Audio Description on Instagram: Evaluating and Comparing Two
Ways of Describing Images for Visually Impaired
João Marcelo dos Santos Marques
1
, Luiz Fernando Gopi Valente
1
, Simone Bacellar Leal Ferreira
1
,
Claudia Cappelli
1
and Luciana Salgado
2
1
Department of Applied Informatics, Federal University of the State of Rio de Janeiro,
Av. Pasteur 458 - Urca, Rio de Janeiro, Brazil
2
Institute of Computing, Federal Fluminense University,
Av. Gal. Milton Tavares de Souza, s/n - São Domingos, Niterói, Brazil
Keywords: Accessibility, Audio Description, Instagram.
Abstract: The social network Instagram encourages interactions among users around audio-visual content (pictures
and short duration videos). However, this type of content still presents itself as a barrier for the visually
impaired. To mitigate this problem, screen readers can be used, but those only work for images which have
texts in the form of subtitles. Audio description, on the other hand, is a technique that describes visual
images into words, allowing the comprehension of these elements. This technique has been used in many
fields, fostering a scenario of inclusion and opportunities for this public. The objective of this paper is to
evaluate and compare these two forms of describing images published on Instagram: one utilizing the
descriptive text read by the screen reader and another utilizing audio description recorded by the image’s
own author. Through an empirical study, we have identified the form of image description preferred by the
visually impaired participants and if the use of audio description on Instagram would encourage its use by
this public.
1 INTRODUCTION
With the evolution of information and
communication technology, new opportunities and
possibilities are open for people who have some type
of disability. Nowadays it is common to find people
with visual impairment using computers to surf the
Internet with the help of screen reading software.
This public is not small. According to the data from
the 2010 IBGE (Brazilian Institute of Geography
and Statistics) census, concerning those with severe
visual disabilities (those with great difficulty in
seeing or who cannot see at all), more than 6.6
million people have claimed to have this type of
disability. Of these 6.6 million, 506.3 thousand have
claimed to be blind. (IBGE, 2010).
Not all interactive systems, however, are
designed to meet the needs of this portion of the
population. Among the types of systems that present
accessibility problems are social networks.
Accessibility is the term used to indicate the
possibility of anyone enjoying the benefits of life in
society, and among them, is the use of the internet
(NBR 9050, 1994; Nicholl, 2001). Despite the
advances, studies show (Piovesan et al., 2013) there
is still much to be done to meet all of the
accessibility criteria.
One of the prominent features of virtual social
networks is the frequent use of user published
images and videos, a behavior which is becoming
more popular in the last few years. In order for
information to be accessible to all, there are
accessibility guidelines and recommendations, some
of which are specific for images and videos.
Guidelines that focus on audio-visual resources
include those which determine that all non-textual
content must be displayed in an alternative format,
that is, images must come with an alternative text
describing it. (W3C, 2008) . Images accompanied by
text can be understood by the visually impaired, as
the screen reader reads the text.
In social networks which are completely based in
pictures and videos, such as Instagram, accessibility
issues it's a barrier, which can prevent people with
seeing disabilities from using them. For this specific
public, there are two fundamental issues: Instagram
Marques, J., Valente, L., Ferreira, S., Cappelli, C. and Salgado, L.
Audio Description on Instagram: Evaluating and Comparing Two Ways of Describing Images for Visually Impaired.
DOI: 10.5220/0006282500290040
In Proceedings of the 19th International Conference on Enterprise Information Systems (ICEIS 2017) - Volume 3, pages 29-40
ISBN: 978-989-758-249-3
Copyright © 2017 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
29

must be completely accessible, according to
accessibility standards which have already been
published, and the user published images should be
accompanied by a descriptive text which can be read
by a screen reader.
This paper has as an objective to evaluate and
compare two forms of describing images on
Instagram, one through the reading of an image’s
descriptive text by the screen reader, and another
through audio description recorded by the image’s
own author which could be heard through the
execution of an audio file, permitting us to identify
if the use of audio description would encourage this
public to have a greater participation in online image
and video based social networks.
This paper is divided into following fashion:
section 2 presents the theoretical framework and
describes the main concepts involved in this
research; section 3 describes how this research was
planned and executed; the results are analyzed in
section 4 and section 5 discusses this research's
conclusions.
2 THEORETICAL FRAMEWORK
2.1 Virtual Social Networks
Broadly speaking, social networks are any type of
relationship among people, mediated or not by
computerized systems. Such relationships involve
interactions which aim to change people's lives, for
the collective or organizations, since such
interactions can occur for private interests, in
defense of others or in the name of organizations
(Tavares and de Paula, 2015). Among the types and
formats of social networks, those which are
established in cyberspace, denominated “virtual
social networks”, represent a new and complex
universe of communicative, social and discursive
phenomena (Recuero, 2014). For Lévy (Lévy,
1999), a virtual social network “is built on the
affinity of interests, knowledge, of mutual projects,
in a process of cooperation or exchange, all of which
are independent of the geographical proximity and
institutional affiliation.
