You are on page 1of 19

Disability and Rehabilitation: Assistive Technology

ISSN: 1748-3107 (Print) 1748-3115 (Online) Journal homepage: http://www.tandfonline.com/loi/iidt20

Assistive technologies for hearing, and speaking


impaired people: a survey

Ghassan Kbar, Akshay Bhatia, Mustufa Haider Abidi & Ibraheem Alsharawy

To cite this article: Ghassan Kbar, Akshay Bhatia, Mustufa Haider Abidi & Ibraheem Alsharawy
(2016): Assistive technologies for hearing, and speaking impaired people: a survey, Disability
and Rehabilitation: Assistive Technology, DOI: 10.3109/17483107.2015.1129456

To link to this article: http://dx.doi.org/10.3109/17483107.2015.1129456

View supplementary material

Published online: 17 Feb 2016.

Submit your article to this journal

Article views: 4

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=iidt20

Download by: [Gazi University] Date: 19 February 2016, At: 22:00


DISABILITY AND REHABILITATION: ASSISTIVE TECHNOLOGY, 2016
http://dx.doi.org/10.3109/17483107.2015.1129456

REVIEW ARTICLE

Assistive technologies for hearing, and speaking impaired people: a survey


Ghassan Kbara, Akshay Bhatiaa, Mustufa Haider Abidib and Ibraheem Alsharawyc
a
Riyadh Techno Valley, King Saud University, Riyadh, Saudi Arabia; bFARCAMT, Advanced Manufacturing Institute, King Saud University,
Riyadh, Saudi Arabia; cIndustrial Engineering Department, King Saud University, Riyadh, Saudi Arabia

ABSTRACT ARTICLE HISTORY


This study presents a novel method for evaluating the scientific research papers in the field of Received 8 June 2015
assistive technologies pertaining to different impairment conditions. The objectives are to Revised 2 December 2015
understand the technologies used for addressing the needs of PWD by identifying relevant criteria Accepted 4 December 2015
for the assessment, explore the implications of these technologies in their lives and identify the Published online 12 January
gaps among certain technologies in assisting PWD. In this article, we reviewed around 40 research 2016
scientific papers in relation to the technologies used to assist PWD in their daily activities. A novel
KEYWORDS
quantitative assessment methodology based on Multi-weighted Scoring Model (MWSM) has been Assistive technology; hear-
developed. It is based on the judgement of clinical experts according to thirteen well-defined ing impaired; multi-
Downloaded by [Gazi University] at 22:00 19 February 2016

criteria. The proposed method is useful because it assesses the scientific studies related to PWD weighted scoring method;
qualitatively according to efficient research coverage, as well as quantitatively in order to have person with disabilities;
good comparative judgment. Moreover, this method recognizes the research gap or areas which quantitative survey; speak-
need further investigation and identifies the research papers that have good coverage of the ing impaired
respective criteria.

ä IMPLICATIONS FOR REHABILITATION


 Human computer interface (HCI) solutions are critical for addressing the main issues facing
people with disabilities (PWD) in their life.
 Assessment of scientific research papers according to well-defined criteria that address PWD
needs would assist in verifying their suitability for PWDs.
 Novel quantitative assessment methodology is used for assessing these research papers using
judgment of experienced researchers according to 13 well-defined criteria that have been
weighted according to relevancy to different impairment groups.
 Identifying research papers that have good coverage of defined criteria and knowing the
research area that needs further investigation by researchers and developers, would ultimately
address the rehabilitation needs for PWD.

Introduction This number is growing each year, which becomes a


greater concern to researchers, governments and deci-
People born with disabilities or who become impaired at
sion makers. The lack of service support can make
some point of their life will experience many difficulties,
individuals with handicaps overly dependent on their
which can impact their quality of life in comparison with
families, which prevents them from being economically
individuals without disabilities. The World Health
active and socially included. In Kingdom of Saudi Arabia
Organization [1] predicted that the number of people
(KSA), it is estimated that 3.73% of the population has a
who will reach the age of 60 will be around 2 billion by
functional disability including cognitive or physical
2020 and that the majority of them will have disabilities
impairment.[2] More than 74.2% of this group experi-
and face difficulties in using computer interfaces. In ence physical, visual or hearing disabilities, but still can
addition, older adults might also suffer from hearing loss participate in knowledge creation.
as well as partial visual and motor impairment. Currently, An inaccessible environment creates additional bar-
over a billion people including children (or about 15% of riers for people with disabilities by making it hard to
the world’s population) are estimated to be living with function independently and to cope with the surround-
disabilities and 3.8% of these have severe disabilities.[1] ing objects. The legislation and policy in regards to

CONTACT Ghassan Ali Kbar gkbar@ksu.edu.sa Innovation Tower, 5th Floor Riyadh Techno Valley, King Saud University, Riyadh, 12373, Saudi Arabia
Supplemental data for this article can be accessed http://dx.doi.org/10.3109/17483107.2015.1129456.
ß 2016 Taylor & Francis
2 G. KBAR ET AL.

supporting the requirements of people with disabilities well as evaluating existing solutions with commercial
has been changed in many countries to improve health products. The research gap and the limitations of the
conditions and outcomes for persons with disabilities.[1] various papers have also been identified in this review,
Almost all people with disabilities need assistance to but to have a thorough evaluation across a range of
participate in social and economic life, and this support articles that have relevant knowledge to offer in the
typically comes from family members, caregivers and/or pursuit of improving technology for the targeted
technology solutions that facilitate interaction with the disability groups a comprehensive literature review or
environment through relevant Human Computer systematic review is required which can be done in the
Interaction (HCI). future. The review covers selective research articles from
Over the past decade, ubiquitous computing-based multiple academic disciplines most notably science,
technologies have emerged that provide a foundation technology and health. The purpose of this literature
for researchers and developers to provide technologies review is to understand the technologies used for
that assist individuals with overcoming or compensating addressing the needs for people with disabilities, the
for their disabilities and improving their quality of life at implication of each technological solution to their life
home, work and all everyday settings. HCI is one and to identify the gap or lack of coverage of certain
particular technology that can assist people with technologies that can assist in providing relevant solu-
disabilities with effective user interaction support. tions in order to address the needs of these people. We
People with disabilities often require supervision and
Downloaded by [Gazi University] at 22:00 19 February 2016

use both a qualitative approach that focuses on the


helps assistance to perform their daily activities, a role contribution of each technology to address the needs of
which family members typically assume. Assistive tech- people with disabilities, and a quantitative approach to
nology helps Persons with Disabilities (PWD) to be assess these papers using a Multi-weighted Scoring
socially included and to become or remain independent. Model according to a set of well-defined criteria. The
Information and communication technologies have a qualitative and quantitative approaches allow us to
key role in providing accessibility and inclusion for PWD. assess the impact of the research in providing relevant
Ongoing advances in Information and Communication solutions to the impairment conditions of PWD. The
Technologies (ICT) are increasing the scope for criteria used for the assessment of different impairment
IT-desktop and mobile assistive technologies to group has been generalized with 13 criteria but with
improve participation, independence, safety and different weights according to their relevancy to a
improved quality of life of for PWD in the workplace. particular group, where the very relevant criteria would
Developing relevant solutions will help PWD in KSA and have a weight of 3 and the non-relevant criteria would
around the world to exercise their creativity and have the Weight of 0 as will be described later in the
contribute to the workplace, instead of being idle and analysis section.
frustrated.[3] In this article, we restrict our attention to population
There can be a number of different approaches to groups that have physical impairments, rather than
carry out a research review. According to the Grant and cognitive impairments. We classify impairment condi-
Booth,[4] an analytical framework, including the search, tions into eight groups: Speaking Impairment (SI),
appraisal, synthesis and analysis (SASA), is required to Hearing Impairment (HI), Visual Impairment (VI), Motor
select the type of the review method. Therefore, among Impairment (MI), Speaking and Motor Impairment (SMI),
the fourteen analyzed review methods, the authors of Visual and Motor Impairment (VMI), Hearing and
the current article has adopted the mixed studies review
Speaking Impairment (HSI) and Hearing and Motor
(or the mixed methods review) method. Although, a
Impairment (HMI). In this article, we review papers for
comprehensive review is not carried out in this paper,
the impairment groups of SI, HI and SHI while other
but it provides a mixed review of scientific contributions
impairments groups related to VI, VMI and MI will be
to the topic of assistive technology for PWDs. Actually,
covered in separate articles.
the focus in this work is on the particular area (selective
or purposive research). Furthermore, a formal quality
assessment and the appraisal instrument and process Assessing research papers using
have been adopted as the appraisal criteria since, a MULTI-weighted scoring model
systematic process is needed. A narrative commentary,
tabular and Graphic, quantitative and qualitative studies In order to assess the suitability of each paper for
are used which defines the synthesis criteria. The providing a relevant solution for assisting PWD users, we
quantitative assessment helps in evaluating research provide qualitative and quantitative evaluations that are
projects that are intended to be complete products as based on well-defined criteria. Hence, we used a Multi-
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 3

weighted Scoring Model (MWSM) to assess and evaluate Motor Speaking Impairments (MSI), Hearing Speaking
approximately 40 research papers. Impairment (HSI), Motor Hearing Impairment (MHI) and
A novel assessment model based on nine steps has Motor Visual Impairment (MVI), where the large number
been used to assess the research papers quantitatively of HI, HSI and SI papers are further assessed in this
using MWSM. The different steps are described in article.
Figure 1 of Appendix 1 in the supplement document.
A brief description of the activities at each step is Assess research papers using guideline and
described below. criteria
Evaluators followed the well-defined guidelines for
Identify the criteria for assessing the technologies
assessment and the results thus obtained are presented
and their additive weight
in the following sections.
The technology performance criteria (Ctx) used for
assessing the required technologies needed for PWD Review of the assessment results
are determined and rated according to the importance
Multiple meetings are conducted to reduce the variance
level as described in Table 1 of Appendix 1 of the
between different evaluator results.
supplement document.
Downloaded by [Gazi University] at 22:00 19 February 2016

Assessment analysis MWSM


Identify the technologies relevant for PWD
The results for assessing the identified research papers
The lists of technologies that can be used to assist PWDs
for HI, HSI and SI are analyzed and presented in the
in interacting with a computer system are defined (see
following sections.
Table 2 of Appendix 1 of the Supplementary material).

