The Rybená Solution
The team’s ultimate goal is to enable communication
between a blind and a deaf in the future through the
use of a cell phone. Since this project requires knowledge
with which the team is not familiar and for the purpose
of obtaining intermediary results, a simplified model
was developed to allow hearing individuals to talk
to deaf individuals through a Telco operator. The
speech is processed through a Via Voice acknowledger
where it is converted into text and then sent to the
LIBRAS word interpretation service. To make it possible,
the system establishes a connection with a Prevalence
Data Model (PDM) (Villela 2002, Wuestefeld 2001, Wuestefeld
2002) where a LIBRAS set of animations is stored.
For each recognized word, a corresponding LIBRAS animation
is transmitted to the client’s cell phone. In
case one word does not have a correspondent symbol
at the PDM, images of all its characters are transmitted
letter by letter (spelling module) to the mobile device
(Fig.1).
Figure 1. Rybena Architecture
The architecture may be currently broken down into
five work forces, the first three of which are in
charge of developing the infrastructure required for
communication services, the fourth will validate the solution’s concept, and the fifth will provide
a universal Portuguese/LIBRAS translator.
Setting up the LIBRAS library
The images were generated according to restrictions on readability and memory costs.
Among the social inclusion programs developed by the
DFJUG in partnership with the CTS, the deaf community
was contemplated with a trainning program in the Java
language. This activity resulted in the need to create
a LIBRAS terminology to represent programming concepts
and, in particular, object-oriented programming.
Words being produced based on the INES dictionary.
INES, Brazilian Ministry of Education, provides a
Brazilian LIBRAS dictionary in the form of AVI-formatted
images, currently consisting of a little more than
two thousand words. The process gained agility and confidence
since animations are currently developed based on the pattern proposed
by INES.
Like the hearing community, the deaf community also
develops regionalisms (slang). The authors noted that
the deaf involved in image development often questioned
the signs presented by the INES. The authors’
decision to maintain the pattern for convenience purposes
has caused some stress among deaf team members. Such
problems will be dealt with in the future to avoid
further conflicts.
JMS service implementation
Access to heterogeneous clients (PC/cell, cell/cell,
cell/PDA, and others) was defined as the desired characteristic
for the Rybena project. Therefore, the team decided
to use a JMS server (JMS, 2004; Monson-Haefel &
Chappell 2001) to provide the messaging infrastructure
enabling client integration within this context.
Such decision was based on the fact that the JMS technology
was the technology that best met the solution needs.
In addition to the messaging service, the JMS technology
will also allow Rybena to integrate with other services
such as databases, third-party services and different
computer platforms. The JMS technology also rules
out the need for services of a telephone operator
since it requires only a TCP/IP connection between
the (heterogeneous) clients and the server.
The Peer-to-Peer connection will be used as required,
always passing through the JMS server, thus allowing
a PC to chat with a PDA, for instance.
An immediate advantage is the total control over Rybena’s
interface with the JMS server, enabling, for instance,
that critical message services are signaled until
the (deaf) client removes it from the informational
context with which it is associated.
The JMS server may provide the Rybena project with
an architecture that supports a set of plug-ins directly
connected with the services available.
Universal translator
Figure 6. Rybena Mobile to Person
While discussing the possibilities raised by the
Rybena project to enable communication between a deaf
and a speaking individual holding a Rybena-enabled
cell phone, the team decided to use the cell phone
as a translator. This would work as follows:
1.The speaking person turns the cell phone
screen to the deaf and speaks on the back of the
equipment (in the test with Sony Ericsson P800 the
cell phone was able to pick up the voice from the
back);
2.The voice is sent to the Telco operator,
which, upon recognizing the equipment as one used
to communicate with deaf individuals, transfers
the voice to the Voice Transformation Module (VTM).
After recognizing the voice, the VTM selects the
indexes of the proper LIBRAS gestures and sends
them to the cell phone;
3.The cell phone then uses the indexes to
retrieve the appropriate sequence of LIBRAS gestures
from its storage and displays the resulting gesture
composition on the screen;
4.The deaf would be able to understand the
message if he/she is conversant in LIBRAS.
Obviously this is only a one-way communication but
it can be upgraded to a two-way communication should
the deaf be able to tap the LIBRAS code into the cell
phone and the Telco operator send the corresponding
sound to the cell phone. This would require that,
upon picking up the cell phone, the speaking person
would tap a command to release the voice to the cell
phone loudspeaker.
