Buletinul Ştiinţific al Universităţii Politehnica Timişoara

Seria Limbi moderne

Scientific Bulletin of the Politehnica University of Timişoara
Transactions on Modern Languages
Vol. 22, Issue 1 / 2023

Classification of Automatic Subtitling Tools.

A Proposal
Cristina VARGA*

Abstract: This article aims to provide a deeper understanding of the automatic subtitling tools
that are currently being used and of their context of application. Based on the assumption that
AVT, in general, and subtitling, in particular, are undergoing fundamental changes, it is our
aim to analyse the range of tools that allow AVT translators to enhance their productivity and
their efficiency. For this purpose, we have analysed 40 different automatic subtitling tools,
currently available and accessible on Internet. Through this analysis, it has been possible to
observe the main features of these tools and observe their functioning. Therefore, different
criteria have been established in order to systemize this extensive inventory based on which 23
categories of software dedicated to automatic subtitling have been identified. These categories
have been illustrated with examples. In this study, the aim is to provide a more accurate and
more systematic understanding of automated subtitling programs. The paper is addressed to
AVT professionals as well as to teachers and students having an interest in the state-of-the-art
of automated subtitling.

Keywords: automatic subtitling, live captioning, classification of automatic subtitling tools,

automatic transcription, real time speech recognition, machine translation, automatic spotting

1. Introduction
Nowadays, with the rapid development of artificial intelligence, professionals in
many fields may have to adapt to new challenges given its significant impact within
their work.
Over the last few decades, with the widespread use of multimedia
communication (Dejica et al. 2020) and due to the need to provide rapid access to its
linguistic content, artificial intelligence (AI) has become increasingly involved in
AVT. The reason often given by companies for implementing automatic subtitling
within this context is that human translators specialized in AVT are not able to cope

*Senior lecturer Cristina VARGA, Applied Modern Languages Department, Faculty of Letters, Babeș-
Bolyai, University, Romania, cristina.varga@ubbcluj.ro
79

BUPT
with the volume of translations required inside a company, and that training of
subtitlers and their work involves very high costs.
Despite the fact that it has been less addressed by researchers, automatic
subtitling has developed significantly as a result of increased influence coming from
industry, Internet and social media, being mostly used by the mainstream audience
and some companies.
The present paper proceeds from the premise that the field of ATV in general,
and subtitling in particular, is now undergoing thorough transformation due to the
involvement of AI, this change has been reflected through the wide range of
automated captioning tools that are now at the disposal of the general public and AVT
professionals. This increasing involvement of automated translation in AVT, rather
than being a threat towards professionals, enables them to enhance their IT skills in
order to become more productive and efficient.
For this reason, this article aims to answer the following research questions:
a) which is the inventory of automatic subtitling tools currently accessible to the
general public and professionals; b) which are the criteria by which these programs
can be classified; c) which categories of automatic subtitling programs exist, and d)
which are their general characteristics.
In order to do this, we started by providing a timeline of the evolution of
automatic subtitling tools, enabling us to understand the wide variety of programs that
have been created and used in this field over the last decades in many different
countries, and involving subtitling from and into many different languages.
Furthermore, the timeline has been segmented into several stages of development
enabling us to focus the analysis on the time period that best illustrates the latest
developments in these tools, meaning the period 2015-2023.
For this purpose, a list of 40 automatic subtitling programs was created which
were analysed and ranked according to multiple criteria. Through this approach, it is
our aim to systemize and provide empirically based knowledge about the different
categories of software that AVT professionals need to know in order to optimally
respond to market demands.
The present article is aimed at professional AVT translators, researchers,
teachers and students who may want to deepen their understanding of the changes that
artificial intelligence brings to the subtitling practice which they can use for their own
benefit.

2. Automatic subtitling. Diachronic developments

Although it has only gained greater visibility in recent years, automatic subtitling is a
field that has been evolving for over two decades. Due to the increasing pressure of
workload in this field (Karakanta et al. 2020, 3727), the audiovisual industry, social
media and social marketers engage in the development of programs that would allow
a more efficient subtitling work. Thus, since 1984, it becomes more and more
frequent to attend academic conferences where computer applications and projects
aimed at optimising translation work in AVT are presented. Early attempts at
80