Identify the gaps in research coverage


Assess technologies using MWSM applied to Step
1 criteria The results found in steps 6, 7 and 8 are analyzed to
determine the lack of research coverage for the average
The MWSM is applied to the identified technologies
criteria that have weights above our specified thresholds.
which determine their score based on the technology
The above technique for the assessment model is
performance criteria (Ctx) as described in Tables 3–13 of
similar to the Assistive Technology Assessment (ATA)
Appendix 1 of the Supplementary material. We note that
process model proposed by Federici and Scherer.[5]
all of the technology ratings were performed by the
During the phase 1 of the ATA model (steps a–d), a
authors. In order to ensure that the ratings are consistent
quantitative assessment technique based on 13 criteria
with the needs of PWDs, the committee consulted with
involving the use of technology and the author’s insight
experts who have specialization in PWD accessibility for and experience was proposed. The rating of these
more than 20 years. The ratings represent the result of criteria have been verified by experts who have partial
extensive discussion within this committee. impairment condition and authors’ committee team, and
they have also been verified through a set of question-
Identify the technological criteria for assessing naire used to collect data from PWD users to know the
research papers importance of technologies that can help them in doing
The MWSM is applied to determine the weight of the their job easily. A multidisciplinary team guided by the
technological criteria (Cx) that can be used for assessing committee collected the data from PWD users (see
the papers as described in Table 3. The scores are Appendix 2 of the Supplementary material). At phase 2
elicited from three experienced researchers, who fol- of the ATA model (step e), the multidisciplinary and
lowed a clear methodology for the assessment through Committee team has evaluated the result associated
checking for evidence of supporting technologies with technologies’ needs for PWD and ranked them
associated with the criteria in the research paper, to numerically to determine the priority of these technol-
report the average scores. ogies for the specific group of PWD and then link them
to technological criteria that can be used to assess the
technological solutions (see Tables 3–12 of Appendix 1
Conduct literature review and classify papers
of Supplementary material). The results from survey
Evaluators conduct a literature review to identify papers were consistent with experts opinion as described in
and classify them according to topics Visual Impairment Appendix 1 and Appendix 2 of the Supplementary
(VI), Hearing Impairment (HI), Speaking Impairment (SI), material.
4 G. KBAR ET AL.

Table 1. Assessment criteria description.


Criterion Explanation Relevance to PWD
Essential system input C1 Speech to text Helps others to communicate with
PWD; Helps in controlling
interfaces
C2 Keyboard or touch screen or Helps enter messages and control
remote interfaces
C3 Gesture control using sensor or Helps observe the environment
camera
Supporting system input C4 Tracking location using sensors or Helps in determining location of
WIFI or RFID PWD for decision-making
C5 Sensors for behavior Helps in tracking PWD movement
for analysis and prediction
Essential system output C6 Text or digital mapping to speech Helps PWD communicate with
other people
C7 Magnification of font and window Helps in reading messages and
size seeing pictures
C8 Color control Helps in seeing frames and
interfaces
C9 Display system Helps interact with systems
Supporting system output C10 Auto adjustable Speaker and Helps with hearing and speaking
Volume control
C11 Haptic or visual feedback (vibra- Help with sensory interfaces
tion or flashing)
Downloaded by [Gazi University] at 22:00 19 February 2016

Supporting system processing C12 Mobile support Helps in carrying smart system for
interfacing with environment
C13 Noise Filtering Improves hearing quality

Table 2. Criteria weightage.


Substitute/Very important Assist/Important Could assist/Less important Not applicable
Impairment type (Weight ¼ 3) (Weight ¼ 2) (Weight ¼ 1) (Weight ¼ 0)
HI C1: speech to text C2: keyboard, mouse C3: gesture control C10: speaker control
C6: text to speech C5: sensors behavior C4: tracking location
C9: display system C13: noise filtering C7: magnification
C11: vibrate, flash C8: color control
C12: mobility
Partial SI (PSI) C2: keyboard, mouse C1: speech to text C8: color control
Person can hardly C3: gesture control C4: tracking location
speak with difficulties C6: text to speech C5: sensors behavior
to explain C9: display system C7: magnification
C11: vibrate, flash C10: speaker control
C13: noise filtering C12: mobility
Full SI (FSI) C2: keyboard, mouse C4: tracking location C8: color control C1: speech to text
Non-speaking person C3: gesture control C5: sensors behavior C13: noise filtering
C6: text to speech C7: magnification
C9: display system C12: mobility
C10: speaker control
C11: vibrate, flash
HSI C1: speech to text C5: sensors behavior C4: tracking location C6: text to speech
C2: keyboard, mouse C7: magnification C8: color control C10: speaker control
C3: gesture control C11: vibrate, flash C13: noise filtering
C9: display system
C12: mobility

The scientific databases including the ScienceDirect, disabilities’’ is a very broad term, and it covers a large
Scopus, IEEE Xplore Digital Library, Google Scholar, number of different research domains. Therefore, this
Emerald and Springer Link were used to collect the survey mainly focuses on the research papers involving
research studies. The keywords that were used to search the assistive technologies pertaining to hearing, speak-
for citations are as follows: assistive technology, hearing ing and motor impairments that have been published
impairment, motor impairment, speaking impairment, a for last 10 years. Other impairments and their combin-
combination of these terms and persons with disabilities. ations, such as visual impairment, visual-motor impair-
Although, the keywords like ‘‘person with disabilities’’ ment, etc. will be covered in the subsequent work.
were used, but the research work that focus on Moreover, the articles have filtered out a particular
impairments other than hearing, motor and speaking impairment condition by obtaining direct evidences
have been filtered out. This is because ‘‘person with from the article. However, if direct evidences for
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 5

Table 3. Weightage for different criteria (criteria with weight ¼ 3 are important).
Relative criteria weight
Criteria C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 Total Weight
HI 3 2 1 1 2 3 1 1 3 0 3 3 2 25
SI Partial Impairment 2 3 3 2 2 3 2 1 3 2 3 2 3 31
SI Full Impairment 0 3 3 2 2 3 2 1 3 3 3 2 1 28
HSI 3 3 3 1 2 0 2 1 3 0 2 3 1 24

an assistive technology application are not provided in The weighted sums are then computed for each paper,
the article, then it is discussed by the research team for as shown in the ‘‘Assessment analysis’’ section. Using
its application and usability. these screening criteria, research papers belonging to the
We introduce 13 criteria, as shown in Table 1, that same group of impairment conditions can then be
address different aspects of challenges facing the eight compared using the weight scores. We have used three
impairments described earlier. The criteria have been approaches for the weight scores and these are explained
divided into different categories, including Essential in the following section: (1) percentage of essential
System Input (includes different technologies where at criteria score, which presents the coverage of essential
least one is needed to provide input signal to the system technologies needed for each paper in order to satisfy the
in order for the PWD user to communicate with the input/output requirements of PWD users, and determine
Downloaded by [Gazi University] at 22:00 19 February 2016

systems, and other people), Supporting System Input how well a particular paper fulfills the essential conditions
(assists in utilizing relevant extra input technologies to of that group; (2) percentage of important criteria score,
communicate with other people, systems or the environ- which presents an overall coverage of the important
ment, and is useful to have for supporting additional input technologies needed for certain impairment groups,
signal to the system), Essential System Output (includes where papers based on the most significant criteria that
different technologies where at least one is needed to have weight of 3 would be considered to determine how
provide output signal to PWD users to observe the well a particular paper fulfills the important conditions of
environment), Supporting System Output (assists in inter- that group; (3) percentage of overall criteria score, which
acting with interfaces and observing the environment, and presents an overall coverage of the technologies related
is useful to have for supporting additional signal to PWD to all criteria, where papers with high-weighted scores
user) and System Processing (improves the accuracy of would be considered appropriate in terms of meeting the
results that will be sent to or read by PWD users). Each overall needs for a particular impairment group. In
criterion is assigned a weight according to its importance addition, two other approaches are used to find the
level as shown in Table 2, and according to the group of overall assessment results of the important criteria relative
impairment conditions summarized in Table 3. to other important criteria and relative to each other, and
We assign each reviewed paper a score value in the it can be understood in the ‘‘Assessment analysis’’ section.
range of 0–10 for every criterion, based on the extent to Using MWSM with multiple discussions to reduce the
which the paper addresses this criterion. The scores are variance in expert scores would have the benefit of
elicited from experienced researchers, who follow a accurate quantitative and qualitative evaluation that
clear methodology for the assessment through checking leads to better judgment of the quality of research
for evidence of supporting technologies associated papers in covering the research criteria.
with the criteria in the research paper, and we report
average scores. The following assessment guidelines
were to the researchers or evaluators for assessing the Assessment analysis
reviewed papers: papers that have clear evidence of
the technologies associated with the criteria with Let us assume that Xi is the score of paper x against
enough explanation and example can have score Criterion i (Ci), [which has a value between 0 and 10, and
between 8 and 10. Papers with clear evidence, but depends on how well the paper supports the features
with less description can have a score between 5 and 7. for criterion (i)].
Papers that might support technologies associated with Wi: is the relative Weight of Criterion-i (Ci), [that has a
the criteria can have a score between 3 and 5. Papers value of 0 (not relevant), 1 (low importance), 2 (import-
that have little support of the technologies associated ant), 3 (very important)]
with the criteria can have a score between 1 and 2, and
X
13
papers that have no evidence of the criteria would have Total assessment score for Paper-x ¼ Wi  xi ð1Þ
a score of 0. i¼1
6 G. KBAR ET AL.