According to a report produced by the Brazilian
Media Research, in 2015 (SECOM, 2015), almost
half of Brazilians use the Internet regularly. The use
of social networks grows each year: in 2014, the
most accessed social networks were Facebook
(mentioned by 83% of people), Whatsapp (58%),
Youtube (17%), Instagram (12%) and Google+
(8%).
One of the fastest growing social networks in the
world and in Brazil, Instagram, focuses exclusively
on the publishing of images and short duration
videos (Quadros, 2015). Its main objective is the
sharing of this content among its participants. Users
can explore published images and choose the people
or organizations they wish to follow. In this
network, the act of following a person or
organization establishes a bond which represents, at
least, and interest in keeping up with their
publications. Interaction between users can also
happen when one user likes or comments on a
picture or video published by another user.
Instagram was created in 2010 and reached the
milestone of a million users in the first two months
in which it was available on the Apple platform
(Paschoal, 2015). This success is, in great part, due
to the ability of applying filters to pictures before
publishing them, which allows users to simulate a
variety of effects on their images. In 2012 Instagram
was also made available for the Android platform,
where it was installed by 1 million users in just 24
hours. Brazil is the second country with the most
active users in this network, following the United
States (Ribeiro, 2015).
2.2 Instagram Web: Promoting
Accessibility
As a vehicle for communication with the internet,
through which a variety of information is transmitted
to people spread in many regions of the world
(Agha, 2008), the interface of systems and
applications such as Instagram must enable the
access by any person, regardless of their physical-
motor and perceptive, culture and social abilities.
That is, they must be designed in conformity with
accessibility guidelines.
However, obtaining interfaces that meet the
needs of many users is not a trivial task, since there
are a variety of people with distinct limitations. In
order to orientate developers in the elaboration of
accessible systems, there are recommendations and
guidelines, such as the “Web Content Accessibility
Guidelines” proposed by the W3C international
committee, which regulate issues related to the
internet. These guidelines address issues which
hinder the access to websites by users with access
characteristics or limitations (W3C, 2008). These
efforts enabled the Internet to play a key role in the
daily lives of people with disabilities, allowing them
to create new forms of relationships, find job
opportunities and leisure options (Queiroz, 2012).
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
30

In early 2013, Instagram announced the launch
of a Web version so that its users could access the
social network through the computer and not only
through the mobile application. In this Web version,
the user can visualize, like and comment on pictures
and videos. However, in order to maintain its initial
strategy, the publication of new images on Instagram
would still be an exclusivity of the mobile
application. This decision was made to preserve the
application’s basic characteristics. According to one
of its executives, “Instagram is about taking pictures
on the spot, in the real world, in real time” (Stivanin,
2015).
The creation of a Web version for Instagram
opened up new possibilities of use, mainly for those
who have seeing disabilities, as they can use their
computer and screen reading software for access
while not having to depend on a smartphone.
Nevertheless, as sight is the main form of interaction
in this type of network, users with accentuated or
total seeing disabilities require an assistive
technology capable of capturing the interfaces and
making them accessible. Therefore, regardless of
how well designed an interface is, if it's not
accessible, it will be a barrier to the social inclusion
of the visually impaired. Furthermore, these users’
access also depends on the characteristics of these
assistive technologies (Ferreira and Nunes, 2008).
Assistive technology is the term used to identify
any tool or resource (like a cane) which provides or
expands the functional abilities of people with
impairments and thus promotes greater autonomy
(Ferreira and Nunes, 2008). In the case of a person
with accentuated or total visual impairment, Internet
access is possible through screen reading software,
applications associated with voice synthesizer
software, which permits users to navigate the
internet, read and send emails and connect with
other people through social networks, including
Instagram. Consequently, interfaces must be
designed to, when accessed by assistive
technologies, provide easy interactions, capable of
being detected and interpreted correctly.
2.3 Inclusive Instagram: Accessible
Content
The WCAG (W3C, 2008) is organized in principles,
guidelines and testable success criteria. Each
guideline and success criterion has its own specific
techniques to evaluate whether they have been met.
In it, it is described that textual information for any
non-textual content, such as graphical information,
in the case of the visually impaired, or sound
information, and the case of the hearing impaired,
must be provided.
Since Instagram's Web version can be used by
people with seeing disabilities, not only accessibility
to the web site's entire structure must be ensured, it
is also necessary to give special attention as to how
these users can understand the images and videos
published, like them and comment on them. In this
case, one of WGAC’s guidelines of the Principle of
Perception must be observed, as it is directly linked
to accessibility issues for the understanding of
images: Guideline 1.1 Text Alternatives: Provide
text alternatives for all non-textual content so that it
can be presented in different ways, according to the
necessity of the users, for example: enlarged
characters, braille, speech, symbols or a simpler
language.
One of the ways of providing textual information
for any non-textual content, such as graphical
information, in the case of the visually impaired, is
through the “alt” attribute, which provides a textual
equivalent for the images; it adds an alternative text
to the image which is read by the screen reader, thus
providing meaning to the image (Ferreira and
Nunes, 2008).