BUPT
automation were designed to assist the subtitler at certain stages of his/her work, such
as automatic transcription based on automatic speech recognition (ASR). As an
example, Damper&Lambourne& Guy (1985) propose the automation of live subtitles
using ASR as a complementary element of the keyboard. Another ASR-based
subtitling application is the VOICE project (Pirelli 2004, 25-29), whose subtitling
application and results are presented at a conference in Vienna, in 1998. The program
is also used in some schools in Italy (Pirelli 2004, 28) and the project develops until
2003 with outstanding results.
The project LINK, also based on ASR, has been developed since 1998 in the
UK. It is the result of a partnership between the University of Hertfordshire and
SysMedia Ltd (Lambourne et al 2004, 270-271). Its aim was also to create real-time
subtitles for television programmes. The authors mention that SpeakTitle, the
application created as part of the LINK project, has been used for subtitling sport
events and live broadcasts (Lambourne et al 2004, 270-271).
The use of automated real-time subtitling systems is starting to widespread
worldwide, for example, the Japanese news programme "News7" of NHK has been
using an ASR-based simultaneous subtitling system since March 2000 (Ando et al.
2000, 195). Since 2002, EU interest in AVT automation has materialised through the
funding of research projects. For example, the MUSA: MUltilingual Subtitling of
multimedia. A content project (https://www.esat.kuleuven.be/psi/projects/archived
/musa), which combines several technologies such as ASR, MT and NLP, is running
from 2002-2005. The automated subtitling software developed in the project is
multilingual and works for English, French and Greek.
Starting with 2003, real-time subtitling systems are being implemented at
national level in different countries. This includes the ASR-based Live Subtitling
System implemented by the British BBC Television to streamline the work of
subtitling broadcast audiovisual content (Marks, M. 2003, 4). In the USA, the
TranslateTV system (www.translatetv.com) has also been in use since 2003 for real-
time Spanish subtitling of television programmes. The same year, AUDIMUS.media,
an ASR-based system for subtitling news programmes in Portugal, was developed
(Meinedo et al 2003, Neto et al 2008). This system is able to identify specific
information in multimedia files, perform audio segmentation, use domain detection
techniques and perform automatic transcription of the audio/video stream.
Between 2004-2006, EU supports another real-time automatic subtitling
project, namely eTITLE which brought together TV producers from the UK, Spain
and the Czech Republic. This is a complex system whose importance lies in the fact
that it integrates MT in the field of subtitling. An innovative element for its time.
Also, eTitle could generate multilingual subtitles, the language pairs for which it was
developed being English-Spanish, Spanish-English, English-Czech, Catalan-Spanish,
Spanish-Catalan English-Catalan, and Catalan-English (Melero et al 2006).
As of 2006, in Canada, in Montreal, a system for automatically generating
French subtitles (Boulianne et al. 2006) is being used in numerous events with large
audience participation (Boulianne et al. 2008, 199-200).
81

BUPT
 Automatic subtitling is also being implemented in Chinese, in the paper by
Gao et al. (2008, 576-577) it is mentioned that first steps have been taken to develop a
text and speech synchronization system for news programs.
The Aragonese Public Television was using an automatic real-time news
subtitling system based on ASR in 2008. However, the unexpected feature of this
system is that it works with satisfactory results without human intervention Ortega et
al. (2009, 2095-2098). Also, in the Spanish space, in 2010, the ApyCA project is
mentioned, which has been developed with the aim of making the translation process
more efficient and faster. The project is being developed due to pressure from the
Spanish audiovisual law which required that by 2013 all television stations must
provide subtitles for more than 90% of the programmes they broadcast (Álvarez 2010,
567). Later, in 2010, in the Czech Republic, a programme for subtitling parliamentary
sessions is being developed (Trmal et al. 2010).
From 2011-2014, automatic subtitling tools are being developed in projects
such as SUMAT (Subtitling by Machine Translation), a multilingual project focused
on 9 European languages (www.fp7-sumat-project.eu/about-us/index.html) and
Translectures (Transcription and Translation of Video Lectures), a multilingual
project dedicated to automatic transcription and translation of educational video
materials (https://www.mllp.upv.es/projects/translectures/) currently hosted by the
Polytechnic University of Valencia.
Between 2012-2015 the EU-BRIDGE platform (https://project.eu-bridge.eu/)
is being developed, providing solutions for many situations where subtitling is
needed, such as: TV broadcasts, translation of academic lectures, translation in the
European Parliament and subtitling for online events (conferences, webinars, etc.).
In 2012, the SAVAS - Sharing AudioVisual language resources for Automatic
Subtitling project is being created by a research group of TV representatives,
subtitling companies and software developers. The aim of the project was to develop
a multilingual automatic subtitling program (Basque, Spanish, Portuguese, Italian,
French, German,) to be used in TAV for news programmes, interviews, debates and
sports broadcasts (Pozo 2014, 432).
As may be seen, most of these tools are either supported by broadcasters or
are developed in an academic setting with EU funding and generally involve
subtitling for TV, conferences, and for events such as: courses/seminars/webinars.
As may be seen, most of these tools are either supported by broadcasters or
are developed in an academic setting with EU funding and generally involve
subtitling for TV, conferences, and for events such as: courses/seminars/webinars.
One of the first automatic subtitling tools for social media was implemented in 2006
for the platform YouTube. Named YouTube Automatic Caption, it allows users to
automatically generate captions for a video file stored in their account (Harrenstien,
2006; 2009). This system was improved in 2013 by implementing a Deep Neural
Network-based speech recognition system (Hank Liao et al., 2013).
In order to provide a comprehensive overview on the evolution of automatic
subtitling tools, we add the diagram (fig. 1) which will allow to illustrate more clearly
82

BUPT
the extent of the process. Many of the programs developed between 1998 and 2015
constitute the basis of the online automatic subtitling applications and platforms that
exist today.

Fig. 1 Overview of the development of automatic subtitling (1984-2015)

All these tools represent only milestones in the evolution of automatic

subtitling, nowadays existing a wide inventory of such tools with various degrees of
complexity.
Depending on the subtitling software being used, the organization of the
workflow and the output obtained may be different. It can also be remarked that these
tools have a rather short useable life, some of them disappearing from the virtual
space after a few years of operation.
There is a possibility that this phase of rapid development of subtitling
platforms that we are currently witnessing will be transient, and that the next phase of

83

BUPT
evolution of these programmes will focus on developing fewer subtitling tools but
with a higher quality a complexity of the services they offer to users.