For example, the total weight of HI paper ¼ The fourth method is to find the overall percentage
P13
i¼1 Wi ¼ 25 as shown in Table 3, and the total score score of an important criterion for all researched papers
of paper: relative to others’ important criteria, using the following
formula.
x ¼ X1  3 þ X2  2 þ X3  1 þ X4  1 þ X5  2
þ X6  3 þ X7  1 þ X8  1 þ X9  3 þ X6  0 VðCjÞ=Max½VðCjÞ ð5Þ
þ X11  3 þ X12  3 þ X13  2 where V(Cj) is the total value of important criteria
The assessment methodology used in this paper is Cj that has a value ¼ 3, for n papers using for-
P P
based on five assessment approaches. mula ¼ ni¼1 ifðWj ¼ 3Þxj Þ= ni¼1 if ðWj ¼ 3Þ, and
Pn P
The first method is to find the paper’s percentage Max½VðCxÞ ¼ i¼1 if ðWj ¼ 3Þ10Þ= ni¼1 if ðWj ¼ 3Þ.
score against all criteria, which gives a comprehensive The fifth method is to find the overall percentage
idea of the paper’s coverage, using the following score of an important criterion for all researched papers
formula. relative to each other’s important criteria using the
formula.
X
13  X13
¼ Wi xi = Wi ð2Þ .X m .X m hX n i
i¼1 i¼1 ¼ VðCrÞ Cj ¼ VðCrÞ xi wi ð6Þ
j¼1 j¼1 i¼1
The second method is to find the paper’s percentage
where Cr is the criteria that have maximum weight ¼ 3,
Downloaded by [Gazi University] at 22:00 19 February 2016

score against important criteria, which explains how the P


important criteria are addressed, using the following and m j¼1 Cj is the total score of all criteria, m is the total
number of important criteria and n is the total number
formula.
of research papers.
X
13  X13
¼ if ðWi ¼ 3Þxi = if ðWi ¼ 3Þ ð3Þ
i¼1 i¼1
Assumption
The score in Formula 3 is considered only, if the
Three different threshold measures were used to assess
logical weight of criterion i ðWi ¼ 3Þ is true or equals 1,
the satisfactory level for the different category (essential,
and ignored if the logical weight equals 0.
important and general) of the criteria. For essential
The third method is to find the paper’s percentage
criteria, the threshold should be 75% to have satisfac-
score against essential criteria, which explains how
tory level since, the evidence of the substitution
essential criteria are addressed, using the following
technologies associated with these criteria must be
formula.
high. While for important criteria, the threshold should
X
13  be 50% since the technologies associated with these
¼ ½if ðWi ¼ 3Þ&Wi 2 fEssentialListgxi =
criteria are needed to assist this group but are not
i¼1
ð4Þ essential. In addition, the threshold for general criteria is
X
13  chosen to be 35% since the technologies associated
½if ðWi ¼ 3Þ&Wi 2 fEssentialListg with these criteria are supplementary which can have
i¼1
additional benefit but are of less importance than other
essential or important criteria.
To determine the score of Essential criteria for system
input or output that have multiple technologies
associated with multiple important criteria, one would
Assistive technologies for hearing and speech
need to consider the maximum score of the related
impaired (HSI)
criteria. For example, if a research paper of Full HI Deafness and hearing impairment affects an individual’s
supports 2 essential important input criteria (C1, C6), ability to communicate with a companion at every
then we need to consider the maximum of these criteria stage in their life, especially when attending classes
as Max(C1, C6) while computing the score. Similarly, if and official meetings. According to the World Health
we have two essential important output criteria Organization,[1] around 360 million people, or 5.3% of
(C9, C11), then we need only the maximum of the world’s population, are suffering from disabling
these criteria Max(C9, C11) while calculating the score. hearing loss. The figure is around 11 million in the
Then the percentage score of the essential criteria Middle East and North Africa region including Saudi
of paper x would be: [Max(C1, C6)*3 + Max(C10, C11)*3]/ Arabia.
Total-Selected-Max-Score ¼ [Max(C1, C6)*3 + Max(C10, As described in the ‘‘Assessing research papers using
C11)*3]/10*(3 + 3). multi-weighted scoring model’’ section, this article
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 7

Table 4. Relevant criteria of HSI.


C1 C2 C3 C9 C12
Speech to text Keyboard or touch screen or remote Gesture control using sensor or camera Display system Mobile support

Table 5. Assessment table of HSI.


Input Output Processing Percentage of
Percentage of Percentage of essential criteria
important criterion. all criterion using (C1 or C2 or C3, and C9)
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 using formula 3 formula 2 using formula 4
Weightage 3 3 3 1 2 0 2 1 3 0 2 3 1
HSI-1 0 0 8 0 0 7 0 0 6 2 1 0 0 28% 18% 70%
HSI-2 7 1.5 8.5 0 0 7 0 1 7.5 3.5 0 0 0 49% 31% 80%
HSI-3 0 0 7 0 1.5 1 0 0 8 1 0 0 0 30% 20% 75%
HSI-4 5 2.5 9.5 0 0 9 0 0 5.5 0 0 8 0 61% 38% 75%
HSI-5 5 3 10 0 0 3 0 0 4.5 0 0 4 0 53% 33% 73%
HSI-6 0 1.5 8 0 0 6 0 0 0 3 0 0 0 19% 12% 40%
HSI-7 7 4.5 5 6 7 8 0 0 8 0 0 7 0 63% 48% 75%
HSI-8 0 1.5 7.5 5 0 9 0 0 10 3 5.5 7.5 0 53% 40% 88%
HSI-9 0 0 1 9.5 9 0 0 0 9 0 0 0 0 20% 24% 50%
HSI-10 0 0 1 10 9 0 0 0 0 0 0 0 0 2% 13% 5%
Downloaded by [Gazi University] at 22:00 19 February 2016

Average percentage result 38% 28% 63%

identified 13 criteria to assess research papers based on huge pre-defined database and lot of training for the
different impairment conditions. The relevant and most user. The paper proposes a new device for deaf and
important criteria (with weight ¼ 3) for the assessment dumb but it is lacking basic support features also for the
group HSI (hearing and speech impairment) are high- hearing disabled as it does not provide features from the
lighted in brown in Table 4, where C1, C2, C3, C9 and defined criteria C1 and C2 to enable deaf to communi-
C12 would be valuable technologies that can assist this cate with others.
group. Paper HSI-2 of Table 5 [7] also proposed a similar
We applied the assessment method described in solution to M. Al-Rousan where the user would need to
Table 5 by computing the average assessment of 2–3 prepare a database for themselves. The computer
experienced researchers. Table 5 lists the average scores employed on this system will be trained to recognize
of those researchers for the 10 papers we assessed and the color of skin of the user which will then process
that are related to HSI according to the multi-weighted through a series of complicated procedures of skin
scoring method discussed in the ‘‘Assessing research detection, extraction, calibration and comparison. The
papers using multi-weighted scoring model’’ section. skin color is detection is used for differentiating it from
This table also presents the percentage of all criterion the background color. An algorithm called Support
using formula 2, the percentage of important criterion Vector Machine handles all the processing. The system
using formula 3 and the percentage of essential criteria was not found to be efficient in recognizing too many
using formula 4. and complex gestures because of technical constraints
Researchers in the past have developed and imple- in the system. The paper addresses C1, C3 and C9 with
mented various technologies to assist people with low complexity but is lacking the support of C2 and C12
hearing and speech impairments for communicating which are important for inputting commands and
with others. Paper HSI-1 in Table 5 [6] developed a sign offering mobility to the user.
recognition system where the input sign language Another type of gesture recognition system using a
gesture is fed to a camera. The signal is then processed camera was developed by paper HSI-3 of Table 5 [8].
by a computer which already has a pre-defined database Their system was built to recognize only 32 gestures
to match and recognize the input gesture. The system where the user would build a database by storing a
provides output as verbal word and text on a screen to particular word with each gesture before using the
send to non-impaired receiver. The system does face a system. The system was made to recognize the up/down
number of limitations. Considering criterion C12, where position of fingers in one hand and then convert the
the served individuals cannot be mobile because it signal to binary numbers of 5 characters (e.g. 1,0,1,1,0). It
works with a computer and stationary camera, it would then be converted to a decimal number and the
recognizes only Arabic language, and it requires a associated word is played and displayed as output. The
8 G. KBAR ET AL.

system attained high accuracy but it was limited to Markov Models (HMM) and the output is displayed as
recognition of only 32 words, which will have to be text at a screen on the system itself. In our assessment,
remembered always by the user making it difficult to be the paper has covered fairly in overall criteria and
practical. In this paper only C3 and C9 are addressed specifically C1, C9 and C12, which can be good example
significantly whereas research gaps could still be for HSI group. Some researchers worked on tracking the
reduced to deliver other support from relevant criteria. movement and performance of the disabled automatic-
A system for mobile phones has also been developed ally using various technologies.
to recognize hand gestures by paper HSI-4 of Table 5 [9]. Paper HSI-8 of Table 5 [13] used flex sensors as the
An Android application was developed where the input device to make gestures. The system contains a
camera would identify the gesture and the phone microcontroller that encodes the incoming sign and
would generate audio as well as text output previously converts it into text and audio signal. The paper strongly
associated with the gesture. Hence, it removes the addressed criteria C3, C9 and C12 where it could still be
requirement of special equipment to be carried. The improved by providing speech to text and touch screen
system lacks any functionality for the hearing disabled, input facility.
though it claims to be serving both deaf and dumb. This Paper HSI-9 of Table 5 [14] demonstrated the use of
article was evaluated to have strongly addressed C3 and wearable sensors for observation of daily routine
C12 but did not provide much support for input like activities of disabled. The sensors monitor the move-
speech to text or keyboard. ment of users continuously. The technology was built
Downloaded by [Gazi University] at 22:00 19 February 2016