2.4 Accessible Content – Audio
Description
Audio description is a process which consists of
transforming visual images into words, which are
then spoken during the silent intervals of audiovisual
programs or live performances (Cintas, 2005). The
audio-visual translation technique encompasses
describing images in words, transmitting feelings,
through intersemiotic translations, that is,
translations which consist of the conversion of one
system of symbols into another, translation of verbal
text into a nonverbal text, such as dance, painting,
music, etc. In the view of the Ministry of
Communications of the Brazilian Federal
Government (Ministério das Comunicações, 2006),
audio description is conceptualized as “a narration,
in Portuguese, integrated to the original sound of the
audiovisual piece, containing descriptions of sounds
and visual elements and any additional information
which is relevant to allow a better understanding of
them by people with visual or intellectual
disabilities”.
Audio description is an accessibility resource
which allows people with visual disabilities to view
and comprehend photography, videos, films,
theatrical plays, TV show, exhibits, musicals, among
others (Queiroz, 2012). Through this resource,
Audio Description on Instagram: Evaluating and Comparing Two Ways of Describing Images for Visually Impaired
31

people with seeing disabilities are able to understand
the scenarios, costumes, facial expressions, body
language and several actions which are not presented
during the spoken words of an artistic exhibit.
Audio description was initially done informally,
by people who accompanied those with seeing
disabilities to shows; these would narrate to them
what they could not listen to in the script of the show
or when people with seeing disabilities asked
questions, they answered doubts, during a movie,
theatrical plays or other types of shows (Queiroz,
2012).
This description technique originated in the
United States, in the 1970s, from the ideas
developed by Gregory Frazier in his thesis “Master
of Arts” from the University of San Francisco -
USA, where, for the first time, the term “audio
description” was used. As this resource was being
publicized, it gained space in media, through the
Japanese TV network, NTV, which began to
transmit it's programming in audio description in
1983. Then, Gaberta (Ofcom, 2010) a TV network in
Catalonia, Spain, also did so. The Cannes Film
Festival also joined the idea in 1989. In Brazil, the
first audio-described movie screenings took place
during the São Paulo International Short Film
Festival, during the 2006 and 2007 editions (Silva,
2009).
The audio description can also be used for the
comprehension of videos and pictures (Queiroz,
2012), it can become a tool to facilitate the
interaction of the visually impaired on the Web. As
the use of screen readers demands and requires a
wide use of key combinations and a steep learning
curve, most people with low sight prefer not using it
whenever possible. Therefore, an alternative
resource, like an audio description, which presents
the details of an image in an audio file, can be an
alternative for these people. Likewise, an audio
description can be useful for blind users, since the
figure’s description wouldn't be provided in a
synthetic manner by a reader, but through a pre-
recorded audio, which, if done correctly, can relay
richer information than a synthesizer for someone
with a seeing disability (mild or total).
In order to obtain universal access applications, it
is fundamental to observe and analyze the
difficulties and abilities of users with limitations, as
they guide the mental model used throughout their
interactions with the system. This evaluation enables
harmonious interaction and, at the same time,
guarantees comprehensible and navigable content
(Queiroz, 2012). The participation of users with
limitations assists in the understanding of how they
interact on the Web and use assistive technologies
(Abou-Zahra et al., 2008). Through the observation
of interaction strategies from different users in
distinct contexts and utilizing several assistive
technologies, difficulties faced can be identified
(Melo, 2007), incorporating the experiences of these
groups as users of the system (Slatin and Rush,
2003).
In this way, the evaluation of users interacting
with Instagram will allow the identification of the
barriers they face and a better assessment of their
experiences while interacting with two distinct
technologies, the screen reader or audio description.
2.5 Related Works
The use of audio description has been addressed in
many academic papers, in which authors analyzed
the benefits of this tool for those with seeing
disabilities. Santos (Santos, 2016), for example, has
contributed, in the field of translation studies, for the
development of the researches and has addressed the
use of audio description as mediation in museums.
By means of a case study in the Indigenous Peoples
Memorial, the author proposes reflections on the
implications of the use of audio description in the
experience of a person with a seeing disability in a
museum, aiming to provide access to visual pieces
and create, through the means of verbal language,
conditions for the inclusion of this public.
Villela (Villela and Losnak, 2016) in his turn,
used audio description to depict pictures of the
Military Dictatorship period, helping to keep the
memory of remarkable facts alive for the Brazilian
society, including for people with visual impairment
The objective of this work was the creation of an
accessible photo-documentary about the fifty years
of the Military Dictatorship in Brazil, presenting the
most significant moments of this period for people
with seeing disabilities through the use of audio
description. Additionally, a script was created for the
presentation of previously selected and audio
describe pictures, focusing on historical scenarios
and characters, depicting people’s physical
characteristics in a very objective manner.
Other works are concentrated in proposing and
developing new resources which allow the visually
impaired to more effectively access websites on the
Internet, such as virtual social networks. At the end
of 2015, Facebook announced that it was working on
an artificial intelligence based object recognition
tool in order to help blind users have an idea of the
pictures people shared on Facebook (Dickey, 2015).
The solution consists in processing the image and
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
32

generating a new alternative text which could then
be described for the user through a screen reader.