3. Classifications of automatic subtitling tools

Nowadays, it is rather difficult to assess the variety and the typology of automatic
subtitling tools. Over the last two decades they have developed in different directions,
seeking solutions for different specific AVT situations, and have often developed in
parallel. For this reason, it is now difficult to determine accurately what place each of
these tools hold in the heterogeneous inventory of automatic subtitling programs.
Thus, given that many of them remain poorly known to the general public and to
specialists, we believe it is important to establish an overview of them, to analyse
them and to establish their typology. This will provide a better understanding of the
tools involved in automatic subtitling and will allow us to know more about their
structure, their functioning, and about the range of services they offer.
Nowadays, it is rather difficult to assess the variety and the typology of
automatic subtitling tools. Over the last two decades they have developed in different
directions, seeking solutions for different specific AVT situations, and have often
developed in parallel. For this reason, it is now difficult to determine accurately what
place each of these tools hold in the heterogeneous inventory of automatic subtitling
programs. Thus, given that many of them remain poorly known to the general public
and to specialists, we believe it is important to establish an overview of them, to
analyse them and to establish their typology. This will provide a better understanding
of the tools involved in automatic subtitling and will allow us to know in more detail
the structure, the functioning, and the services these tools have to offer. This
knowledge is particularly important for AVT professionals who need to have the
required skills to select the optimal working tool and to make the appropriate
decisions in AVT.
Both the literature and our analysis of automatic subtitling tools indicate that
the evolution of this field has taken place in a number of stages, each of which
presents its own characteristics and has influenced the development of AVT tools.
Considering these inputs, we propose the following stages of automatic subtitling
development:
a. the beginnings (1984-2000);
b. development (2000-2015);
c. widespread use and diversification (2015-present);
d. standardisation and specialisation of software (forthcoming stage of
development).
Thus, the first stage is characterised by the emergence and development of
theoretical models describing automatic subtitling and/or the automatization of some
stages of the process. There are no automatic subtitling tools as such yet, but work is
underway to implement some of their components. Even though it is known that ASR
technology is essential to automate subtitling, there are still steps towards its
development.
84

BUPT
 In the second stage, the evolution is characterized by the automation of the
whole subtitling process. The first automatic subtitling tools and research projects in
this field are presented at conferences and attempts are made to promote their use in
schools. A number of countries develop automatic subtitling tools which are
successfully used by national TV stations. Software is beginning to diversify and the
question of automatic subtitling in new contexts such as social networks is raised.
We are currently at the third stage of the evolution of automatic subtitling.
During this stage, the technological evolution of automatic subtitling tools is
influenced by AI, which improves the quality of both ASR and MT. The use of
automatic subtitling platforms become widespread, especially in the use of automatic
subtitling programs for social networks. Browsers use live subtitling software. Thus,
automatic subtitling is no longer limited to the needs of TV companies, it is no longer
carried out only by translators specialised in AVT. Virtually anyone can use the
services offered by these automatic tools in case of need.

3.1. Overview of Automatic Subtitling Tools

In order to better understand automatic subtitling, an overview of the most important
tools and their systematic analysis is a first step. For the identification of these tools,
relevant literature was consulted and a number of searches were carried out in Internet
using a search engine and advanced search operators. This allowed us to create a list
of 40 automatic subtitling tools of different types. This first list used for analysis was
enriched during the entire duration of the investigation.

1. Ai-Live Captioning - https://www.ai- 21. Live Caption Chrome

live.com/ 22. Maestrasuite – https://maestrasuite.com/
2. Animaker - 23. MateSub - https://matesub.com
https://app.animaker.com/subtitle/ 24. Media.io - https://www.media.io
3. ASG - https://www.autosubgen.com/ 25. Ooona - https://ooona.net/
4. AudioType - 26. Otter.ai - https://otter.ai/
https://www.audiotype.org/subtitles/ 27. Rev – https://www.rev.com/checkout/capti
5. Auto-Subtitle Firefox extension onfiles
6. CapCut - https://www.capcut.com/ 28. Simplified - https://simplified.com/video-
7. Checksub - https://www.checksub.com/ editor/auto-subtitle-generator
8. Clideo - https://clideo.com/add-subtitles-to- 29. Subtitle video - https://www.subtitlevideo.
video com/auto-subtitle-generator-online.php
9. Diy captions - https://www.diycaptions.com 30. SubtitleBee - https://subtitlebee.com/
10. EasySsub - https://easyssub.com/ 31. Transkiptor - https://transkriptor.com/
11. Edge Live Caption Feature 32. Veed.io - https://www.veed.io
12. EVE - https://starteve.ai/ 33. Wave.video - https://wave.video/video-
13. Facebook’s Automatic Caption Feature captioning
14. Filmora - https://filmora.wondershare.com 34. Wavel.ai - https://wavel.ai/studio/auto-
15. Flexclip - subtitles-generator/
https://www.flexclip.com/tools/auto- 35. Wearenova - https://wearenova.ai/nova-
subtitle/tomatic-subtitles/ tools/au
16. Flixier - https://flixier.com/ 36. Webcaptioner - https://webcaptioner.com/
17. HappyScribe - 37. YouTube Automatic Caption
https://www.happyscribe.com/ 38. Zubtitle - https://zubtitle.com
85