Researchers in this field have also implied neuro-fuzzy upon a behavior modeling method based on HMMs. The
system for detection of sign language as described by system claimed to detect even small movements while
paper HSI-5 of Table 5.[10] This system uses a camera for sleeping and could display and activity pattern on the
video input and converts the image into devanagari screen. Nevertheless, our assessment found the paper
spoken language as audio output. It detects and has covered well C9 but it has limitation in serving deaf
compares extreme points of fingers to the existing and dumb specifically for communication.
database. This article got high score on C10 but was not Different approaches for monitoring PWD was
found strong in all other criteria for easing communica- adopted by paper HSI-10 of Table 5.[15] They used
tion for the hearing disabled, providing display system several wide angled cameras installed around a home.
or mobility. However, the system was evaluated to have Five routine activities like standing, sitting, etc. were
addressed the necessary criteria on a medium scale targeted to be monitored. Support Vector Machines
which means there is still scope of advanced research in (SVM) and hybrid classifiers were used for formulating a
this technology. detection algorithm. Despite being successful in detect-
Another project targeted at gesture recognition ing movements, these papers did not meet any of the
where the user does not have to stay in front of essential criteria defined by us as it did not offer any
camera is presented by paper HSI-6 of Table 5.[11] The features for the deaf and dumb. This paper covered C4
user would just remain in a specified range of the Wi-Fi (tracking location) and C5 (sensor behavior) well, which
network and make a gesture. The gesture would be could be an example for tracking behavior that is not
detected by the system using Doppler shift phenom- critical for HSI but it is relevant for the general criteria
enon and played on a speaker after matching with a defined for HSI.
predefined pattern. Only C3 criterion is served in this As shown in Table 5, the average of overall percent-
article while all others are ignored, and the paper does age covering the essential criteria was 63%. Coverage of
not offer any relevant feature for the hearing impaired as important criteria was 38%, and coverage of overall
claimed. criteria was 28%. By applying the terms explained in the
Other technologies have been used and developed by ‘‘Assumption’’ section, the number of papers covering
researchers for gesture generation and recognition.  75% of essential criteria on an average is 5 out of 10 or
Some have employed flex sensors. Paper HSI-7 of 50%. The number of papers covering  50% of import-
Table 5 [12] created a prototype for deaf and dumb. ant criteria on an average is 4 out of 10 or 40%. And the
The speech disabled would wear flex sensors that number of papers covering  35% of all criteria on an
change resistance according to a degree of finger average is 3 out of 10 or 30%. These results indicate that
bend. Different voices were pre-stored in the system not all papers covered the essential criteria well for HSI
and played on a speaker with associated gestures. For group which would not satisfy the needs of this group.
the deaf people, communication is established through In addition, since 40% of the papers covered important
an Automatic Speech Recognition System (ASR). A non- criteria reasonably, there is a room for further research to
disabled person needs to train the system using Hidden cover multiple technologies for system input/output to
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 9

Table 6. Assessment results of HSI group.


Assessment parameters C1 C2 C3 C9 C12
Total value of important criteria 72 43.5 196.5 175.5 79.5
Percentage of coverage level of important criteria relative to maximum 24% 15% 66% 59% 27%
of specific criteria by applying Formula (5)
Percentage of coverage level of important criteria relative to each 13% 8% 35% 31% 14%
other by applying Formula (6)

provide extra flexibility to PWD users in choosing relation to essential criteria for HSI and identifying the
relevant technology for them to substitute missing research gaps for this group, a more comprehensive
sensation caused by their impairment conditions. There review of the literature need to be done. Table 6 shows
is low percentage of the 10 research papers in providing that C3 and C9 have highest coverage among these
a comprehensive solution to address all criteria. relevant criteria relative to maximum score with values
Covering all criteria would add a great flexibility to of 66 and 59%, respectively. These are followed by C1,
PWD users in practicing multiple technologies in one C2 and C12 which have a coverage value of 24, 15 and
solution. Hence, further research should be done to find 27%, respectively which are considered low. This leads to
more comprehensive solutions that can address the the conclusion that further research in relation to C1, C2
need of HSI group through the support of multiple and C12 should be conducted to improve the conditions
Downloaded by [Gazi University] at 22:00 19 February 2016

technologies associated with important criterion as well for HSI individuals.


as overall relevant criterion which make HSI group more
comfortable in utilizing such solution. The technologies
Assess the coverage of important criteria for HSI
associated with relevant criteria for HSI people that need
relative to each other’s score
further investigation would be analyzed and determined
in the next section.
Table 6 shows that C3 and C9 have the highest coverage
among the relevant criteria with values of 35 and 31%,
Level of coverage of the features associated with respectively. These are followed by C12, C1 and C2
important criteria for HSI group
which have relatively low scores with value of 14, 13 and
In this section, we described the coverage of different 8%, respectively. This low coverage of C1, C2 and C12
criteria in relation to the sum of maximum scores of leads to the same conclusion as stated above that these
important criteria, as well as comparing the coverage of criteria need further research.
each criterion in relation to other criteria (Table 6). In
addition, the percentage of coverage for the overall
Assistive technologies for hearing impaired
criteria of the 10 papers is also presented where the
average of overall percentage of all criteria for HSI is 28% For hearing impaired (HI) individuals, we defined some
[by applying Formula 2], as shown in Table 5. This result important technological parameters which are necessary
indicates that further research is required to improve the for this individual to behave normally in a public
coverage of all criteria relevant to the HSI group; the environment. We define the parameters as: Speech to
specific criteria that require further research are text (C1) to understand spoken text, Text to Speech (C6)
explained in the sub-sections below. to type and convert text to audible voice, Display System
(C9) to read system input/output and Haptic Feedback
(C11) which helps HI to receive alert signals through
Assess the coverage or relevant criteria in
vibration in the device or a flashing light as shown in
relation to the maximum value for HSI
Table 7. In addition, importance of having a mobile
Table 6 illustrates the results by applying formulas (5) support (C12) using phone, tablet or any wearable
and (6) to determine the average percentage for equipment has been identified for free movement of the
covering each criterion in relation to the accumulated user.
sum of all criteria, as well as in relation to each other. Table 8 lists the average scores of those researchers
For hearing and speech impaired, criteria C3 (gesture for the 17 papers we assessed, and that are related to HI
control) and C9 (display system) are covered well among according to the multi-weighted scoring method dis-
these 10 papers while criteria C1 (speech to text), C2 cussed in the ‘‘Assessing research papers using multi-
(keyboard or touch screen) and C12 (mobile support) are weighted scoring model’’ section. This table also pre-
not addressed well. To get more accurate result in sents the percentage of all criterion using formula 2, the
10 G. KBAR ET AL.

Table 7. Relevant criteria for HI.


C1 C6 C9 C11 C12
Speech to text Text or digital mapping to speech Display system Haptic or visual feedback Mobile support

Table 8. Assessment table of HI.


Percentage of
Percentage of Percentage of essential criteria
Input Output Processing important criterion. all criterion (C1, and C6 or C9)
using formula 3 using formula 2 using formula 4
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13
Paper HI-1 9.5 10 0 0 0 0 6 0 9 0 1.5 0.0 0.0 40% 34% 93%
Paper HI-2 0.0 7.7 0.0 0.0 0.0 0.0 7.7 4.7 9.0 0.0 0.0 0.0 0.0 18% 22% 45%
Paper HI-3 0.0 7.5 0.0 0.0 0.0 0.0 0.0 0.0 10.0 0.0 0.0 0.0 0.0 20% 18% 50%
Paper HI-4 10.0 1.5 0.0 0.0 0.0 0.0 0.0 0.0 10.0 0.0 0.0 1.5 0.0 43% 27% 100%
Paper HI-5 9.7 0.0 10.0 0.0 0.0 6.0 0.0 0.0 10.0 1.0 3.3 0.0 0.0 58% 39% 98%
Paper HI-6 9.7 10.0 0.0 0.0 0.0 8.7 0.0 0.0 8.7 9.3 9.3 10.0 9.7 93% 71% 92%
Paper HI-7 4 4 0 0 3 5 1 0 4 5 4 3 0 41% 30% 45%
Paper HI-8 0 6 1 0 9 0 0 0 9 0 0 0 0 17% 23% 43%
Paper HI-9 0.0 1.5 0.0 0.0 0.0 0.0 0.0 0.0 1.5 6.5 0.0 0.0 10.0 3% 11% 8%
Paper HI-10 8.0 1.5 0.0 0.0 5.0 8.0 0.0 0.0 2.0 5.5 0.0 0.0 8.0 24% 26% 50%
Paper HI-11 6.5 2.0 0.0 0.0 7.5 7.0 0.0 0.0 0.0 7.0 0.0 0.0 8.0 16% 24% 40%
Downloaded by [Gazi University] at 22:00 19 February 2016

