert
• Context based alert
These alerts can be run in parallel in all three different
scenarios. The user can enable or disable each scenario.
Two techniques are used here, Automatic Speech
Recognition (ASR) and Text-To-Speech (TTS). This feature,
along with the power of Pocket Sphinx speech recognition
engine enables a great choice for incorporation of the
proposed architecture. Thus it helps to develop an application
that makes the headphone users to be a more attentive person.
1. INTRODUCTION
2. LITERATURE SURVEY
A mobile application or a mobile app is software
designed to run on mobile devices such as smart phone’s and
tablet computers. Most such devices are sold with several apps
bundled as pre-installed software’s, such as a web browser,
email client, calendar or an app for buying music, tickets,
shopping and many more.
Most of the people in this world love to listen to
music as they are travelling or walking or whenever they don't
have a companion to talk with. These people sometimes don't
have any concentration about their surroundings. This may
lead to major accidents or minor injuries which in most of the
scenarios end up with loss of life. Now, this android
application called snappy surrounding alert helps the
headphone users by raising the alert under different situations
like presence of horn sounds in the surroundings or when the
In [1], the Speech is acquired at run time through a
microphone and is processed to recognize the uttered word.
The recognized word is stored as a text in a file. This is
developed on android platform using Android Studio.The
Speech-to-Text system directly acquires and converts speech
to text. It is implemented using a third party library called as
Pocket Sphinx. Pocket sphinx is a lightweight speech
recognition engine. This Automatic Speech Recognition
(ASR) engine is specifically tuned for handheld and mobile
devices. The Speech-to-Speech translation is also used here. It
allows the user to record and communicate between a Sinhala
and Tamil, crossing the language barrier. Thus the users can
pass the messages to another native language person. Speech
recognition for voice uses a technique based on hidden
978-1-5090-4929-5©2017 IEEE
509
2017 IEEE 3rd International Conference on Sensing, Signal Processing and Security (ICSSS)
Markov model (HMM - Hidden Markov Model). It is
currently the most successful and most flexible approach for
speech recognition. This speech recognizer also works in the
offline mode and allows much faster data processing. Another
advantage is the much larger databases that can be used.
Our application uses this method to recognize Speech
without the Internet. Then the user’s voice input is converted
to text using Pocket Sphinx speech recognition library.
Finally, the converted text is read aloud naturally in the
calling alert module.
The author in [2] has proposed Text-To-Speech
(TTS) synthesis system that is based on phonetic
concatenation for input text. The input text is first converted
into phonetic transcription using the Letter-to-Sound rules.
For synthesis of a new speech, TTS system selects the
recorded phoneme units from database and modifies the
duration according to the rule based on spelling using Time
Domain Pitch Synchronous Overlap Add. The PUs that is
modified is then concatenated by synchronizing pitch-periods
at juncture and by smoothening the transitions in order to
remove the audible discontinuity and spectral mismatches.
The pitch of PUs is kept to original neutral sounding.
This paper describes a simple, flexible and efficient
procedure that uses much lesser memory spaces. This method
is used in our proposed system to convert Text to Speech in
calling alert module. With this implementation our system
was found to be running perfectly and flawlessly.
In [3], the author has discussed about the Personal
Safety app developed by him. Technology and the
advancements in technologies have been utilized here. A
smart phone is used here to provide an interactive solution
using the speech technology. This application helps the user
by asking him to utter a keyword, which the application
recognizes as a call for emergency and sends out an SMS to
the preset list of contacts as set by the user. The system also
ensures that their call for help reaches the concerned people
and an immediate response is generated. This type of
application can enhance the security of general public
especially women and people with certain chronic health
conditions such as heart patients, people in danger or
emergency situation. The smart phone based app uses the
open source Android API’s and developer tools.
Different approaches were used by different
researchers for recognition of voice. Various voice recognition
systems were implemented in different fields. The voice
recognition process has several well defined steps as follows.
• Divide the process into evenly spaced blocks.
• Each block is processed separately and produces a
string.
To always stay connected with the loved ones and friends
where ever they are people prefer mobile phones. This means
not only talking but for other purposes like e-mailing, texting
and so on. In paper [4], the author introduces a new android
application which will recognize voice commands. This
application uses desired equipment and a service which can be
controlled through voice without touching the screen of the
android smart phone. Reliable speech recognition is a hard
problem, requiring a combination of many techniques.
However modern methods have been able to achieve an
impressive degree of accuracy. At the end, this paper also
introduced some new set of features for voice actions on
Android platform smart mobile phones. The device proposed
here is an interactive android smart phone, which is capable of
recognizing spoken words. Voice recognition is used as a
matching context for the headphone user, to convey this
information to user as voice message in order to alert the
headphone users to someone calling in emergency time.