BUPT
 18. Hei.io - https://www.hei.io/subtitle-generator 39. Zoom’s closed captioning feature
19. Instagram Automatic Captions 40. Meet’s closed captioning feature
20. Kapwing Auto-Subtitle -
https://www.kapwing.com/

Fig. 2 List of automatic subtitling tools used in the present investigation

Without being exhaustive, the list that has been created reveals a rather startling
variety of automatic subtitling tools that can be used by professionals and/or
amateurs. Of these, some are of a rather general nature, while others are used in
specific settings such as the Internet, television, conferences, etc. Each of the selected
tools has been tested and analysed in order to see which features they have. As a
result, several categories of automatic subtitling software were identified, which will
be presented hereafter.

3.2. Automatic Subtitling Tools. Criteria of Classification

Based on the analysis of the profile of automatic subtitling tools, the following
classification criteria were considered: structure of the software (embedded or stand-
alone), mode of operation (fully automatized, partially automatized), type of user
(amateur, general public, professional), medium for using subtitles (browser, social
networks, online conferencing, browser, general use), type of access (free, demo or
commercial), linguistic criteria (monolingual, multilingual). In this section, we
present the categories of automatic subtitling tools observed, describe their most
important characteristics and mention which of the programs in the list analysed fall
into these categories. The analysis will allow us to establish a clear profile of each
automatic subtitling tool.

3.2.1. The structure of the software is an important criterion because it is the first clue
as to whether the subtitling tool we are using is dedicated to professionals or to the
general public. Depending on this, two categories of software are distinguished:
embedded software and stand-alone software. The difference between them lies in the
different way they are implemented and in the importance of the automatic subtitling
activity within the software. Thus, in the case of embedded programs, we are talking
about multifunctional software in which subtitling plays a secondary role, often
reduced to the level of a simple option. In general, these are studio-type software
where the main activity is video editing and which offers its users, among other
options (filters, text insertion, transitions, etc.), also subtitling. One can usually
choose between automatic subtitling, uploading an already made subtitle file and
creating subtitles manually. In other cases, it is a matter of options integrated into a
browser, a conferencing program or a social network. In this category fall: Animaker,
AudioType, Auto-Subtitle Firefox extension, CapCut, Edge Live Caption Feature,
Facebook’s Automatic Caption Feature, Filmora, Flexclip, Flixier, HappyScribe,
Instagram Automatic Captions, Kapwing Auto-Subtitle, Live Caption Chrome,

86

BUPT
Media.io, Otter.ai, Simplified, Veed.io, Wave.video, Wearenova, și YouTube
Automatic Caption.
Stand-alone programs are those in which automatic subtitling is the primary
activity, whether they offer one type of subtitling or several. These are dedicated
software programs such as: Ai-Live Captioning, ASG, Checksub, Clideo, Diy
captions, EasySsub, EVE, Hei.io, Maestrasuite, MateSub, Ooona, Rev, Subtitle video,
SubtitleBee, Transkiptor, Wavel.ai, Webcaptioner, Zubtitle.

3.2.2. Depending on their working mode, automatic subtitling programs fall into two
categories: programs that are fully automatized and programs that are partially
automatized. The first category includes software that performs all the steps necessary
in order to obtain a subtitle (transcription, spotting, and translation), their output being
comparable in form to TV/cinema/DVD subtitling. This category includes the
following software: Ai-Live Captioning, Animaker, ASG, AudioType, Auto-Subtitle
Firefox extension, CapCut, Checksub, Clideo, EasySsub, Edge Live Caption Feature,
EVE, Facebook’s Automatic Caption Feature, Filmora, Flexclip, Flixier,
HappyScribe, Hei.io, Instagram Automatic Captions, Kapwing Auto-Subtitle, Live
Caption Chrome, Maestrasuite, MateSub, Ooona, Rev, Simplified, Subtitle video,
SubtitleBee, Veed.io, Wave.video, Wavel.ai, Wearenova, YouTube Automatic Caption.
Despite the fact that the quality of automatic subtitling nowadays requires post-
editing by a human translator; the capability of such software to perform complex
tasks such as real-time speech recognition, machine translation, text-to-speech
synchronization, and spotting, enables us to consider them as fully automatized.
Among them, there are tools that can perform less common types of subtitling, such
as CapCut, which has an option for automatic subtitling music videos.
Partially automatized tools are those that can display text on the screen yet
without going through all the above-mentioned stages. This is their ability to display
text in a multimedia file, which is not a subtitle in itself from the point of view of
professional translators. In some cases, it is simple ASR-based transcription as carried
out by: Otter.ai, Transkiptor, Webcaptioner și Zubtitle. Despite not being as
advanced, this type of software can be used successfully by professional translators,
especially for simultaneous subtitling, used in television and conferences.
Of the same category, but with a higher degree of complexity because they
can transcribe and spot the spoken content of a video file, but without having to go
through all the steps required for a proper subtitling, are the programs such as: Diy
captions and Media.io.