The engineers involved in the project believe that,
despite not completely describing the images with
all their details, the level of engagement of the
visually impaired could increase.
As in this paper, the Facebook initiative
described above aims to provide the visually
impaired with new opportunities of participation in
social networks, describing the images published by
the users.
The main difference is in the reach and
assertiveness of each method: the use of audio
description that has been recorded by the content’s
own author could present richer details and allow a
greater understanding by those with seeing
disabilities.
3 RESEARCH DEVELOPMENT
This research’s method was qualitative-
observational, (Cresswell, 2009; Denzin and
Lincoln, 2003) based on a case study in the “União
de Cegos do Brasil” (Brazilian Union of the Blind)
institute and involved a public composed of both
young and elderly participants. This study aims to
address the following research question: “How to
evaluate and compare two forms of describing
images published on Instagram: one utilizing the
descriptive text read by a screen reader and another
utilizing audio description recorded by the image’s
own authors?”. The development of this work was
done by two researchers and organized in four
stages: a) test preparation; b) participating user's
profile selection; c) execution of the tests and d)
analysis of the results.
a) Test Preparation: We took into consideration
features regarding free screen readers for computers
running the Windows operating system.
For the realization of the tests, the NVDA
(NonVisual Desktop Access) was chosen for being
familiar to the participants and also for having more
than forty-three language options, including
Portuguese, and for being able to be used from a
USB drive, with no need of installing this reader in a
computer (NVDA, 2014).
The tests were done in the União dos Cegos
institute, a Federal, State, and Municipal public
utility institution, founded in 1924, whose mission is
to ensure that a person with seeing disabilities is able
to reach its potential as a full citizen.
To capture the opinion of the visually impaired
participants, two questionnaires were formulated.
The first questionnaire (pre-test questionnaire)
addressed questions about the profile of each
participant, such as educational level, type of
disability, profession, age, gender and computer
using habits. Regarding computer use, the following
questions were asked: Do you have any experience
with screen readers? Do you use computers to access
the Internet?; Do you know any social networks?;
Do you have an account on any social network?;
With what frequency do you use social networks? If
you did not use any social network, what would be
the reason?
The second questionnaire (post-test
questionnaire) encompassed questions related to the
comprehension of the audio descriptions of the
shown images, such as: How would you grade your
understanding of the image of the 1st test? How
would you grade your understanding of the image of
the second test? Based on the two image description
tests, which was the best for your understanding? To
answer the questions about the comprehension of the
images in both tests, it was necessary for the
participant to attribute a grade in a scale from 0 to
10.
For the realization of the audio description tests,
two images were selected from a public Internet
base. Despite being different, the images should
present a similar context and theme, restricting only
the manner in which they would be described, so
that the participants would influence the results. In
this way, two images were selected (figure 1 and
figure 2), representing a family composed by father,
mother and two children in a moment of leisure.
These images are referenced as “image 1” and
“image 2”.
The redaction of the descriptive texts of each
image followed the same style and format. The full
text of each image is reproduced below.
Audio Description on Instagram: Evaluating and Comparing Two Ways of Describing Images for Visually Impaired
33

Figure 1: Image described by the screen reader (image 1).
Descriptive text of Figure 1: “A family is playing
on a green lawn. The father and his son 8 years-old
are standing. The son is flying at colorful kite and
the father proudly watches it fly. The mother is
sitting on the grass with her younger daughter on her
lap, watching father and son play. The day is lovely,
with very blue skies, no clouds, and they all seem
very happy”.
Figure 2: Image described through audio description
(image 2).
Descriptive text of figure 2: “A family is walking
along the sand on a beautiful beach. The mother
carries her youngest son on her shoulders. The father
is just behind and plays with his other son, throwing
the boys up in the air to catch him soon after. The
seawater is quite blue with a few small white waves.
On the horizon, there's a mountain with green trees
and a few houses. They're all smiling and seem quite
happy”.
At first, one of the limitations of this research
could be the fact that the audio descriptions were
produced by the researchers themselves; they were
not made by professionals specialized in audio
description. But the choice to make them in a
personalized way was deliberate, as, if applied to
Instagram, they would be generated by the users
themselves. The recording of the audio description
was made by a collaborator. She read the descriptive
text in Figure 2, in natural speech, respecting the
pauses predicted on the text’s punctuation.
Then, two local copies of the Instagram
website’s HTML files were made, maintaining all
their layout and visual identity. One of the copies
was modified to present “image 1” and its
descriptive text, as if it were a normal user
publication, and the other copy received “image 2”,
but inserting below it, a button that, when pressed,
would play an audio with the recording of their
descriptive text made by human collaborator.
b) Profile definition and selection of participants:
In order to participate in the research, the participant
would have to be an adult over 35 years old, and
have severe visual impairment. It was decided not to
pre-screen the participants as to avoid their
commenting to one another, which could influence
the results of the research. The participants were
invited according to their availability in the activities
and programming at the União dos Cegos do Brasil
institute.
