Mintzatek, Text-to-Speech Conversion Tool Adapted to Users with
Motor Impairments
J. Eduardo Pérez, Myriam Arrue and Julio Abascal
EGOKITUZ: Laboratory of HCI for Special Needs, University of the Basque Country (UPV/EHU)
Informatika Fakultatea, 20018 Donostia, Spain
Keywords: Adapted User Interface, Assistive Technology, Limited Mobility, Speech Disorder, Text-to-Speech
Conversion, User-Centered Design.
Abstract: Text-to-speech (TTS) conversion software tools are capable of generating synthetic voice from written text.
These tools are essential for some groups of impaired users who have speech difficulties. In some cases, this
limitation is caused by some kind of motor impairment. However, current TTS tools are not fully accessible
as contain barriers for those users with limited mobility in upper extremities. This paper presents the most
significant accessibility barriers detected for this specific user group. In addition, an accessible TTS tool,
Mintzatek, has been implemented based on User-Centered Design (UCD) process. The user interface of the
developed tool is adapted to users with limited mobility in upper extremities. All the development process
has been guided by two real motor impaired users with plenty of experience in the use of assistive
technologies.
1 INTRODUCTION
In the last few years, plenty of text-to-speech
conversion software tools have been developed.
They are capable of generating synthetic voice from
written text and are meant to help a wide range of
impaired users in their everyday lives.
Most of text-to-speech conversion tools are
oriented to users who need speech assistive
technology due to any functional diversity limiting
their oral communication capacity. In some cases,
this limitation is caused by some kind of motor
impairment.
However, the development process of the user
interfaces of such tools does not always consider
universal access paradigm. Consequently, they may
contain numerous accessibility barriers for those
users with specific functional diversity for example
for users with limited mobility in upper extremities.
In general, the developed user interfaces require
use of mouse events for activating the different
functionalities. This interaction mode implies
mobility, precision and strength in upper extremities,
something not possible for diverse groups of users.
The objective of this paper is to analyse the
accessibility barriers in current text-to-speech
conversion tools and to develop an accessible and
usable tool for users with motor impairments. User-
Centered Design (UCD) (Newell et al., 2000)
process has been applied. It is a user interface design
approach that takes into consideration final users’
characteristics during project development. The
products obtained are optimized for those users. This
can be only achieved with the participation of final
real users at different phases of the iterative design
process of a product. Following this approach, we
recruited end-users and they were involved during
all the process starting at early phases of the analysis
and design.
Several accessibility barriers were observed and
directly discussed with users. New features were
arised from interviews with users and several testing
sessions were planned during the development
process. As a result, a text-to-speech conversion tool
has been developed which considers universal
access principles, includes functionalities for
improving users’ performance and enables motor
impaired users to easily interact with the user
interface.
The paper has been structured in the following
way: several text-to-speech conversion tools are
presented in section two. In section three, an
accessibility evaluation of selected three tools is
presented. The accessibility guidelines and user
112
Eduardo Pérez J., Arrue M. and Abascal J..
Mintzatek, Text-to-Speech Conversion Tool Adapted to Users with Motor Impairments.
DOI: 10.5220/0004897701120119
In Proceedings of the 16th International Conference on Enterprise Information Systems (ICEIS-2014), pages 112-119
ISBN: 978-989-758-029-1
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
interface adaptation techniques considered in the
development of the text-to-speech conversion tool
accessible for users with motor impairments are
discussed in section four. Section five is devoted to
the development process based on UCD. Finally,
conclusions and future work are drawn.
2 RELATED WORK
Numerous text-to-speech (TTS) conversion tools
have been developed in the last few years. Some of
them are devoted to mobile devices while others to
personal computers. We have analysed three tools
selected based on their popularity in user
communities and the features they include in their
user interface. All of them are standalone desktop
software applications. This type of application
provides valuable aid to those users experiencing
difficulties for oral communication. The only
requirement is to install the software in any
computer (for instance, their personal laptop) so
there is no need to purchase any speech-generating
dedicated device.