3.2.3. Classifying software according to the type of user is a worthwhile distinction,

given that subtitling has become a very common practice nowadays due to the
increased attention paid to accessibility. Thus, there are the following types of users
of automatic subtitling tools: the general public (people who occasionally need
subtitles in their daily lives), amateurs (people who are enthusiasts of subtitling but
have no professional training in this field, they create the so-called funsubs), and
87

BUPT
professionals (people working in the AVT field). Depending on these categories of
users, a distinction is also made between software for the general public, for amateurs
and for professionals. The first category includes software offering generic subtitling
services, for which no professional training is required. The general public may
include, for example, influencers, content creators and multimedia artists. They may
use software such as: Animaker, ASG, Diy captions, Facebook's Automatic Caption
Feature, Filmora, Flexclip, Flixier, Instagram Automatic Captions, MateSub,
Media.io, Wavel.ai and YouTube Automatic Caption.
Programs for amateurs have a simple interface and offer a minimum set of
options for subtitling. However, they differ from those intended for the general public
in that the user can intervene and modify the subtitles that have been created. On the
other hand, some essential options for editing subtitles are not implemented in these
programs, which makes them unsuitable for professionals. The following fall into this
category: Ai-Live Captioning, AudioType, Auto-Subtitle Firefox extension, CapCut,
EasySsub, Edge Live Caption Feature, EVE, HappyScribe, Hei.io, Instagram
Automatic Captions, Kapwing Auto-Subtitle, Live Caption Chrome, Maestrasuite,
Otter.ai, Simplified, Subtitle video, SubtitleBee, Transkiptor, Veed.io, Wave.video,
Wearenova, Webcaptioner, and Zubtitle.
Professional software is distinguished by a more complex user interface and
by providing all the necessary elements for editing and spotting subtitles. In this
category we mention: Checksub, Clideo, Ooona, and Rev.

3.2.4. The context of use of subtitling is also an important classification criterion

because it indicates to users which programmes should be used in a specific situation.
For example, for a conference, for browsing the Internet or for social networking it is
not necessary for the users to rely on professional software that requires training and
involves high purchase costs. At present, there are four categories of automatic
subtitling software used in specific contexts: browser subtitling tools, social
networking subtitling tools, online conferencing subtitling tools, and
TV/DVD/streaming subtitling tools. Examples in the browser software category are:
Auto-Subtitle Firefox extension, Edge Live Caption Feature, and Live Caption
Chrome. They are rather complex browser extensions, for example, the latter can do
automatic subtitling in 22 languages.
In the category of subtitling software dedicated to online conferences, the
following software has been identified: EVE, Zoom's closed captioning feature, and
Meet's closed captioning feature. These are recently developed programs that use
speech recognition to perform simultaneous subtitling. At the moment, they can only
perform simultaneous intralingual subtitling and provide accessibility for online
conferences.
Automatic subtitling has also been implemented on social networks through
options such as Facebook's Automatic Caption Feature, Instagram Automatic
Captions, and YouTube Automatic Caption. These features perform intralinguistic
captioning, which is highly required by users as the time spent watching videos with
88

BUPT
captions is increasing on social media. Thus, it is estimated that 69% of social media
users watch video content without sound in public spaces, preferring the use of
captions.
A final category consists of automatic subtitling tools for
TV/DVD/streaming. These are most similar to the concept of professional subtitling
as they are tools designed to generate subtitle files that will later be integrated into
various media such as TV/DVD, streaming, theatre, opera, and so on. Within this
category fall the following programs: Ai-Live Captioning, Animaker, ASG,
AudioType, CapCut, Checksub, Clideo, Diy captions, EasySsub, Filmora, Flexclip,
Flixier, HappyScribe, Hei.io, Kapwing Auto-Subtitle, Maestrasuite, MateSub,
Media.io, Otter.ai, Ooona, Subtitle video, SubtitleBee, Transkiptor, Veed.io,
Wave.video, and Wavel.ai.

3.2.5. Access to automatic subtitling tools is a criterion that distinguishes between

several categories: open access, commercial+demo and commercial programs. Most
of the open access programs are browser-integrated (Auto-Subtitle Firefox extension,
Edge Live Caption Feature, Live Caption Chrome) or social media subtitling features
(Facebook's Automatic Caption Feature, Instagram Automatic Captions, and
YouTube Automatic Caption), where automatic generation of captions involves no
cost for users. There are also a number of generic subtitling tools that are currently
free to access: ASG, CapCut, Diy captions, Simplified, and Zubtitle. Some of which,
such as Webcaptioner, allow users to support the development of software through
donations.
At the other end of the spectrum there are the professional, commercial
subtitling tools, which are usually available on a subscription basis, such as: Ai-Live
Captioning, Kapwing Auto-Subtitle, Ooona, and Rev. Generally, they are the most
advanced automatic subtitling tools amd illustrate the new trends in the development
of up-to-date subtitling software.
Most of the automatic subtitling tools offer a quite flexible access, consisting
of an option for limited access or a demo version of their software. Thus, the user can
put the program to the test and decide whether he/she wants to continue using it. This
category includes: Animaker, AudioType, Checksub, EasySsub, EVE, Filmora,
Flexclip, Flixier, HappyScribe, Hei.io, Maestrasuite, MateSub, Media.io, Otter.ai,
Subtitle video, SubtitleBee, Transkiptor, Veed.io, Wave.video, Wavel.ai, and
Wearenova.