The author in [5] proposes optimized approach for
implementing the Global Positioning System (GPS). The
Global Positioning System (GPS) is a reliable, available and
accurate positioning technology that we can be used from
anywhere across the globe. GPS is the most effective
navigation tool that can be used for a wide variety of
application areas like air, sea, land, and space navigation. The
Google Location Services API, part of Google Play Services
provides a more powerful, high-level framework that
automatically handles location providers, user movement and
location accuracy. It also handles location update scheduling
based on power consumption parameters provided.
In GPS tracker, there are two ways to get the
coordinates. They are by using GPS PROVIDER and
NETWORK PROVIDER. GPS PROVIDER is the most
accurate method, which uses the built-in GPS receiver of the
device. GPS is a system of satellites in orbit. It provides
location information from almost anywhere on earth. It can
sometimes take a while to fix a GPS location when we move
faster in outdoors. On the other hand, NETWORK
PROVIDER method determines a device's location using data
collected from the surrounding cell towers and WiFi access
points. While it is not as accurate as the GPS method, it
provides a quite adequate representation of the device's
location.
In our application, we have used both the options of
getting the GPS location. The NETWORK PROVIDER is
used when there is no need for the user to turn on the mobile
data connection. The GPS PROVIDER is used when the
mobile data is turned on and an accurate position is required.
We also get the coordinates at a faster rate in this method.
978-1-5090-4929-5©2017 IEEE
510
2017 IEEE 3rd International Conference on Sensing, Signal Processing and Security (ICSSS)
3. SYSTEM ARCHITECTURE
The system initially checks for the connectivity of
the headset. Later it continuously gathers the latitude and
longitude of the user’s location. It then checks the database for
careful zone and if it be so then it will raise a voice alert like
“You have reached the careful zone. Please take out your
headphone”. The user can also add new locations in the
repository so that the next time the user goes there trigger will
be raised.
The app also continuously senses the sound
frequency in the surroundings and compares it with a user
defined or standard given threshold value, for a high pitch
sound in the surrounding. If the sound is above threshold then
a voice alert “You have a High pitch noise present in the
surrounding. Check your surrounding “, is raised.
Also the system continuously senses the surroundings for
the common text or username like hello, John etc. This is
compared with a username and common text in the database
and raises a calling alert like “Someone is calling you. Please
check out”, when someone calls the user. This is achieved by
using Speech to Text and Text to Speech conversion.
• Location Based Alert
• High Pitch Alert
• Calling Alert
A. Headphone Connectivity Checking
This module is used to help the user to install the
application in their mobile phones. Once the user installs the
application it asks the user to connect their headphone. The
headphone contains a pair of small speaker which can fit to
human ears and a microphone.
If headphone is connected with mobile then the user
is taken to the main activity of the application which contains
three modules and each module can be enabled by user.
Otherwise, the application display the message on the screen
as “No Headset is detected please connect and try again”. This
constraint is most important for this project. Without headset
this application will not work. It also intimates the user about
the status of the app like Enabled/Disabled, whenever the
headset is plugged into the mobile phone.
Figure 2. Headphone Connectivity Checking
B.
Figure-1. System Architecture
4. MODULE DESCRIPTION
The modules in our Snappy Surrounding Alert for Android are
as listed.
• Headphone Connectivity Checking
Location based alert
The second module uses the GPS Location and
SQLite database. After the user enables this module, it asks
the user to enable the GPS. Then, the GPS device searches for
the user’s current mobile coordinates for this application. Here
we have used the concept of multithreading because this
application has totally three modules and all the three modules
have to work at same time. In this module the threads are first
initialized. Then the GPS continuously searches for every five
seconds and the thread is integrated with one more function
that does the checking of headset connectivity within the same
time interval.
Another component in this module is database which
consists of table with two columns named as place and
978-1-5090-4929-5©2017 IEEE
511
2017 IEEE 3rd International Conference on Sensing, Signal Processing and Security (ICSSS)
coordinates. Interestingly, this application has a functionality
of adding additional coordinates of careful zone by the user.
Apart from the user input, the database contains default set of
particular coordinates. Every time the user position is changed
as per the GPS position, these coordinates are searched in the
database. If the search result is true, it builds a voice message
with concatenation of respected place of searched coordinates.
After this the voice message will be delivered to the user
using Text To Speech engine (e.g., "You are near Central
Railway Station, can you take out your headphone").
A database is maintained in this system that contains the
location coordinates and their respective name. This can be
updated manually by the user, for other locations where he
wants to be alerted. Now in the location based alert the GPS is
initialized and returns the coordinates of user’s present
location. The system or the mobile application continuously
monitors, by searching for a matching coordinate of the
current GPS location in the database. If the coordinates match
then application will trigger a voice alert using TTS.
Enter the location name and
coordinates to be added to the
database.
ratio. This is now compared with the fixed threshold and if it
is greater than threshold level then instantly a voice alert is
raised. This voice alert helps the user to divert his attention
and look back to what is happening in the surrounding.
The application here recognizes the high pitch sounds
such as vehicle horns, building construction noise and so on.