An interesting example of such TTS tools is the
BJ Hermes commercial software (BJ Adaptaciones,
2011). It was created by a Spanish company
specialised in the development of assistive
technology tools for people with disabilities.
However, it presents serious accessibility problems
for users with specific motor impairments. Its
interaction model is based on a user who precisely
uses the mouse device and keyboard access to the
functionalities may be quite difficult for some users.
Another remarkable TTS system is the software
called “Verbose Text to Speech” (NCH Software,
2013). Despite the fact that this shareware tool
implements some screen reader features (not of
interest for our research), it presents other useful
ones like the lateral command bar with stretchtext
characteristics. However, the tool presents some
functionality only reachable by mouse device so
keyboard-only users cannot access to them.
TTSReader tool has also been considered
(SpheNet, 2011) during this research work. It
includes some interesting features such as the
tracking of the speech within the typed text.
Activating this feature ensures that the cursor
position is synchronized with the speech process
highlighting the word the tool is reading aloud.
Nevertheless, there is not any personalization option
in the tool and some features require users to use a
mouse device.
3 ACCESSIBILITY BARRIERS
ON ANALYSED TTS TOOLS
This section presents an accessibility evaluation of
the three different TTS tools described in the
previous section (Figure 1 shows their main
windows): “BJ Hermes PC” (BJHt), “Verbose Text
to Speech” (VTSt) and “TTSReader” (REAt). This
evaluation is focused on accessibility issues
concerning physically impaired users that suffer
from limited movements in upper extremities
(causing imprecise use of mouse or even keyboard-
only access).
Figure 1: Main windows of three different TTS tools analysed during this work (left: commercial “BJ Hermes PC”, middle:
shareware “Verbose Text to Speech” and right: freeware “TTSReader”).
Mintzatek,Text-to-SpeechConversionToolAdaptedtoUserswithMotorImpairments
113
Several accessibility principles and guidelines have
been considered in order to detect accessibility
barriers in the analysed tools: (Kurniawan et al.,
2006), (NIA and NLM, 2002) and (Gajos et al.,
2010). It can be noticed that some accessibility
guidelines considered in this analysis are related to
elderly people who may suffer ageing-related motor
impairments as well as users temporarily disabled or
interacting in unusual postures.
3.1 Buttons Issues
3.1.1 Reduced Dimensions
The analysed tools contain buttons with
inappropriate size. They are too small, producing
many difficulties when the mouse is used with a low
precision. The VTSt application also implements
buttons like links on its lateral command bar (see
Area 3 in Figure 1 - middle), reducing even more
dimension of clickable areas.
Another problem present in the three tools is that
horizontal and vertical sequences of buttons are too
close to each other (see Areas 2 and 3 in every
capture of Figure 1 - left, middle and right). This
issue causes motor impaired users click the incorrect
option very often when using the mouse device.
3.1.2 Lack of Textual Information
The REAt tool implements five buttons related with
file options that do not incorporate any text (see
Figure 2). This issue can prevent users from
knowing the function of each button.
Figure 2: File options related buttons with no label tags on
"TTSReader" application.
3.2 Other Components Issues
3.2.1 Cascading Menus
Every studied TTS tool contains hierarchical
cascading menus on the upper area of its main
window (see Areas 1 in every capture of Figure 3 –
up, middle and down). These walking menus
introduce accessibility barriers when the user cannot
easily move the mouse pointer onto the desired entry
(being especially difficult the navigation through
secondary menus). In addition, it makes difficult to
select an option by keyboard-only users as they have
to go through a lengthy sequence of moving keys.
Although every entry is reachable with a keyboard,
it is recommended to minimise the use of this type
of menu bars.
Figure 3: Same distribution of cascading menus above [1]
and toolbar below [2], for the three TTS tools analyse
d
(up: “BJ Hermes PC”, middle: “Verbose Text to Speech”
and down: “TTSReader”).