3.2.6. We conclude with the last classification criterion, the linguistic criterion, which
refers to the working languages available within automatic subtitling tools. Therefore,
a distinction is made between monolingual and multilingual subtitling tools. The first
category consists of subtitling tools using a single working language, where the
source and target languages of the subtitling of a multimedia file are identical1. These
are tools which provide transcription and/or subtitling in the same language
(captions). This category includes: AudioType, Auto-Subtitle Firefox extension, Edge
89

BUPT
Live Caption Feature, Diy captions, Facebook's Automatic Caption Feature,
Instagram Automatic Captions, Live Caption Chrome, Otter.ai, Zubtitle, Transkiptor,
and Webcaptioner.
The second category encompasses subtitling software that can provide
speech-to-text transcription, intralingual and interlingual subtitling. That means that
the software can support a wider range of subtitling types and provide access to the
communicative content of multimedia files in multiple languages, which makes them
a more suitable solution for different subtitling environments. In this category,
programs with a wide range of working languages include Ai-Live Captioning,
Animaker, ASG, Checksub, Clideo, EasySsub, Filmora, Flexclip, HappyScribe,
Hei.io, Kapwing Auto-Subtitle, Maestrasuite, MateSub, Media.io, Ooona, Rev,
Simplified, Subtitle video, SubtitleBee, Veed.io, Wave.video, Wavel.ai, and YouTube
Automatic Caption.
Furthermore, it is noteworthy to mention that although all these programs are
multilingual, their subtitling capacity differs from one software to another since their
working language inventory is different. In addition, within a multilingual subtitling
program, differences will be observed in the quality of machine translation for
primary and secondary languages. For example, while we expect a multilingual
automatic subtitling program to have primary languages such as English, French,
Spanish or German, for secondary languages the situation is different. There are a
number of programs that do not have Romanian as a working language, such as
CapCut, EVE, Flixier, and Wearenova. Also, for multilingual automatic subtitling
programs, the quality of subtitling in English will always be much better than the
quality of subtitling in Romanian.
With this overview of automatic subtitling software, it can be stated that they
present a wide variety of programs which have been developed under different trends
and represent solutions for specific subtitling situations. Following the classification
criteria, we can therefore distinguish 6 main categories of automatic subtitling
software:

No. Criteria of Typology of Examples

classification automatic
subtitling tools
1. software embedded Animaker, AudioType, Auto-Subtitle Firefox
structure extension, Edge Live Caption Feature, CapCut,
Facebook’s Automatic Caption Feature, Filmora,
Flexclip, Flixier, HappyScribe, Instagram Automatic
Captions, Kapwing Auto-Subtitle, Live Caption
Chrome, Media.io, Otter.ai, Simplified, Veed.io,
Wave.video, Wearenova, și YouTube Automatic
Caption
standalone Ai-Live Captioning, ASG, Checksub, Clideo, Diy
captions, EasySsub, EVE, Hei.io, Maestrasuite,
MateSub, Ooona, Rev, Subtitle video, SubtitleBee,
Transkiptor, Wavel.ai, Webcaptioner, Zubtitle

90

BUPT
2. working mode fully automatized Ai-Live Captioning, Animaker, ASG, AudioType,
Auto-Subtitle Firefox extension, Edge Live Caption
Feature, CapCut, Checksub, Clideo, EasySsub, EVE,
Facebook’s Automatic Caption Feature, Filmora,
Flexclip, Flixier, HappyScribe, Hei.io, Instagram
Automatic Captions, Kapwing Auto-Subtitle, Live
Caption Chrome, Maestrasuite, MateSub, Ooona, Rev,
Simplified, Subtitle video, SubtitleBee, Veed.io,
Wave.video, Wavel.ai, Wearenova, YouTube
Automatic Caption
partially Otter.ai, Transkiptor, Webcaptioner și Zubtitle, Diy
automatized captions și Media.io
3. user type general public Animaker, ASG, Diy captions, Facebook’s Automatic
Caption Feature, Filmora , Flexclip, Flixier, Instagram
Automatic Captions, MateSub, Media.io, Wavel.ai și
YouTube Automatic Caption
amateurs (fans) Ai-Live Captioning, AudioType, Auto-Subtitle
Firefox extension, Edge Live Caption Feature,
CapCut, EasySsub, EVE, HappyScribe, Hei.io,
Instagram Automatic Captions, Kapwing Auto-
Subtitle, Live Caption Chrome, Maestrasuite, Otter.ai,
Simplified, Subtitle video, SubtitleBee, Transkiptor,
Veed.io, Wave.video, Wearenova, Webcaptioner, și
Zubtitle
professionals Checksub, Clideo, Ooona, Rev
4. the context of browser Auto-Subtitle Firefox extension, Edge Live Caption
use of Feature, Live Caption Chrome
subtitles social networks Facebook’s Automatic Caption Feature, Instagram
Automatic Captions și YouTube Automatic Caption
online conferencing EVE, Zoom’s closed captioning feature și Meet’s
closed captioning feature
TV/DVD/streaming Ai-Live Captioning, Animaker, ASG, AudioType,
CapCut, Checksub, Clideo, Diy captions, EasySsub,
Filmora, Flexclip, Flixier, HappyScribe, Hei.io,
Kapwing Auto-Subtitle, Maestrasuite, MateSub,
Media.io, Otter.ai, Ooona, Subtitle video, SubtitleBee,
Transkiptor, Veed.io, Wave.video, and Wavel.ai
5. acces type Open acces Auto-Subtitle Firefox extension, Edge Live Caption
Feature, Live Caption Chrome, Facebook’s Automatic
Caption Feature, Instagram Automatic Captions, and
YouTube Automatic Caption, ASG, CapCut, Diy
captions, , Simplified, Webcaptioner, Zubtitle
commercial + Animaker, AudioType, Checksub, EasySsub, EVE,
demo Filmora, Flexclip, Flixier, HappyScribe, Hei.io,
Maestrasuite, MateSub, Media.io, Otter.ai, Subtitle
video, SubtitleBee, Transkiptor, Veed.io, Wave.video,
Wavel.ai, and Wearenova
commercial Ai-Live Captioning, Kapwing Auto-Subtitle, Ooona,
and Rev
6. linguistic monolingual - SL Auto-Subtitle Firefox extension, Edge Live Caption
and TL are the Feature, AudioType, Diy captions, Facebook’s