Paper HI-12 0.0 2.0 0.0 0.0 7.5 4.0 7.5 7.5 10.0 6.5 0.0 0.0 2.5 28% 32% 50%
Paper HI-13 10 0 0 0 0 0 0 0 10 4.0 0.0 0.0 0.0 40% 24% 100%
Paper HI-14 0 0 0 0 10 0 0 0 1 6.5 0.0 4.0 0.0 10% 14% 5%
Paper HI-15 8 0 0 0 7 8 0 0 0 9 10 0 7 50% 41% 75%
Paper HI-16 0 6 0 0 8 0 0 0 10 10 10 0 8 40% 41% 50%
Paper HI-17 0 0 0 0 8 0 0 0 0 5 9 0 0 18% 17% 0%
Average percentage result 33% 29% 55%

percentage of important criterion using formula 3 and enhance motivation for a hearing impaired person while
the percentage of essential criteria using formula 4. navigating virtual environments as described by paper
Software was developed to work on a computer that HI-3 of Table 8.[18] The system had pre-designed stories
converts speech to text and enables deaf people to talk that would display text and play videos with sign
over VoIP calls as described by paper HI-1 (Table 8).[16] language for the learning of students. This article was
The user would normally speak and receive text also found to be addressing only parameter C9. Again a
messages from the other end at every pause in limited coverage of all important criterion could be
speech. According to HI parameters, this article scored associated to limited scope of interfacing of HI user to
high on C1 and C9 but it did not serve the individuals web application.
with disabilities on the remaining parameters C6, C11 Other types of systems were created where speech
and C12 where there is an opportunity to develop this would be converted to sign language for the hearing
software for mobile phones with an added feature of disabled. Paper HI-4 of Table 8 [19] developed software
text to speech conversion. This allows the user to just to capture speech through a recognition mechanism
type the message in a noisy environment and the and convert it to Arabic sign language. The system
receiver can listen to it synthetically. worked on a PC where avatars making sign language
Paper HI-2 of Table 8 [17] studied the effect of gestures would be displayed for the deaf. It contained
using graphical icons instead of hypertext links in pre-defined databases of images of words of sign
accessing a website. They designed a website using language and another one of avatars to match with
graphical icons for the hearing disabled. They analyzed corresponding word and play to the user. This article
the system by measuring response time and disorienta- addresses C1 and C9 significantly but it could be
tion accomplishment time of the users. This paper improved further by supporting C6 and C12.
addresses only one essential criterion C9 whereas all A system called Tessa was fabricated and tested paper
other criteria were ignored. The limited coverage of HI-5 of Table 8 [20] to aid communication between a
important criteria could be related to focusing on web deaf person and a post-office clerk. It would convert the
interface only to evaluate interaction with HI user which clerk’s speech to British sign language and be played by
is fine for addressing less important criteria to assist an avatar on a screen. The speech recognition could be
further the HI group. trained for the speaker’s voice, which would increase the
Another research group developed a website espe- system’s recognition performance. The paper addressed
cially for the deaf with sign language and icons to C1 and C9 significantly and covered C6 reasonably.
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 11

Instead of introducing new technologies some Paper HI-11 of Table 8 [26] also analyzed the
researchers focus on evaluating existing works. Paper magnitude of problems in listening with various types
HI-6 of Table 8 [21] evaluated main utility features of 10 of background noises. The paper found some factors
mobile applications that claimed to assist as replace- that contribute to difficulty beside natural ailment like
ment of hearing aids. The applications were evaluated Reduced Frequency selectivity and Loudness recruit-
on services offered like hearing tests, possibility of ment. It was also attempted to compensate the effect of
creation of personal profile, mic support etc. The paper these problems using signal processing and amplitude
ignored user feedback in usage of applications. All compression methods. The paper covered C1 and C6 as
parameters were assessed to be significantly addressed it focused more on C5, C7 and C10 which are non-
in this paper including C1, C6, C9, C11 and C12. This relevant criteria for HI but it could be relevant for HI with
article presents a good example for addressing the need low vision.
for HI people. Another study was conducted by paper HI-12 of
Other researchers evaluated effect of environment Table 8 [27] on response of phrases played to HI people
factors such as inadequate support systems such as masked by noise on pupil of human eye. Significant
hearing aids on people’s career choices and develop- dilation in the pupil was observed by the study when the
ment as described by paper HI-7 of Table 8.[22] The user listened to the phrases with fluctuating noise or
survey was done by using forced choice and open- disturbance by a single-talker. The paper provided an in-
Downloaded by [Gazi University] at 22:00 19 February 2016

ended questions about disabled’s work experience since depth analysis on pupil response of HI people while
leaving their university. The average score of this article having conversation in noisy environment. This article
is related to the purpose of having survey instead of focused on studying the behavioral effect of noise on
providing relevant solution to HI. eye pupil dilation. This article did not address all relevant
A concept was built to study an online community essential criteria except C6 and C9 and focused its
that provides education support for individuals with research on non-relevant less-relevant criteria C5, C7, C8
hearing impairment as described by paper HI-8 of and C10 as this paper focused on studying the behav-
Table 8.[23] Factors were grouped into categories ioral effect of noise on eye pupil dilation.
including hearing loss and education to link to aspects Paper HI-13 of Table 8 [28] created a video watching
such as identity, community building and connections system for the hearing disabled by generating word-by-
between communities. The aspects and factors were word scripts so that HI could track the speech. The voice
aligned into the model in a way that the aspects were is also controlled by the system so that the user can
built upon each other, whereas the impacts of the understand the emotion of played character. Solution of
factors were also investigated. This article was found to this article would be improved by providing this system
address only C9 and neglecting others. on mobile devices or by supporting text entered by user
Paper HI-9 of Table 8 [24] studied the usage of a to be converted to speech so user can see their spoken
certain noise reduction algorithm to improve sound words. The paper covered C1 and C9 significantly but
quality for the partial-severely impaired disabled. It was did not address other relevant criteria as these were not
found that output of noise reduction algorithm was not covered by its scope.
significantly affected by degree of hearing impairment of A study on children’s use of and attitudes towards
the user. This article focuses only on noise reduction hearing devices was conducted by paper HI-14 of
effect and not on whole solution for HI. This article did Table 8.[29] The study found out that males view
not cover well the all relevant criteria but addressed C10 hearing technology more positively than do females.
and C13. Having severe haring loss also promoted positive
Another investigation was done by paper HI-10 of attitudes toward hearing aids and cochlear implants.
Table 8,[25] to assess whether auditory training in noise As this article just surveyed the attitudes of users, it was
would improve speech recognition in noise for chil- not able to properly address any of the relevant essential
dren with hearing impairment. The disabled were criteria for HI. However, it covered C5 significantly which
divided in 3 groups and trained in Auditory Training is behavior analysis.
Interrupted (ATI), Auditory Training Continuous (ATC) Paper HI-15 of Table 9 [30] proposed music hearing
and Control. It was found that people trained on ATI system for partially deaf. The system was fitted on a chair
performed better than other groups. This paper which also gave haptic feedback according to the beats
addressed C1, C6 and ignored C9, C11 and C12 where and notes in the music to enhance the disabled person’s
these features could provide additional support for the listening experience. Along with haptic feedback, visual
disabled person. display was provided that played abstract animations
12 G. KBAR ET AL.

Table 9. Assessment results of HI group.


Assessment parameters C1 C6 C9 C11 C12
Total value of important criteria 209 119.5 311.5 141.5 56.5
Percentage of coverage level of important criteria relative 41% 23% 61% 28% 11%
to maximum by applying Formula (5)
Percentage of coverage level of important criteria relative 28% 14% 37% 17% 7%
to each other by applying Formula (6)

from the components of music which offered better feel choosing relevant technology for them to substitute
of hearing to the deaf person. This article addressed C9 missing sensation caused by their impairment conditions.
and C11 significantly for offering a display system and There is low percentage of the 17 research papers of 18%
haptic feedback. However, covering other relevant in providing a comprehensive solution to address all
criteria C1, C6 and C12 would make this solution more criteria that add great flexibility to PWD users in practicing
relevant for HI users. multiple technologies in one solution. Hence, further
Another research by paper HI-16 of Table 9 [31] research can be done to have more comprehensive
featured a chair with haptic feedback that helped in solution that can address the need of HI group through
providing speech therapy to the person with partial the support of multiple technologies associated with
hearing impairment. The chair produced vibrations with important criterion as well as the overall relevant criterion
Downloaded by [Gazi University] at 22:00 19 February 2016

voice of user as well that of speech therapist. Also, the which make HI group more comfortable in utilizing such
disabled would hear amplified voice through head- solution. The technologies associated with relevant
phones. The system tended to improve the vocal ability criteria for HI that needs further investigation would be
of the user in a few weeks. Among relevant criteria, C1, C6 analyzed and determined in next section.
and C11 were covered reasonably leaving gap for further
research in C9 and C12 where user could use feature of
Level of coverage of the features associated with
mobility and a display system for improving accessibility.
important criteria for HI group
Paper HI- of Table 9 [32] developed a prototype
system of a haptic chair to assist the speech therapy In this section, we described the coverage of different
process for deaf and hearing impaired people. The chair criteria in relation to the sum of maximum scores of
produces the vibrotactile stimulation to certain body important criteria, as well as comparing the coverage of
parts (palms and fingers) based upon the sound each criteria in relation to other criteria (Table 9). In
produces by the impaired person. The authors tried to addition, the percentage of coverage for the overall
evaluate that how much the haptic chair will have criteria of the 17 papers is also presented where the
positive effect on deaf individuals during speech therapy average of overall percentage of all criteria for HI is 29%
process. A performance evaluation study on 20 students (by applying formula 2) as shown in Table 8. This result
were performed, and it was found out that the students indicates that further research is required to improve the
who used the haptic chair give better result with coverage of all criteria relevant to the HI group; the
significant differ starting from week 12. specific criteria that require further research are
As shown in Table 8, the average of overall percentage explained in the sub-sections below.
covering the essential criteria was 55%, covering the
important criteria was 33%, and covering all criteria was
Assessing the coverage or relevant criteria in
29%. By applying the terms explained in the
relation to the maximum value for HI
‘‘Assumption’’ section, the number of papers covering
essential criteria by average 75% is 6 out of 17 or 35%, Table 9 illustrates the results by applying formulas (5)
the number of papers covering important criteria by and (6) to determine the average percentage for
average 50% is 3 out of 17 or 18%, and the number of covering each criterion in relation to the accumulated
papers covering all criteria by average  35% is 4 out of sum of all criteria, as well as in relation to each other.
17 or 24%. These results indicating that the majority of For hearing impaired, C1 (speech to text) and C9
the 17 papers did not covered the essential criteria well (display system) are covered well among these 17 papers
with 35% for HI group which would not satisfy the needs while criteria C6 (text to speech), C11 (haptic feedback)
of this group. In addition, since 18% of the papers and C12 (mobile support) are not addressed well.
covered important criteria reasonably, there is a need for Table 9 shows that C1 and C9 have highest coverage
further research to cover multiple technologies for system among these relevant criteria relative to maximum score
input/output to provide extra flexibility to PWD users in with values of 41 and 61%, respectively. These are
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 13