The answers from the pre-test questionnaire
revealed that 50% of the blind declared to have
completed High School. Of those with low vision,
33% declared to have higher education and 70%
declared to have finished Elementary School. As for
the disability, 60% of the respondents had a type of
severe visual impairment (total or low vision) that
consists of the total lack of visual perception of any
type of light. 80% of the participants reported to
have made use of a computer for Internet access.
Concerning screen readers, 67 % of participants
utilize this type of software and the other 33% are
aware of it. All of them reiterated that they have
heard of social networks such as: Facebook,
Instagram, Twitter, Whatsapp. However, 33% have
never accessed any of these social networks. 67% of
participants access social networks twice a week, on
average, and are mostly female. Regarding the
profession of these participants, they were the most
varied (pensioner, retirees, early education teachers
and the medical area).
For the sake of maintaining the anonymity of
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
34

these participants, we sought to preserve their
names, which were coded as: P1, P2, P3, P4, P5 and
P6. The profession, age, type of disability and
computer use of the participants are illustrated in
Table 1.
Table 1: Code, Profession, Age, Visual Disability,
Computer Use of Participants.
Code Professional Age Visual
Disability
Computer
Use
P1 Pensioner 35 Total No
P2 Retired 45 Total Yes
P3 Doctor 59 Low Vision Yes
P4 Retired 67 Total No
P5 Retired 51 Total Yes
P6 Teacher 68 Low Vision Yes
c) Execution of the Tests: All of the research’s
details, its objective and mainly the benefits that
could be expected from this work were explained to
the institute's coordinators by the researchers. The
coordinators requested that the duration of the test
did not exceed three hours in total so as not to
compromise the participant’s programmed activities
for the day. A room was made available for the
research team, containing two laptops (one for the
execution of the tests, another for support), sound
speakers and headphones.
As they arrived at the institution, the participants
were directed in pairs to the test room by one of the
institutions coordinator. This type of approach (in
pairs) was a request of the institutions coordinators
for the realization of the tests. In the test room, the
participants received the initial information from
both researches so they would understand clearly
what would be done and what was expected from
each one of them. Besides ensuring that all
participants would receive the same information in a
standardized fashion, the objective of this initial
explanation was to reassure the participants and
make them more comfortable during the test run.
The first activity was the application of the pre-
test questionnaire. The questions, whose objective
was to collect each participant's profile information,
were read by one of the researchers and the answers
written down on printed forms.
In the second activity, each participant listened to
the text description of “Image 1” by the NVDA
screen reader. Then, before answering or making
any comments, the participant listened to the text
description of “Image 2” through the execution of an
audio description which had been previously
recorded by a human collaborator. The choice of the
description being made first by the screen reader
was due to the fact that 67% of the participants were
already familiarized with it and that it had already
been extensively studied in the literature. At any
moment the participants were informed about which
description method was being used.
Lastly, after listening to the image descriptions,
the post-test questionnaire was applied. Once again,
the researcher read the questions to the participants
and wrote down the answers on printed forms.
During the test run, the main impressions,
difficulties and reactions from the participants were
registered, and are described below.
The participant P1 reported that after listening to
the audio description of image 2, she had the
sensation of being part of the scene, since the
intonation put by the human voice was very real, as
if the family described in the image was by her side;
If she had to opt between the two descriptions, she
would opt for the audio description. It was noticed
that this participant had no greater difficulties in
carrying out this test. As far as emotional reactions
are concerned, it can be said that she was very
secure, determined and alert in carrying out the test
and, at the end, she still said: “it's already over!”
For participant P2, who wasn't used to screen
readers, there was no difficulty in doing the test,
even while being tense, determined and attentive. At
the end of the test, she said that application
developers should be more concerned about building
tools with audio description, aiming to include the
visually impaired who are largely forgotten by that
professionals, thus ratifying her choice for audio
description.
Participant P3 was the most enthusiastic about
taking part in the tests. Before starting, she
mentioned that she loved screen readers and audio
description and asked: which movie are you going to
show us? Even though she had been oriented on how
the tests would be performed, that she would not be
shown a movie, she did not lose her good mood and
determination when she realized it was not a movie.
So, good mood was another defining trait of this
participant. There were no noticeable difficulties in
the handling of the equipment during the tests, as
she felt very secure with them. According to her, the
audio description was so real, clear, enriched in
details, as, for example, the sound of the waves, that,
if she could, she would like to be able to play with
the couple's children. She concluded her
participation by saying: “audio description can be
seen as a way of the visually impaired getting to
know a world which can't be seen or explored by
many”.
Volunteer P4 was somewhat tense in the
expectation of what would happen, but was
Audio Description on Instagram: Evaluating and Comparing Two Ways of Describing Images for Visually Impaired
35

determined to conclude the tests. For this voluntary
there was no difference between the descriptions of
images 1 and 2 made by the screen reader and the
human voice, even though she was not used to
working with screen readers. When asked about
which description she would prefer, she said she
would opt for the audio description, as it more
closely depicts the reality of the facts to the
spectator.