91

BUPT
 same Automatic Caption Feature, Instagram Automatic
Captions, Live Caption Chrome, Otter.ai, Zubtitle,
Transkiptor, and Webcaptioner
multilingual - Ai-Live Captioning, Animaker, ASG, Checksub,
different SL and Clideo, EasySsub, Filmora, Flexclip, HappyScribe,
TL Hei.io, Kapwing Auto-Subtitle, Maestrasuite,
MateSub, Media.io, Ooona, Rev, Simplified, Subtitle
video, SubtitleBee, Veed.io, Wave.video, Wavel.ai,
and YouTube Automatic Caption
multilingual – CapCut, EVE, Flixier, Wearenova
without Romanian
language2
Fig. 3 Classification of the automatic subtitling tools

The development of automatic subtitling software is not going to stop at this

point. These programs will continue to develop and enrich their services according to
technological advances and the language requirements of users and the industry. We
believe that the classification criteria we have established will prove reliable in the
future, allowing for a comprehensive on-the-ground approach no matter how much
automatic subtitling software evolves and changes in the future.

4. Conclusions
In this article, we hope that we have succeeded in illustrating the wide variety of
automatic subtitling software currently available to the general public and AVT
professionals. Furthermore, we believe that the extensive list of subtitling tools
presented in this article provides valuable guidance for the training and skills
development of translators, allowing each of them to choose the tools best suited to
their specific professional context.
Perhaps the most surprising finding of our research is the significant number
of automatic subtitling tools that had to be analysed and classified. This indicates that
in future the training of subtitlers should include computer skills related to the use of
automatic subtitling tools. The fact that the list of subtitling tools has continued to
grow over the course of the study is indicative of the interest in AVT automation,
mainly on the part of the industry. In this regard, we are aware that in Europe, in
addition to the various subtitling tools developed at national level, an automatic
subtitling tool is being developed for use by EU bodies.
In light of this continuous development, the classification of the translator's
working tools in the AVT field is necessary. And our analysis of existing programs
has revealed the possibility of distinguishing 6 classification criteria for automatic
subtitling programs. Although there are certainly other classification criteria,
however, we have selected the most important ones in our research, namely:
structure, working mode, type of user, context of use of subtitling, access, and
linguistic aspects of automatic subtitling programs.

92

BUPT
 Based on the established criteria, a complex classification was made, in which
23 categories of automatic subtitling programs were highlighted as follows:
embedded, stand-alone, fully automated, partially automated, general audience,
amateur, professional, browser, social networking, online conferencing,
TV/DVD/streaming, free, commercial + demo, commercial, monolingual,
multilingual, and multilingual (without Romanian language support). Due to their
development, it is possible for different programs to switch from one category to
another over time.
The brief description of each category of software as well as their illustration
with concrete examples not only provides a structured and clear knowledge of the
field but also allows the researcher to understand the future development directions of
these programs. The structured knowledge about the tools used in the field of AVT
can also be useful as a teaching tool in the training of subtitlers.

5. Bibliography

1. Álvarez, Aitor, A. d. Pozo and A. Arruti. 2010. "APyCA: Towards the automatic
subtitling of television content in Spanish," Proceedings of the International
Multiconference on Computer Science and Information Technology, Wisla, Poland,
2010, pp. 567-574, doi: 10.1109/IMCSIT.2010.5680055.
2. Álvarez, Aitor, Carlos Mendes, Matteo Raffaelli, Tiago Luís, Sérgio Paulo, Nicola
Piccinini,Haritz Arzelus, João Neto, Carlo Aliprandi, Arantza Pozo. 2006.
“Automating live and batch subtitling of multimedia contents for several European
languages”, Multimedia Tools and Applications, Volume 75Issue 1801 September
2016, pp 10823–10853https://doi.org/10.1007/s11042-015-2794-z.
3. Ando, Akio, Toru Imai, Akio Kobayashi, Haruo Isono, K. Nakabayashi. 2000. “Real-
time transcription system for simultaneous subtitling of Japanese broadcast news
programs”. IEEE Trans. Broadcast, 46(3): 189-196 (2000).
4. Bartoll, Eduard. 2008. Paràmetres per a una taxonomia de la subtitulació. Tesi
doctoral, URL: http://hdl.handle.net/10803/7572.
5. Bartoll, Eduard. 2012. La subtitulació. Aspectes teorics i practics, Vic, Eumo
Editorial, 2012, 215 pp.
6. Boulianne, G., Boisvert, M., Osterrath, F. 2008. “Real-time speech recognition
captioning of events and meetings.” 2008 IEEE Spoken Language Technology
Workshop, pp. 197-200.
7. Damper, Robert, Lambourne, A., Guy, D. 1985. “Speech input as an adjunct to
keyboard entry in television subtitling.” Shackel, B. (ed.) Human-Computer
Interaction---INTERACT'84. pp. 203-208.
8. Boulianne, G., J.-F.Beaumont,M.Boisvert,J.Brousseau,P.Cardinal, C.Chapdelaine,
M.Comeau, P.Ouellet, F.Osterrath. 2006. "Computer-assisted closed-captioning of
live TV broadcasts in French", Interspeech Sep., pp. 17-21.
9. Dejica, Daniel & Carlo Eugeni, Anca Dejica-Cartis (eds.) 2020. Translation Studies
and Information Technology - New Pathways for Researchers, Teachers and
Professionals. Timișoara: Editura Politehnica, Translation Studies Series.