followed by C6, C11 and C12 which have a coverage assessed and that are related to SI according to the
value of 23, 28 and 11%, respectively. This leads to the multi-weighted scoring method discussed in the
conclusion that further research in relation to C6, C11 ‘‘Assessing research papers using multi-weighted scoring
and C12 should be conducted to improve the conditions model’’ section. This table also presents the percentage
for hearing impaired people. of important criterion using formula 3, and the percent-
age of essential criteria using formula 4.
Paper SI-1 of Table 12 [33] presented an artificial
Assess the coverage of important criteria for HI
speaking mouth for speech impaired persons. PWDs
relative to each other’s score
with speech disabilities wear flex sensors for commu-
Table 9 shows that C9 and C1 have the highest coverage nicating by making gestures which gets converted into
among the relevant criteria for HI with values of 37 and text and then be played as voice. The gestures are
28%, respectively. This are followed by which have low English Alphabets only. It is a standalone system. The
scores with value of 14, 17 and 7%, respectively. This low system is based on motion sensor. It works in coaching
coverage of C6, C11 and C12 leads to the same mode and operation mode. In coaching mode, a
conclusion as stated above that these criteria need database was created with template motions using
further research, as supporting these criteria for HI alphabets and stored in a microcontroller. The user
would be beneficial for utilizing the text to speech needs to remember particular finger angles for each
facility, for providing haptic feedback and for supporting alphabet, where there are total 256 gestures to be
Downloaded by [Gazi University] at 22:00 19 February 2016

mobility application for these people. learned. The PWD can benefit by using this system for
communication with normal person. In addition, the
system is compact and can be carried around. The
Assistive technologies for speech impaired (SI)
criterion that allows PWD with SI to use adjustable
Speech impairment is a condition in which the ability to speaker level and getting visual feedback are not
produce speech sounds that are necessary to commu- covered well in this article.
nicate with others is impaired. Speech impairments can Paper SI-2 of Table 12 [34] presented a new form of
be mild, such as occasionally mispronouncing a couple augmentative and alternative communication (AAC), as
of words, to severe, such as not being able to produce well as the Voice-Input Voice-Output Communication
speech sounds at all. Aid (VIVOCA). The proposed VIVOCA recognizes dis-
We have assessed seven articles concerned with the ordered speech and converts it into synthetic speech
technological solutions developed for supporting peo- used to communicate with normal person. They used a
ples with speech impairments. new methodology for building small vocabulary and
The significant criteria that are relevant to full SI and speaker dependent Automatic Speech Recognition (ASR)
partial SI group which have maximum weight value of 3 with reduced amount of training data. The system was
and value of 2 for C1 for SI are as following (Tables 10 developed on a PDA and is in prototype stage. The
and 11, respectively): processing consists of speech recognition, message
The rest of criteria are of less importance for the building, speech synthesis and audio output. The user
speech impairment group such as ‘‘C1: Speech to text, speaks disordered speech into a mic and speech is
C4: Tracking location using sensors, etc. The assessment recognized and processed by speech synthesizer. The
of each paper of this impairment group has been done recognized words are sent to a message building
by at least two researchers. Table 12 lists the average module. The module sends feedback to synthesizer
scores of those researchers for the seven papers we where the audio is played. Small vocabulary is required

Table 10. Relevant criteria for full SI.


C2 C3 C6 C9 C10 C11
Keyboard or touch Gesture control using Text or digital Display Auto adjustable Haptic or visual
screen or remote sensor or camera mapping to speech system Speaker and feedback (vibration
Volume control or flashing

Table 11. Relevant criteria for partial speech impairment.


C1 C2 C3 C6 C9 C11 C13
Speech Keyboard or Gesture control Text or Display Haptic or Haptic feedback
to text touch screen using sensor digital mapping system visual feedback (vibration or flashing)
or remote or camera to speech (vibration or flashing
14 G. KBAR ET AL.

Table 12. Assessment results of SI.


Input Output Processing
Criteria Total scores for each paper
Weight (PSI) 2 3 3 2 2 3 2 1 3 2 3 2 3 Percentage Percentage percentage of percentage of
Weight (FSI) 0 3 3 2 2 3 2 1 3 3 3 2 1 coverage coverage essential coverage essential coverage
of important of important (C1 or C2 or C3, and (C1 or C2 or C3, and
criteria for PSI criteria for FSI C6 or C9), PSI C6 or C9), FSI
SI-1 0 0 8.5 0 0 8.5 0 0 7.5 3.5 0 8 0 37% 47% 85% 85%
SI-2 9.5 7.5 0 0 0 9 0 0 9 1.5 0 8 1.5 50% 45% 92% 83%
SI-3 0 0 8.5 0 0 7 0 0 0 2 0 0 0 23% 29% 78% 78%
SI-4 0 1 8.5 0 0 0 0 0 4.5 0 0 0 0 21% 23% 43% 43%
SI-5 0 0 10 0 0 9 0 0 0 10 1.5 0 6 40% 51% 95% 95%
SI-6 0 10 0 0 0 10 0 0 10 6.5 0 6 5 53% 61% 100% 100%
SI-7 0 1 8.5 0 0 0 0 7 7.5 0 0 0 0 26% 28% 43% 43%
Average percentage result 36% 41% 76% 75%

to be stored by each user. These samples were used to touch screen, while at the same time it lacks the support
train the system using Hidden Markov Models with an of feedback to user and mobility which are important for
algorithm named Baum–Welch algorithm. The advan- SI people.
tages include the ease with which the device can Paper SI-4 of Table 12 [36] tried to build a neural
Downloaded by [Gazi University] at 22:00 19 February 2016

facilitate the communication between persons with network that could be used to recognize only static
speech disorder by allowing them to build phrases hand gestures of Japanese sign language. They used
using word-by-word recognition system. However, the data gloves with flex sensors for each finger. For testing
disadvantages include: low accuracy of 67%, the inter- purpose, a data glove was worn by an expert Japanese
face can hold only few words at one time, and that the sign language user. The glove was fitted with five
system was not found user friendly by all participants in sensors and was used as the input device for the system.
testing. A switch had to be pressed each time the user Three experts were asked to wear the gloves and make
wanted to start communicating. This article has a good static Japanese sign language gestures. The neural
coverage for partial SI, whereas still relatively acceptable network was built by the authors using commercially
for full SI, the other relevant criteria to full SI, like (C3, available neural network development tool. The input
C10 and C11) are not covered well. Supporting Haptic or data from gloves was used to train the system. Training
visual feedback as well as vibration in this article would was conducted 10 times per training set of particular
make the solution friendlier. gestures. Total 25 gestures out of 41 were used for
Paper SI-3 of Table 12 [35] proposed a new device training purpose. Testing was conducted after prepar-
called ‘‘Virtual Voice’’ that makes use of a sensor which ation of each training set. The testing gestures were not
the user fits on their lips. The user then would just speak used for training. The disadvantages of the proposed
the word without any sound. The device utilized AG-500 system associated with the system that works on static
Articulograph sensor. The sensor tracked the space gestures only. That is, the user has to hold the gesture
between several predefined points on the lips and fed for the system to read, in addition to very low accuracy
this data to a software for processing and generating of 45% of recognition. Furthermore, the sensors failed to
output voice. The space was plotted with its corres- recognize the differences in gesture directions. Finally,
ponding Speech Assessment Methods Phonetic the technology is still in early stages of development
Alphabet (SAMPA) code. The software was called and also does not hold any special feature over other
PHONWEB to which the SAMPA code was fed in as existing technologies such as flex sensors. This article
input. The main advantage of the proposed system is developed a prototype for training and testing purpose
that it can be used by illiterate person. However, the and not for providing a comprehensive solution for SI.
limitation of the system is that, to make it friendlier by The coverage of important criteria for PSI and FSI are
developer, it was developed for PC only. Also, the relatively low since some of the relevant criteria to
algorithm is not available for JAVA mapping. The partial SI, like (C1, C2, C6, C9, C11 and C13) are not
coverage of important criteria for this PSI and FSI are covered well, and also, the relevant criteria to full SI, like
relatively low since some of the relevant criteria to (C2, C6, C9, C10 and C11) are not covered well.
partial SI, like (C1, C2, C9, C11 and C13) are not covered Paper SI-5 of Table 12 [37] developed a glove system
well, and also, the relevant criteria to full SI, like (C2, C9, for deaf and dumb persons, which can convert the sign
C10 and C11) are not covered well. The papers focuses language to synthesized speech. The system consists of
on sensors as input instead of supporting keyboard and a leather glove, flex sensors, AVR ATMEGA32L
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 15