A new thread is started after this module is enabled by the
user. This thread handles a lot of functionalities within the
time interval. The recorder continuously gathers surrounding
sounds through the microphone. These recorded sounds are
converted to amplitude signals. Sound consists of more
number of amplitude data points which are not in a neatly
arranged manner. So, now we apply the EMA (Exponential
Moving Average) filter which smoothen the amplitude data
points and finally gets the highest amplitude point from the
particular interval recorded sounds. Immediately the resultant
amplitude is converted into dB (decibel) and this dB is
compared with user-defined threshold level and it is measured
in dB scale. If converted dB is greater than threshold level
then the mobile device indicate this alert to user via voice
message (e.g., "High Pitch sound is present can you check the
surroundings").
This enables the app to rise
Location Alert offline.
C.
User can adjust the
threshold level based
on surrounding
sounds.
High pitch alert
In the High pitch based scenario, there are three tasks
to be performed. They are,
• Sound Recognition
• Amplitude Measurement
• Fixing Threshold
• TTS
In this scenario, the system frequently records the sound near
the user. This sound is converted to amplitude signals. It is
analyzed to determine the peak amplitude. This peak
amplitude is converted to dB by using the noise-to-signal
D.
Calling alert
The last scenario in our application is the context
based alert. There are three sub divisions. They are,
• Speech Recognition.
• Speech-To-Text engine
• Grammar file
• Text-To-Speech engine
This module alerts the user when someone is calling a user
from behind him and he is unaware of it. Voice around the
978-1-5090-4929-5©2017 IEEE
512
2017 IEEE 3rd International Conference on Sensing, Signal Processing and Security (ICSSS)
user is recognized and recorded often. This is converted to
string using the Speech-To-Text engine. The converted strings
are presented in a grammar file that triggers a voice message.
This module of our application uses the Automatic
Speech Recognizer (ASR). Pocket sphinx has offered ASR
engine which comprises of a dictionary and a grammar files
that works offline. Once the user has enabled this module, the
system creates a thread when starting a speech recognizer. The
Recognizer builds with dictionary and acoustic sounds model
and it continuous searches for a particular context nearest to
the headphone users. These contexts are like sir, mam, excuse
me, and it might also be the name of the user.
Headphone user can add the custom context in the
grammar file.So, if the system finds a person calling then this
context is captured by the microphone and is compared with
the database. Then this information reaches out to human ears
in the form of voice message (e.g., "Someone is calling you" )
using Text To Speech.
These are all common
context to all users
who are using this app
•
User can know exactly how far he is from the careful
zones.
• User can get the information of which vehicle horn it is.
Because the vehicle may be a car, bus, train, and so on.
• Intimation about the failure of the app to the headphone
users.
7. REFERENCES
[1]. Layansan R., Aravinth S. Sarmilan S, Banujan C, and G.
Fernando, “Android Speech-to-speech Translation System For
Sinhala”, International Journal of Scientific & Engineering
Research, Volume 6, Issue 10, October-2015
[2]. Mahwash Ahmed, ShibliNisar, “Text to Speech
Conversion with Phonematic Concatenation”, International
Journal of Scientific Engineering and TechnologyVolume
No.3, Issue No.2, pp : 193 – 197 (February 2014)
[3].
NikamTanmay,
GangurdeVidya,
PatilChaitalee,
VidyaKawtikwar, “Personal Safety App using Speech
Recognition API”, International Journal of Application or
Innovation in Engineering & Management (IJAIEM)Volume
4, Issue 3, March, 2015
[4]. RenuTarneja, Huma Khan, Prof. R. A. Agrawal, Prof.
Dinesh. D. Patil, “Voice Commands Control Recognition
Android Apps”, International Journal of Engineering Research
and General Science Volume 3, Issue 2, March-April, 2015
[5]. Smriti Sharma, Rajesh Kumar, PawanBhadana,
“Terrestrial GPS Positioning System ” IJRET: International
Journal of Research in Engineering and TechnologyVolume 2,
Issue 4, April, 2013
[6].https://developer.android.com/guide/topics/location/strateg
ies.html
5. CONCLUSION
[7].https://github.com/HelloSpoon/PocketSphinx_Sample
The Snappy Surrounding Alert application can
reduce the accident in highways, railway track, etc. It always
keeps the headphone users informed about their surroundings.
Also it makes the headphone users be more attentive during
listening to something in the headphone.
[8].
http://www.androidhive.info/2012/01/android-text-tospeech-tutorial/
[9].http://stackoverflow.com/questions/16072185/decibels-inandroid
[10].http://www.informit.com/articles/article.aspx?p=2731932
6. FUTURE ENHANCEMENT
In future enhancements we have planned to include
AI along with the current technologies like GPS, ARS, TTS.
After integrating AI with this application it can include the
following features.
978-1-5090-4929-5©2017 IEEE
513
Документ
Категория
Без категории
Просмотров
2
Размер файла
324 Кб
Теги
2017, ssps, 8071649
1/--страниц
Пожаловаться на содержимое документа