3.2.2 Click-and-Drag Components
Every TTS tool studied uses the same spinner
component (see Figure 4 – up left, up right and
down). This component is provided for altering
synthesis parameters like the speed, pitch or volume
of the generated speech. These “click-and-drag”
components are extremely difficult to use for motor
impaired users. These spinner components are
strongly not recommended by accessibility
guidelines. In addition to that, only BJHt tool (see
Figure 4 – up left) implements a spinner component
reachable by keyboard-only users.
Figure 4: Spinner button components highlighted for each
of the three TTS tools analysed (left: “BJ Hermes PC”,
right: “Verbose Text to Speech” and down:
“TTSReader”).
ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems
114
3.3 Scrolling Issues
3.3.1 In Relation to Dialog Windows
Every TTS tool studied implements the same
standard “open file dialog window” (see Figure 5)
and “save file dialog window”. The scroll bars
included to navigate within the file system add an
accessibility barrier for people with low precision
with mouse device. The scrolling buttons are
difficult to access due to their small size.
Figure 5: Standard “open file dialog window” as
implemented by the three studied TTS tools with scroll
bars tagged.
3.3.2 In Relation to Text Editor
None of the three studied TTS tools incorporates
specific controls for easily moving the cursor
through the text editor content (see Area 1 in every
capture of Figure 1 – left, middle and right). Users
have to manage with the small sized scroll buttons.
Navigating through long texts becomes tedious for
some users who experience difficulties with motor
coordination and fine movements.
3.4 Keyboard Issues
3.4.1 Out of Reach Controls
The analysed TTS tools include many out of reach
controls for a keyboard-only user. It means a total
inaccessibility to those functionalities for those
users. An analysis has been performed in order to
determine the out of reach controls included in each
tool. The Areas 2 highlighted in Figure 3 show the
inaccessible functionalities for keyboard-only users
in the main window. This presents accessibility
barriers for some specific group of users even
though the functionalities are accessible from other
menus.
In addition, VTSt and REAt tools implement not
reachable lateral command bars (see Areas 3 in
Figure 1 - middle and right) for keyboard-only users.
All these not accessible functionalities can be
performed in an alternative way. However, the
alternative is selecting the option in cascading
menus that are not recommended for some groups of
users.
Moreover, some components for speech
adjustments are completely unreachable for
keyboard-only users in VTSt and REAt tools (see
Figure 4 - up right and down).
3.4.2 Hotkeys Assignments
VTSt and REAt tools allow the user to define
custom hotkeys to perform several commands (read,
pause or save text...) by pressing only one key. Both
tools present the same accessibility issue related to
this functionality: the assigned hotkey is not
displayed over the correspondent command button
on the GUI. Therefore, users have to investigate by
themselves the way to define them and memorize
each one in order to get to use them.
On the other hand, BJHt implements not
customizable hotkeys. These values are correctly
displayed next to the appropriate label (see Area 3 in
Figure 1 - left).
Table 1: This table summarizes the results obtained in the
accessibility evaluation of the TTS tools (x: accessibility
barrier detected, -: not detected, xx: issue detected without
alternative way)
Issue BJHt VTSt REAt
Buttons
Reduced size X X X
Lack of text - - X
Other components
Cascading menus X X X
Click-and-drag X X X
Scrolling
Dialog window X X X
Text editor X X X
Keyboard
Out of reach
controls
X XX XX
Hotkeys
assignments
- X X
Dialog windows X X X
Other
Lack of text editor
personalization
- X X
Mintzatek,Text-to-SpeechConversionToolAdaptedtoUserswithMotorImpairments
115
3.4.3 In Relation to Dialog Window
Every analysed TTS tool presents the same
accessibility issue regarding dialog windows.
Moving to a specific control requires going through
a lengthy sequence of tabbing, so the time required
to complete tasks increases.