93

BUPT
 10. Gao, J., Zhao, Q., Li, T., Yan, Y. 2009. “Simultaneous Synchronization of Text and
Speech for Broadcast News Subtitling”. Yu, W., He, H., Zhang, N. (eds) Advances in
Neural Networks – ISNN 2009. ISNN 2009. Lecture Notes in Computer Science, vol
5553. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-01513-7_63
11. Pirelli, Giuliano. 2004. “Storia del respeaking – Evoluzione del progetto europeo
VOICE”, Lambourne, Andrew & Ltd, Sysmedia & Riverdale, & Hewitt, Jill & Lyon,
Caroline & Warren, Sandra & Hewitt@herts, J & Uk,. Speech-Based Real-Time
Subtitling Services. International Journal of Speech Technology. 7. 269-279.
10.1023/B:IJST.0000037071.39044.cc.
12. Meinedo, H., D. Caseiro, J. Neto, and I. Trancoso. 2003. "AUDIMUS.media: a
Broadcast News Speech Recognition System for the European Portuguese
Language", Proceedings of PROPOR'2003, Faro, Portugal.
13. Neto, João, H. Meinedo, M. Viveiros, R. Cassaca, C. Martins, D. Caseiro. 2008.
"Broadcast news subtitling system in Portuguese," 2008 IEEE International
Conference on Acoustics, Speech and Signal Processing, Las Vegas, NV, USA,
2008, pp. 1561-1564, doi: 10.1109/ICASSP.2008.4517921.
14. Karakanta, Alina, Negri, Matteo, Turchi, Marco. 2020. “MuST-Cinema: a Speech-to-
Subtitles corpus”, in Proceedings of the 12th Conference on Language Resources and
Evaluation (LREC2020), pp. 3727–3734.
15. Marks, M. 2003. “A distributed live subtitling system”. BBC R&D White Papers,
WHP 070, London. [url=http://www.bbc.co.uk/rd/pubs/whp/whp-pdf-
files/whp070.pdf]http://www.bbc.co.uk/rd/pubs/whp/whp-pdf-files/whp070.pdf[/url]
16. Meinedo, Hugo, Caseiro, Diamantino, Neto, João, Trancoso, Isabel. 2003.
“AUDIMUS. (MEDIA): A broadcast news speech recognition system for the
European Portuguese language”. 9-17. 10.1007/3-540-45011-4_2.
17. Melero, Maite, Oliver, Antoni, Badia, Toni. 2006. “Automatic Multilingual Subtitling
in the eTITLE project”, TC 2006; Linguistics, pp. 1-16.
18. Ortega, A., Garcia, J.E., Miguel, A., Lleida, E. 2009. “Real-time live broadcast news
subtitling system for Spanish.” Proc. Interspeech 2009, 2095-2098, doi:
10.21437/Interspeech.2009-600
19. Pozo, A.D., Aliprandi, C., Álvarez, A., Mendes, C., Neto, J.P., Paulo, S., Piccinini,
N., & Raffaelli, M. 2014. “SAVAS: Collecting, Annotating and Sharing Audiovisual
Language Resources for Automatic Subtitling”. Proceedings of the Ninth
International Conference on Language Resources and Evaluation (LREC'14),
Reykjavik, Iceland, pp. 423-435.
20. Trmal, J., Pražák, A., Loose, Z., Psutka, J. 2010. “Online TV Captioning of Czech
Parliamentary Sessions”. Sojka, P., Horák, A., Kopeček, I., Pala, K. (eds) Text,
Speech and Dialogue. TSD 2010. Lecture Notes in Computer Science, vol 6231.
Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-15760-8_53.
Notes:
1 It is important that intralinguistic subtitling should not be confused with subtitling for the hearing
impaired (SDH). The former is one category of captioning, while the latter is a specific type of
intralinguistic subtitling. For a classification of subtitles see Bartoll (2012, 2008)
http://hdl.handle.net/10803/7572.
2 Due to our research interest in Romanian, we have decided to create this category in order to point out

to Romanian translators in the AVT field which automatic subtitling programs are not working into this
language.
94

BUPT