Impressions and reactions of participant P5 drew
the most attention, as it was the participant who had
lost almost all sight five years before (he became
blind at age 46) due to complications of a glaucoma
caused by diabetes mellitus. He had graduated in
programming and was very determined, enthusiastic
and attentive during the tests; he distinguished
himself from the others by his professional
experience in handling computer equipment and
social networks and, therefore, did not have any
difficulties during the tests. Regarding the
description of the images, he said the audio
description was far superior when compared to the
one made by the screen reader. He reiterated that
more effort and investment should be made so that
software developers could build more tools that use
audio description. According to him, if there were
more investments in audio description tools, the
social inclusion of the visually impaired would be
better promoted.
The last participant, P6, was tense but attentive
to instructions and handling of the equipment. When
asked about the best image description, after the
realization of the test, she answered that she would
prefer the screen reader’s description, even though
she was not used to working with them. However, in
her opinion, the description of the images utilizing
the audio description tool could stimulate the use of
social networks.
The time established by the research method for
the realization of each participant's tests was 15
minutes. On average, the tests lasted approximately
10 minutes, which included: the objective of the
research; the profiling of the participants; the
understanding of the method utilized in the two
image descriptions; the questionnaires (pre-test and
post-test); and the listening of the descriptions.
At the end of the data collection, the information
was consolidated and analysed.
Research limitations: One of the limitations of
this research was the fact that only images were
analyzed. No work was done regarding videos. One
of the accessibility recommendations determines that
all real time (live) or pre-recorded audio and/or
video content, must be made available through
alternative content which presents transcribed or
described information.
4 EVALUATION OF TEST
RESULTS
For the purpose of result analysis, it is possible to
divide participants in two groups according to their
level of experience with screen reading software.
Out of the six participants, three had some
experience with screen readers and the other three
didn't have any contact with this type of software. Of
the participants who had some experience with
screen readers, P1, P3 and P5 are highlighted. P1
revealed that she has been using this type of
software for over 10 years. P3, who had lost her
vision when she was young, reiterated that she has
been using the screen reader for over 15 years. P3,
who has not been able to see for over 5 years, began
using this application after losing her sight.
The first two questions of the questionnaire had as an
objective make a direct comparison between the two
methods of image description used in the test. Figure 3
illustrates the grades attributed by the participants
regarding the description made by the screen reader, on a
scale from 0 to 10.
Figure 3: Grades attributed by the participants regarding
the description made by the screen reader.
The lowest grades for the description were given
by the participants of the group that didn't have any
prior experience with screen reading software. When
restricted to this group, the average grade for the
description falls to 6.3. According to the
participant’s own comments, the frequent use of
screen reading software increases the comprehension
level of what is listened to during computer use.
This justifies the higher grade given by the group
which had experience with screen readers. In this
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
36

group’s opinion, the average grade for the screen
reader’s description was 9.7.
Figure 4 shows the grades given by the
volunteers in regards to the audio description, on a
scale from 0 to 10.
Figure 4: Grades attributed by the participants to the audio
description.
When asked about which of the two methods
favored a better understanding of the images, only
one of the participants opted for the screen reader’s
description. Every other participant thought image
comprehension was better through the use of audio
description, which represents 83% of total
participants.
It is worth mentioning that, in the group which
had prior experience with the screen reader, two
participants had given grade 10 to both the screen
reader’s description as well as the audio description,
that is, they had classified both methods the same
way. However, if they had to choose between one of
methods, they would choose the audio description.
The last question of the post-test questionnaire
had as an objective to understand whether, in the
opinion of participants, it the audio description to
describe imagens would encourage the visually
impaired to use social networks. All the participants
answered yes to this question, resulting in a 100%
approval rating.
The graph of Figure 5 illustrates the participants'
perceptions of emotions, difficulties, and
impressions during the tests, where enthusiasm,
determination, attention, tension, and safety
regarding the use of the screen reader were recorded.
From the data in this figure, it was verified that
31% of the participants were determined, that is,
they were convinced that they could carry out the
tests. The visually impaired who had attention
during the experiment represent 25%. The facial
expressions regarding tension could be observed in
19% of the participants. The perception of
enthusiasm depicted 12% of users. And the safety
during the tests was perceived in 13% of the
volunteers.
Figure 5: Perception of participant’s emotions.
With respect to the question “do you know any
social networks?”, the participants mentioned they
knew the following social networks: Facebook,
Instagram and WhatsApp. Without exception, they
mentioned their knowledge of Facebook. Figure 6
illustrates the percentage of the participant’s
knowledge of these social networks.
Figure 6: Knowledge of social networks.
When analyzing the use of social networks it was
found that not all participants access them.
Comparing the participants in relation to the use of
this entertainment channel on the seven days of the
week, it was noticed that only two of the participants
accessed this type of channel. Participants P1 and P4
were not considered as they do not access any social
network.
Regarding the voice of the screen reader, the
majority of the participants were of the opinion that
it was a very computerized, synthesized voice, and
that for a better understanding and clarity of speech,
the screen reader would have to be well configured,
as voice quality is determined by its similarity to the
human voice.