3.5 Other Issues
Related to the GUI personalization, only BJHt
implement a customizable option within text editor
(font and background features). All the analysed
tools implement the possibility of showing or hiding
controls (lateral panel, toolbar and status bar).
3.6 Summary Table
Table 1 summarizes the accessibility barriers
detected in the analysed TTS tools. The results
shown in this table determines the existence of
accessibility barriers in each tool and whether the
tool provides an alternative way for users with
limited mobility in order to avoid those barriers.
4 ACCESSIBLITY ISSUES
CONSIDERED IN THE
DEVELOPMENT PROCESS
The implemented system called MintzaTek (see
Figure 6) is aimed at achieving an accessible TTS
tool for motor impaired users with limited
movements in upper extremities. The goal is that the
implemented functionalities can be accessed and
used in an efficient way without using any additional
assistive technology.
Figure 6: Main window of the MintzaTek TTS tool
implemented within this work.
User interface adaptation techniques have been
considered in the development in addition to
accessibility guidelines. According to Knutov et al.
(Knutov et al., 2009), adaptation techniques can be
classified in three main groups: content adaptation
techniques, presentation adaptation techniques and
navigation adaptation techniques. This section
presents the techniques considered in the
development of MintzaTek.
4.1 Content Adaptations
These techniques involve changes in the content
display within the interface. Some interaction
components have been replaced with other more
accessible.
4.1.1 Avoiding Cascading Menus
We have implemented independent static menus (see
Figure 7) avoiding the use of walking menus.
Figure 7: Alternative menu design.
4.1.2 Avoiding Click-and-Drag Components
We have replaced the spinner components with two
buttons (see Figure 8): one of them to increase the
corresponding parameter and the other one to
decrease it.
Figure 8: Two buttons design instead of spinner control.
ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems
116
4.2 Presentation Adaptations
This group of techniques involve style changes in
the interface layout. Some aspects of the layout have
been changed in order to avoid inaccessible
interfaces.
4.2.1 Avoiding Scrolling Buttons
Accessibility issues related with scrolling (see
Section 3.3.1) have been solved by applying an
alternative design. The total information to present
in the user interface has been broken into data
chunks (see Figure 9). Additional buttons (previous
page and next page) are incorporated when
necessary so users can navigate backward and
forward through the content.
Figure 9: Dictionary dialog window.
4.2.2 Accessible Buttons
The buttons included in MintzaTek fulfil these
necessary characteristics: minimum space between
consecutive buttons, minimum dimensions, textual
description (see visible buttons in Figures 6 to 10).
4.2.3 Personalization Features
The implemented system allows each user to
localize the user interface in two possible languages,
to change within different colours schemes (for texts
and backgrounds), to resize texts and buttons, or to
alter the distribution of the interface main group of
elements.
Although this group of adaptations are not
related with physical disabilities needs, users with
low vision or interacting with computer in unusual
postures can benefit from them.
4.3 Navigation Adaptations
This group of techniques involve changes dealing
with interface structure and behaviour in order to
facilitate the navigation of users. The following
aspects have been included in MintzaTek tool in
order to facilitate user interaction.
4.3.1 Fast Navigation within Text Editor
To move faster through texts within the editor when
only using a keyboard, several commands have been
implemented. These are aimed at jumping forward
and backward variable amount of characters through
text (next/previous word, phrase, paragraph and all
text). Also another command has been included to
navigate to next text editor tab.
4.3.2 Keyboard-only Full Accessibility
In order to create an accessible tool for both
keyboard-only users and low precision mouse users,
every button present within the interface can be
accessed via direct keyboard button and mouse.
4.3.3 Fast Keyboard Interaction
Every command within all dialog windows also
implements direct access via keyboard (see Figure
10). This feature makes more dynamic the
interaction of keyboard-only users through dialog
windows, reducing the time required to complete
any task.
Figure 10: Alert dialog window for deleting file
confirmation.