microcontroller, SpeakJet sound synthesizer and LM386 support SI people. The proposed system has the benefit
sound amplifier. Software used to develop the system of supporting gesture control that was found to be user
are; AVR Studio 4, Magnevation Phrase Translator for friendly and beneficial, especially for SI. The algorithms
SpeakJet V1.4, OrCAD 10.0 and ISP Programming. The presented in this article could be integrated with other
glove that equipped by flex sensors need to be worn by solutions to improve the support of input control for
user’s finger in order to generate input signal even with people with SI that would be helpful for SI people to use
small bend of the finger. The signal is handled by circuit sensor for gesture control along with keyboard and
for amplification and signal processing, and then fed to touch screen.
AVR microcontroller for digital conversion. The output As shown in Table 12, the average of overall
signal of the AVR is sent to Speak-Jet that has percentage covering the essential criteria for PSI and
Mathematical Sound Architecture (MSA) and predefined FSI were 76 and 75% consecutively, and covering the
spoken words according to the received signal from important criteria for PSI and FSI were 36 and 41%,
AVR. The spoken words are combined to make sentence, respectively. By applying the terms explained in the
where the SpeakJet uses synthesizer to enhance its ‘‘Assumption’’ section, the number of papers covering
volume. The benefits of the proposed system are that it essential criteria by average  75% for PSI is 5 out of 7 or
is easy to use, portable and requires low power. Among 71%, and for FSI is also 5 out of 7 or 71%. The number of
the systems’ disadvantages, are that the system is not papers covering important criteria by average  50% for
water resistant and relatively, few number of messages PSI and FSI are 2 out of 7 or 29%. These results indicate
Downloaded by [Gazi University] at 22:00 19 February 2016

(32) had been programmed for 32 gestures with that the majority of the seven papers for PSI and FSI
maximum 243 messages can be programmed. The covered the essential criteria well for SI group which
paper could provide additional benefit to SI group if a would satisfy the needs of this group. However, since
display system (C9) and the support of feedback (C11) 29% of the papers covered important criteria reasonably,
are supported well. In addition, the input method using there is a need for further research to cover multiple
sensor could complement other technology needed for technologies for system input/output to provide extra
SI group such as keyboard and touch screen as well as flexibility to PWD users in choosing relevant technology
speech to text for partial SI. for them to substitute missing sensation caused by their
Paper SI-6 of Table 12 [38] designed a speech impairment conditions. The low percentage of covering
synthesizer, which could receive text input through a important criteria by average of 29% indicate that
touch screen or a cell phone keypad and convert the majority of the papers did not provide a comprehensive
text into a speech to help speech impairment individuals solution to address important criteria that add great
to speak. The system consists of a cell phone keypad to flexibility to PWD users in practicing multiple technol-
input text and a screen to monitor the input text by user, ogies in one solution. Hence, further research can be
a speech synthesizer IC is used to process the data to a done to have more comprehensive solution that can
speech signal and a speaker to produce the speech address the need of SI group through the support of
output of the system, and microcontroller board is multiple technologies associated with important criter-
responsible for the communication between the hard- ion which make SI group more comfortable in utilizing
ware boards. The authors designed an algorithm that such solution. The technologies associated with relevant
processes each letter symbol of input text at a segment criteria for SI that needs further investigation would be
and then compounds each phoneme to generate the analyzed and determined in next section.
final speech as one signal. The system could help the
speech impairment to make conversations with others.
Level of coverage of the features associated with
The disadvantage is that the system has no option to
important criteria for SI group
save templates for the user and each time the user has
to enter all text he needs to convert to a speech. This In this section, we described the coverage of different
article has a very good coverage for the relevant criteria criteria in relation to the sum of maximum score of
for both partial and full SI. important criteria, as well as comparing the coverage of
Paper SI-7 of Table 12 [39] presented a design of three each criteria in relation to other criteria. In addition, the
action recognition algorithms for gaming consoles for percentage of coverage for the overall criteria of the
public places. They also proposed the design of gesture seven papers is also presented. where the average of
based interface for a game for public setting. Algorithms overall percentage of all criteria for FSI is 75% (by
were developed for recognition of the complex actions applying Formula 2), and the average of overall percent-
using a webcam only. The system is claimed to have fast age of all criteria of PSI is 76% (by applying Formula 2) as
and robust interpretation of hand gestures which can shown in Table 12. This result indicates that further
16 G. KBAR ET AL.

Table 13. Assessment results of SI.


Relative important criteria for full SI and partial SI
Assessment parameters C1 C2 C3 C6 C9 C10 C11 C13
Total value for important criteria for FSI 58.5 132 130.5 115.5 70.5 4.5
Total value for important criteria for PSI 19 58.5 132 130.5 115.5 4.5 37.5
Percentage of coverage level of important criteria relative 28% 63% 62% 55% 34% 2%
to maximum value for FSI by applying Formula (5)
Percentage of coverage level of important criteria relative 14% 28% 63% 62% 55% 2% 18%
to maximum value for partial SI by applying Formula (5)
Percentage of coverage of important criteria relative to each 11% 26% 26% 23% 14% 1%
other FSI by applying Formula (6)
Percentage of coverage of important criteria relative to each 6% 12% 26% 6% 23% 1% 7%
other Partial speaking by applying Formula (6)

research is required to improve the coverage of all criteria and 11% consecutively. While the coverage of other
relevant to the SI group; the specific criteria that require criteria C11 of SI is very low of 1%, C1 (speech to text)
further research are explained in the sub-sections below. and C13 (noise filtering) have low percentage for partial
SI. This leads to the same conclusion as stated above
that the criteria C2, C11 (haptic feedback) for SI, and C1
Downloaded by [Gazi University] at 22:00 19 February 2016

Assessing the coverage or relevant criteria in


and C13 for partial SI need further researches to support
relation to the maximum value for SI and
better the conditions for SI and partial SI.
partial SI
Table 13 illustrates the results by applying formulas (5)
Discussion
and (6) to determine the average percentage of covering
each criterion in relation to the accumulated sum of all In this article, we surveyed around sixty scientific papers
criteria as well as in relation to each other. related to the use of technologies for addressing the
For both full SI and partial SI, criteria C3 (Gesture needs for different impairment groups of PWD by
Control), C6 (Text to speech) and C9 (display) are reviewing and evaluating these papers using Multi-
covered well among these seven papers, while criteria weighted Scoring Model. Three impairment groups of
C11 (Feedback to user), C2 (Keyboard or touch screen) PWD have been assessed in this paper, which are HI, SI
and C10 (auto speaker volume) are not addressed well. and HSI, where the other impairment groups of Visual
Criteria C3, C6 and C9 have the highest coverage among Impairment, Motor Visual, Motor Impairment, Motor
these criteria in relation to the maximum score for full Hearing and Speech Impairments are covered in separ-
speech impairment, with values of 63, 62 and 55%, ate article. The assessment methodology of this litera-
respectively, as shown in Table 13. These are followed by ture survey is based on qualitative assessment according
C10, C2 and C11 (haptic feedback) which have a to general criteria that are associated with relevancy to
coverage value of 34, 28 and 2%, respectively. For the use of technology for disable people, paper
partial SI, C3, C6 and C9 are followed by C1, C2, C11 and credibility, the practical applicability of these papers’
C13 which have coverage value of 14, 28, 2 and 18%, solution in meeting the people needs, and the evidence
respectively, and therefore are not covered well. This in supporting their claims. In addition, the methodology
leads to a conclusion that further research in relation to also includes quantitative assessment based on Multi-
C2, C10 as well as C11 should be conducted to improve weighted Scoring Model according to defined 13 criter-
the conditions for people with SI, and further research in ion that are weighted and mapped to different impair-
relation to C1 and C13 is also required for partial SI. ment groups, as well as they have been assessed by
three researchers. Thirteen criteria related to system
input, system output and system processing have been
Assess the coverage of important criteria for SI
identified that cover the technological requirements of
and partial SI relative to each other’s score
the various combinations of impairment conditions of
We can see from Table 13 that C3 (gesture control), C6 the PWDs. We have assessed these papers based on
(text to speech) and C9 (display system) has the highest essential criteria to satisfy at least one technology for
coverage among these relevant criteria for SI and partial system input and one for system output. We also
SI with value of 26, 26 and 23% consecutively. These assessed the papers base on the overall percentage of
followed by C10 (auto volume control), C2 (keyboard or important criterion to have more comprehensive solu-
touch screen) for SI which has a coverage value of 14 tion that would add extra flexibility to PWD in having
ASSISSTIVE TECHNOLOGY REVIEW FOR DISABLE PEOPLE 17

multiple input and output technologies to serve their determine the focus research area needed to improve
needs. Furthermore, we also assessed these papers for the technological solution for the six impairment groups
full comprehensive solution that apply the 13 criteria at of PWDs.
different weight according to their importance level. This
will give us more comprehensive approach and higher
flexibility for PWD users to utilize multiple technological Declaration of interest
solutions that can enhance the performance of the
The authors whose names are listed in the first page certify that
system input, system output and system processing. This they have no affiliations with or involvement in any organiza-
analysis enabled computing in overall the level of tion or entity with any financial interest (such as honoraria;
coverage of individual assessment criteria with respect educational grants; participation in speakers’ bureaus; mem-
to each group of impairment, and which represent the bership, employment, consultancies, stock ownership, or other
equity interest; and expert testimony or patent-licensing
main target information of this article assessment. In
arrangements), or non-financial interest (such as personal or
order to improve the assessment results, few meetings professional relationships, affiliations, knowledge or beliefs) in
have been conducted to discuss the survey results, in the subject matter or materials discussed in this manuscript.
order to verify the ratings of technologies and their This Project was funded by The National Plan for Science
associated criteria, as well as discussing the results of the Technology and Innovation (MARRIFAH), King Abdul-Aziz City
for Science and Technology, Kingdom of Saudi Arabia, Award
assessment. This series of meetings led to reduction in
Number (12-ELE3220–02).
the variance between the assessment’s results of differ-
Downloaded by [Gazi University] at 22:00 19 February 2016

ent researchers for the same paper.