Audio Description on Instagram: Evaluating and Comparing Two Ways of Describing Images for Visually Impaired
37

As it was mentioned in the previous section, the
two participants that dealt with screen readers almost
daily, were the ones who most questioned the choice
for the synthesized voice for the realization of the
test, as in their opinion, the voice was not very
appropriate for use in the image description.
Participants P1 and P4 declared that they were
considered to be people of “low-income”, who
depended on the technological resources available at
the União dos Cegos institute for internet access and,
consequentially, social networks. As such, they need
to commute from their residences to the institute,
aiming at socialization and, therefore, digital
inclusion. Participant P1, who is a pensioner and
depends on the government's financial resources for
her livelihood, said: “A computer could be a
Christmas gift”. The retired participant, P4,
reiterated that she does not have the financial
conditions to buy a computer, and is not able to
connect to social networks. For these participants,
access to social networks would open new forms of
interaction and communication, gradually decreasing
their digital exclusion, as well as enriching their
studies.
5 CONCLUSIONS
The objective of this paper was to evaluate and
compare two forms of describing images on
Instagram, one through the reading of an image’s
descriptive text read by the screen reader and
another through an audio description recorded by the
picture’s own author, which is heard through the
execution of an audio file. Through the realization of
tests involving a group of people with seeing
disabilities, four with total impairment and two with
low sight, it was possible to obtain important
information about the participant’s preferred method
of image description and if the inclusion of audio
description resources on Instagram could encourage
the participation of people with visual impairment.
The analysis of the data collected during tests
shows that the use of audio description allowed
better image comprehension. The fact that the audio
description of an image is narrated by a human (the
speech of the screen reader is created by a sound
synthesizer, which sounds somewhat artificial), was
fundamental for the understanding, resulting in no
difficulty of comprehension by the participants.
Even among participants who already had previous
experience with screen readers, the audio description
was chosen as the best option. All the participants
stated that having the possibility to listen to an audio
description of an image that has been recorded by its
own author (giving a greater personal focus to the
content), would increase the participation of the
visually impaired on Instagram, which, as it is
completely image and video based, is currently
barely inclusive for this public.
As it has been demonstrated throughout this
article, audio description has shown itself to be an
excellent tool for the inclusion of the visually
impaired, permitting greater access and participation
in cultural and leisure activities and education.
Furthermore, accessibility standards for Internet web
sites help developers make them accessible, ensuring
access to all, including people who have some type
of the visual disability. The results of this paper
show that the use of audio description, allied to the
fulfillment of accessibility requirements, can be
decisive for these people’s access to image based
social networks, such as Instagram.
As in the description made by screen readers –
Which depend on the production of text or subtitle
that explains the image - the use of audio description
on Instagram would also depend on the collaboration
of users who publish pictures, as they would be
responsible for recording the audio description of
their own images.
In future research, besides evaluating possible
accessibility limitations in Instagram's WEB version,
it would be important to study the modifications and
new functionalities that would be necessary to be
able to implement the correct use of audio
description on Instagram.
In the application for smartphones, for example,
new functionalities could be created which allow
users to record the audio description in a quick and
simple fashion. Currently, the publication of images
on Instagram is done exclusively through
smartphones, which already offer hardware and
software tools for audio recording. As such, a user
could publish a picture on Instagram and, shortly
after, record the audio description with their own
voice on their own devices.
As for Instagram's WEB version (referred to in
this paper as an opportunity for the visually impaired
to access this network), modifications should be
made in order to offer new audio description
resources. In this case, the focus would be on
offering users forms of searching and identifying
images which have audio description and allowing
users to listen to them. Additionally, it would be
interesting to create a new form of interaction in
which the visually impaired user could send a
request to the author of an image so that he would
record an audio description, in case it had not been
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
38

made yet. Besides being a form of increasing the
volume of audio described images, this resource still
establishes a new form of contact between the
visually impaired and other users on Instagram.
New studies will be done to efficiently plan and
define the set of changes on Instagram’s systems,
materializing the benefits of the use of audio
description for the visually impaired pointed out in
this research.
Another aspect that could be explored is the
interest of volunteers in participating in social
networks using audio description.
REFERENCES
Abou-Zahra, S., Bjarno, H., Duchateau, S., Restrepo, E.,
Henry, S., McGee, L., Pouncey, I., Rush, S., Sutton, J.
and Wassmer, S. (2008) Evaluating Websites for
accessibility: Overview ◦ web accessibility initiative ◦
W3C. Available at: http://www.w3.org/
WAI/eval/Overview.html (Accessed: 15 December
2015).
Agha, G. (2008) Computing in pervasive cyberspace.
Proceedings of the ACM Communications of the
ACM, 51, 1.
Cintas, Jorge Díaz (2005), Audiovisual translation today: a
question of accessibility for all. Translating Today,
London, n. 4, p. 3-5, July 2005.
Cresswell, J.W. (2009), Research Design: Qualitative,
Quantitative, and Mixed Methods Approaches. 3rd
Edition. Thousand Oaks: SAGE Publications.
Denzin, N.K. and Lincoln, Y.S. (2003) The landscape of
qualitative research: Theories and issues. Thousand
Oaks, Ca. Sage Publications, Inc.