5 TOOL DEVELOPMENT BASED
ON USER-CENTERED DESIGN
During the development of MintzaTek we have
collaborated with two real end-users. These users
experience reduced mobility, precision and strength
in upper extremities. In addition, they have
difficulties in oral communication due to their motor
impairments. They both are usual users of text-to-
speech conversion tools and have huge experience
with assistive technology. Therefore, both users
were involved in our UCD process so we benefit
from their knowledge and experience. The objective
was to develop a tool accessible for them but also
for other users in a similar situation.
Mintzatek,Text-to-SpeechConversionToolAdaptedtoUserswithMotorImpairments
117
5.1 Collaborations with Users
The UCD process applied in the development of
MintzaTek was totally oriented to the collaboration
of both real users. Every feedback obtained in one
step of the process was immediately adopted in next
steps. The main steps of the process with direct
interaction with users were the following:
1. Preliminary analysis of existing TTS tools
2. Gathering Initial Requirements of MintzaTek
3. User Interface Design
4. User Testing
The first step is related to analyse existing TTS
tools. The involved users are experienced in using
“BJ Hermes PC” tool. We had the opportunity of
observing direct interaction of both users with the
system. They both use different strategies for
interacting with the interface: one of them emulates
mouse device by keyboard and the other uses direct
access keyboard buttons for accessing the
functionalities. We simulate an interview using the
TTS tool and annotate any strategy they applied for
the interaction.
Table 2: Initial requirements gathered from users involved
in UCD process.
Initial requirements
Interaction
Full keyboard access to all implemented
functionalities
Simplified GUI that allows reaching any
function with the fewer steps
Speech synthesis
Speech synthesis of text in Basque language
Customizable synthetic communication with
several voices, and speech speeds
Customizable amount of text send to the
speech generator (word, phrase, paragraph
or everything)
Others
Allow saving texts typed before closing the
tool
Customizable GUI (resolution, language,
colors, distribution,...) and text editor (font,
font size, font color and background color)
The second step was determinant in order to define
the functionalities MintzaTek should provide. We
carried out interviews with both users and perform a
brainstorming technique in order to detect essential
functionalities for the new TTS tool as well as other
complementary functionalities which could be
interesting in order to facilitate the interaction. For
this purpose, results from our TTS tools analysis
were used. We presented functionalities found in
other tools which were not included in “BJ Hermes
PC” to users (for example, the speech tracking
feature of “TTSReader”). In addition, we discuss
functionalities only implemented in “BJ Hermes
PC” which were not in other tools (for example,
personalization issues).
Table 3: Improvements and corrections detected after user
testing sessions for prototypes evaluation in UCD process.
Improvements/corrections
Interaction
Allow defining abbreviations to speed up
the typing process
Add keyboard navigation assistance to help
moving faster through text editor, and
keyboard selection and edition text options
Speech
synthesis
Stop the synthetic speech in any moment
Generate the synthesizer output (synthetic
speech) faster
Add more speed levels to choose from
Others
Allow changing directly a text selection
between capital letters and lowercase letters
Rename several textual information from
the GUI
The third step was intended to test the first interface
designs of the MintzaTek prototypes. We
implemented just the user interface and perform a
user evaluation with both users. We observed the
interaction and gathered all comments aroused in the
sessions. Results from this evaluation sessions were
applied in next iterations (for example, a new
functionality for defining abbreviations to speed up
the typing process was proposed by a user). This
step was repeated several times until the design was
satisfactory for both users.
The fourth step is related to the user testing of the
implemented tool. Users directly interacted with
MintzaTek while we observed them and annotated
any comment for improving the system. Results
from this step were considered in next iterations (for
example, in an evaluation session it was detected the
need of adding navigation assistance functionality so
they could easily navigate through the text). This
step was repeated several times until the
implemented prototype was satisfactory for both
users.
ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems
118
5.2 Results
After all the meetings we had with both users, we
can summarize the following results as part of the
contribution of the UCD process to our research.