References
[1] World Health Organization. World report on disability
Conclusions ‘‘The way forward: recommendations, chapter 9’’. 2011.
[cited 2015 May 3] Available from: http://www.who.int/
The assessment result of HSI indicates that there is still
disabilities/world_report/2011/chapter9.pdf.
scope of more research in HSI according to the [2] Centre on Human Policy. The right to health in Saudi
important criteria (C1, C2, C3, C9 and C12), and that Arabia ‘Right to health in the Middle East’ project, Law
are related to text to speech, gesture control, display School, University of Aberdeen; 2009.
system and mobility support. The assessment result of HI [3] Kbar G, Aly S, ElSharawy I, et al. Smart Help At The
Workplace For Persons With Disabilities (Shw-Pwd),
indicates that average score of all-important criteria for
ICIES 2015: XIII International Conference on Intelligent
HI of the 17 papers is 33%. This score for HI is not high, Environments and Systems, Paris, France.
which indicates that more research is required to [4] AL-Rousan M, Assaleh K, Tala’a A. Video-based signer-
address the requirements needed for HI conditions independent Arabic sign language recognition using
according to the important criteria. The criterions that hidden Markov models. Appl Soft Comput. 2009;9:
have low score for HI are C6, C11 and C12, and are 990–999.
[5] Yuvarani M, Vinothkumar S, Prabu R, et al. Shrug and
related to text to speech, visual feedback and mobility
Voice Recognition System For Dumb And Hearing
support. In terms of covering the essential criterion, both Impards. Int J Res Advent Technol. 2009;2:36–39.
FHI and SHI papers indicate acceptable results of [6] Subha Rajam P, Balakrishnan G. Real time Indian sign
average greater than 55%. The assessment result of SI language recognition system to aid deaf-dumb people.
indicates that the total score of all important criteria of IEEE Trans Instrum Mesure 2011;58:2501–2506.
[7] Sangeetha K, Barathi Krishna L. Gesture detection for deaf
the seven papers is between 36 and 41%. This value is
and dumb people. Int J Dev Res. 2014;4:749–752.
relatively low, which indicates that more researches are [8] Misal PK, Pathan. MM. Design & implementation of ANFIS
required to address the requirements needed for FSI and system for hand gesture to Devanagari conversion. Int J
PSI conditions according to the important criteria. These Eng Res Gen Sci 2014;2:207–212.
important criterions that have low score are C2, and C11 [9] Swaroop T. Wireless and uninstrumented communication
for SI, and C1 and C13 for P SI, and are related to speech by gestures for deaf and mute based on Wi-See technol-
ogy. IOSR J Electron Commun Eng. 2013;7:31–35.
to text, keyboard or touch screen, visual feedback, and
[10] Yella Reddy G, Ramana Reddy M. Design of ultrasonic
noise filtering. In terms of covering the essential spectacles, flex sensor based voice generation and
criterion, SI papers indicate a very good results of speech to text (data) conversion techniques for physically
average greater than 75%. Hence, the common criteria disable people. Int J Res Electron Commun Eng. 2013;1:
among this six impairment group that need further 12–22.
[11] Radha HG, Shruti SD. Design and development of an
research are C1, C2, C4, C5, C11, C12, C13 and the
assistive device for speech and hearing impaired. Int J
specific criteria for HSI and HI that needs more research Innov Technol Res 2013;2:859–862.
and investigation are C6 and C9. It becomes obvious that [12] Atallah L, ElHelw M, Pansiot J, et al. Behaviour profiling
the quantitative assessment allowed us to clearly with ambient and wearable sensing. 4th International
18 G. KBAR ET AL.

Workshop on Wearable and Implantable Body Sensor response in adults with hearing impairment. J Acoust Soc
Networks (BSN 2007), Vol 13. Berlin: Springer;2007. Am. 2014;135:1596–1606.
p. 133–138. [26] Hong R, Wang M, Yuan X-T, et al. Video accessibility
[13] Brdiczka O, Reignier P,, Crowley JL. Detecting Individual enhancement for hearing-impaired users. ACM Trans
Activities from Video in a Smart Home. 11th International Multimedia Comput Commun Appl. 2011;7:24.
Conference on Knowledge-Based and Intelligent [27] Rekkedal AM. Assistive hearing technologies among
Information and Engineering Systems (KES), 2007. p. students with hearing impairment: factors that
363–370. promote satisfaction. J Deaf Stud Deaf Educ. 2012;17:
[14] Shirley BG, Thomas J, Roche P. VoIPText: Voice Chat for 501–517.
Deaf and Hard of Hearing People. In: Proceedings of 2012 [28] Nanayakkara S, Wyse L, Taylor E. Enhancing
IEEE Second International Conference on Consumer musical experience for the hearing impaired using
Electronics, Berlin (ICCE-Berlin), Berlin (Germany): IEEE; visual and haptic displays. Hum Comput Interact.
2012. 2013;28: 115–160.
[15] Fajardo I, Canas JJ, Salmeron L, et al. Improving deaf [29] Nanayakkara S, Wyse L, Taylor E. The Haptic Chair as a
users’ accessibility in hypertext information retrieval: are speech training aid for the deaf. ACM CHI Conference on
graphical interfaces useful for them? Behav Inform Human Factors in Computing Systems, Melbourne,
Technol. 2006;25:455–467. Australia; 2014. p. 405–410.
[16] Bottoni P, Borgia F, Buccarella D, et al. Stories and signs in [30] Nanayakkara S, Wyse L, Taylor EA. Effectiveness of the
an e-learning environment for deaf people. Univ Access haptic chair in speech training. 14th international ACM
Inform Soc. 2012;12:369–386. SIGACCESS conference on Computers and accessibility,
[17] Halawani SM, Zaitun AB. An avatar based translation Austin, Texas, USA. 2012 p. 235–236.
Downloaded by [Gazi University] at 22:00 19 February 2016

system from arabic speech to arabic sign language for [31] Padmanabhan V, Sornalatha M. Hand gesture recognition
deaf people. Int J Inform Sci Educ. 2012;2:13–20. and voice conversion system for dumb people. Int J Sci
[18] Cox S, Lincoln M, Tryggvason J, et al. TESSA, a system to Eng Res. 2014;5:427–431.
[32] Hawley MS, Cunningham SP, Green PD, et al. A voice-
aid communication with deaf people. In: The proceedings
input voice-output communication aid for people with
of the fifth International ACM conference on ASSETS and
severe speech impairment. IEEE Trans Neural Syst Rehabil
SIGACCESS Conference on Computers and Accessibility.
Eng 2013;21. p. 23.
New York: ACM Press; 2002. p. 205–212.
[33] Rajeswari K, Jeevitha E, Kalai Selvi VKG. ‘‘Virtual Voice’’ –
[19] Offiah MC, Rosenthal S, Borschbach M. Assessing the
the voice for the dumb. International Conference on
utility of mobile applications with support for or as
Computational Intelligence and Computing Research,
replacement of hearing aids. Proc Comput Sci 2014;34:
Coimbatore, India; 2010. p. 1–3.
593–600.
[34] Machacon HTC, Shiga S. Recognition of Japanese finger
[20] Punch Ŕe, Hyde M, Power D. Career and workplace
spelling gestures using neural networks. J Med Eng
experiences of australian university graduates who are
Technol. 2010;34:254–260.
deaf or hard of hearing. J Deaf Stud Deaf Educ. 2004;12: [35] Ahmed SF, Ali SMB, Qureshi SM. Electronic speaking
504–517. glove for speechless patients: a tongue to a dumb. The
[21] Kozuh I, Hintermair M, Ivanišin M, et al. The concept of IEEE Conference on Sustainable Utilization and
examining the experiences of deaf and hard of hearing Development in Engineering and Technology, Selangor,
online user. Proc Comput Sci. 2013;27:148–157. Malaysia; 2010. p. 56–60.
[22] Kim J, Nama KW, Yook S, et al. Effect of degree of [36] Venkatraman S, Padmavathi TV. Speech for the disabled.
sensorineural hearing impairment on the results of The International Multiconference of Engineers and
subjective evaluations of noise-reduction algorithm. Computer Scientists, Hong Kong; 2009.
Speech Commun. 2015;68:1–10. [37] Roccetti M, Marfia G, Semeraro. A. Playing into the wild: a
[23] Sullivan JR, Thibodeau LM, Assmann. PF. Auditory gesture-based interface for gaming in public spaces. J Vis
training of speech recognition with interrupted and Commun Image. 2012;23:426–440.
continuous noise maskers by children with hearing [38] Grant MJ, Booth A. A typology of reviews: an analysis of 14
impairment. J the Acoust Soc Am. 2012;133:495–501. review types and associated methodologies. Health Inform
[24] Moore BCJ. Speech processing for the hearing-impaired: Lib J. 2009;26:91–108
successes, failures, and implications for speech mechan- [39] Federici S, Scherer MJ. The assistive technology assess-
isms. Speech Commun. 2003;41:81–91. ment model and basic definitions. In: Federici S, Scherer
[25] Koelewijn T, Zekveld AA, Festen JM, et al. The influence of MJ, editors. Assistive technology assessment handbook.
informational masking on speech perception and pupil Boca Raton (FL): CRC Press; 2012. p. 1–10.

You might also like