Dickey, M.R. (2015) Facebook’s working on A tool to
help the blind ‘See’ images. Available at:
http://techcrunch.com/2015/10/13/facebooks-working-
on-a-tool-to-help-the-blind-see-images (Accessed: 14
May 2016).
Ferreira, S. B.L e Nunes, R (2008). e-Usabilidade. Rio de
Janeiro: LTC Editora.
IBGE (2010). Sala de imprensa | notícias. Available at:
http://www.ibge.gov.br/home/presidencia/noticias/noti
cia_visualiza.php?id_noticia=2125&id_pagina=1
(Accessed: 14 May 2016).
Lévy, P. (1999). Cibercultura, São Paulo, Editora 34.
Coleção Trans.
Melo, A. (2007) Design Inclusivo de Sistemas de
Informação na Web. Doctoral Thesis, Universidade
Estadual de Campinas, Instituto de Computação,
Campinas.
Ministério das Comunicações. Portaria nº 310, de 27 de
junho de 2006. (2006) Available at:
http://www.anatel.gov.br/legislacao/normas-do-
mc/442-portaria-310 (Accessed: 15 October 2015).
NBR 9050. (1994). NBR 9050 Associação Brasileira de
Normas Técnicas. Acessibilidade de Pessoas
Portadoras de Deficiências a Edificações, Espaço,
Mobiliário. Rio de Janeiro: ABNT.
Nicholl, A. (2001). O Ambiente que Promove a Inclusão:
Conceitos de Acessibilidade e Usabilidade.
Assentamentos Humanos Magazine, 3, 2.
NVDA (2014). Manual do Utilizador do NVDA 2014.3.
Available at: http://www.nvda.pt/files/html/
userGuide.html (Accessed: 20 November 2015).
Ofcom. Guidance on standards for audio description.
(2010). Available at: http://www.ofcom.org.uk/
static/archive/itc/itc_publications/codes_guidance/audi
o_description/introduction.asp (Accessed: 18
November 2015).
Paschoal, Mariana. Instagram: O Aplicativo que
Revolucionou o Mundo da Fotografia (2015).
Available at: http://blog.emania.com.br/instagram-
revolucionou-o-mundo-da-fotografia (Accessed: 20
November 2015).
Piovesan, S. D., Wagner, R. and Rodrigues, L. (2013),
Acessibilidade em redes sociais: em busca da inclusão
digital no Facebook. Informática na educação: teoria
& prática, 2013.
Quadros, Yves. 6 redes sociais para prestar atenção em
2015 (2015). Available at: http://www.
8020mkt.com.br/6-redes-sociais-para-prestar-atencao-
em-2015 (Accessed: 28 November 2015).
Queiroz, M. A. (2012). Bengala Legal. Available at:
http://www.bengalalegal.com (Accessed: 16 May
2016).
Recuero, Raquel (2014). Contribuições da Análise de
Redes Sociais para o estudo das redes sociais na
Internet: o caso da hashtag# Tamojuntodilma e#
CalaabocaDilma. Fronteiras-estudos midiáticos 16.2:
60-77.
Ribeiro, Igor. Instagram: 29 milhões de usuários no Brasil
(2015). Available at: http://www.meio
emensagem.com.br/home/midia/2015/11/09/instagram
-chega-a-29-milhoes-de-usuarios-no-brasil.html
(Accessed: 28 November 2015).
Santos, L. D. S. (2016). Audiodescrição em museus: a
experiência em acessibilidade no memorial dos povos
indígenas.
SECOM. Secretaria de Comunicação Social da
Presidência da República. Pesquisa Brasileira de
Mídia 2015 (2015).
Silva, M. (2009). Com os olhos do coração: estudo acerca
da áudio descrição de desenhos animados para o
público infantil. 218f. Dissertação (Mestrado em
Letras e Lingüística) –Universidade Federal da Bahia,
Salvador.
Slatin, J.,Rush,S. (2003) Maximum Accessibility: Making
Your Web Site More Usable for Everyone.
Massachusetts: Addison-Wesley.
Stivanin, T. (2015) Fato em Foco - em 2015, Instagram se
consolida como umas das redes sociais mais populares
do Brasil. Available at: http://br.rfi.fr/geral/20150105-
em-2015-instagram-se-consolida-como-umas-redes-
sociais-mais-populares-do-brasil (Accessed: 28
November 2015).
Audio Description on Instagram: Evaluating and Comparing Two Ways of Describing Images for Visually Impaired
39

Tavares, Wellington, e Ana Paula Paes de Paula (2015).
Movimentos Sociais em Redes Sociais Virtuais:
Possibilidades de Organização de Ações Coletivas no
Ciberespaço.
Villela, L. M., & Losnak, C. J. (2016). Abrindo os olhos
sobre a Ditadura Militar: audiodescrição como recurso
de manutenção da memória brasileira. Cadernos de
Tradução, 46-65.
W3C (2008) Web content accessibility guidelines
(WCAG) 2.0. Available at: http://www.w3.org/
TR/2008/REC-WCAG20-20081211 (Accessed: 28
November 2015).
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
40