5.2.1 Initial Requirements
After the first couple of interviews with both users,
we established the next requirements for the user
interface of our TTS tool (see Table 2).
5.2.2 Improvements and Corrections
After several meetings in which we have analysed
the way both users responded and interacted with
every prototype designed, the following
improvements and corrections have been done (see
Table 3).
6 CONCLUSIONS AND FUTURE
WORK
The existing TTS conversion tools present several
accessibility barriers for users with limited mobility
in upper extremities. The motor impairments causing
these mobility limitations sometimes cause also
speech disabilities. Therefore, there is a specific user
group who could greatly benefit from TTS
conversion tools if they were accessible. The
accessibility barriers detected in this research work
make user interaction difficult in many cases even
impossible in others.
User interface adaptation techniques have been
studied for creating alternative interaction
components and techniques for solving the detected
accessibility barriers. Several content adaptation
techniques, presentation adaptation techniques and
navigation techniques have been considered for the
development of MintzaTek user interface.
The development process of the TTS conversion
tool has been an iterative one and based on UCD.
Two real motor impaired users with large interaction
experience with this type of assistive technology
have been involved in the process. The initial
requirements were gathered with the aid of these two
users and applying techniques such as observations,
interviews and discussion activities. User interface
design have been tested and commented with both
users and prototypes of the system have been
evaluated. Feedback obtained from this UCD
activities have been always considered in the
iterative development process.
We are currently contacting more users with
motor impairments in order to conduct formal user
evaluation of the developed TTS conversion tool.
We also plan to apply eye-tracking methods for
obtaining additional information from user
interaction. This comprehensive user evaluation will
serve to demonstrate the utility of the adapted user
interface components included in the tool.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support of
the Spanish Ministry of Science and Innovation
through Project ModelAccess (TIN2010-15549).
The EGOKITUZ Research Laboratory is supported
by the Department of Education, Universities and
Research of the Basque Government (Eusko
Jaurlaritza/Gobierno Vasco) through Grant# IT-395-
10.
REFERENCES
BJ Adaptaciones, 2011. BJ Hermes PC, version 1.1.2
[software]. Trialware available from: http://www.bj-
adaptaciones.com/descargas/BJ_Hermes/BJHermesSe
tupDemo.exe (Retrieval: October 2, 2013).
Kurniawan, S. H., King, A., Evans, D. G., Blenkhorn, P.
L., 2006. Personalising web page presentation for
older people. Interacting with Computers. ELSEVIER
SCIENCE INC. Volume 18. Issue 3. Pages 457-477.
Newell, A.F., Gregor, P., 2000. User Sensitive Inclusive
Design: in search of a new paradigm. CUU 2000 First
ACM Conference on Universal Usability, 2000.
National Institute on Aging and National Library of
Medicine, 2002. Making You Web Site Senior
Friendly: A Checklist. Available from:
http://www.nlm.nih.gov/pubs/checklist.pdf (Retrieval:
October 2, 2013). NIH & NLM.
Gajos, K.Z., Weld, D.S. and Wobbrock, J.O. 2010.
Automatically Generating Custom User Interfaces for
Users with Physical Disabilities. Journal Artificial
Intelligence, Vol. 174, N.12-13, 910-950. Elsevier.
Knutov, E., De Bra, P. and Pechenizkiy, M. AH 12 years
later: a comprehensive survey of adaptive hypermedia
methods and techniques. New Review of Hypermedia
and Multimedia 5 (1), Pp. 5-38. 2009
SpheNet, 2011. TTSReader, version 1.30 [software].
Available from: http://www.sphenet.com/Files/TTSRe
ader.exe (Retrieval: October 2, 2013).
NCH Software, 2013. Verbose Text to Speech, version
2.01 [software]. Shareware available from:
http://www.nch.com.au/verbose/vbsetup.exe
(Retrieval: October 2, 2013).
Mintzatek,Text-to-SpeechConversionToolAdaptedtoUserswithMotorImpairments
119
