Live Subtitling and Speech-to-Text Interpreting - 5 truths

(Copyright notice at the end of the post.)

ATLAS E.P. in cooperation with Prosvasis - Universal Access Research and Innovation Lab for Technology, Society and the Arts bring to the surface the social impact of ECOS 2024 to launch a discussion on the relationship between science and the public.

In the summer of 2024, Prosvasis hosted the international conference ECOS about Speech-to-Text Interpreting (STTI) and live subtitling (live CC), hosting leading figures in the field and inviting the Greek audience to meet these new services, which have been introduced and are being applied professionally in Greece by ATLAS E.P.

The services are offered by ATLAS E.P. since 2017, when they were experimentally applied and endorsed within the arts environment by the users themselves and in 2021 in a conference environment by the Greek Institute for the Deaf, and since then, they have been applied at a wide range of conferences and events for the purposes of information, education and entertainment.

In order to introduce the public and the media to these services, in collaboration with Prosvasis, we are using this post to share information that helps the public and professionals distinguish between the services and related products, so that they can request them, discuss them and be educated about them, producing accurate, truthful and scientific information.

The following constitute the views of ATLAS E.P. and Prosvasis, based on scientific knowledge and practical application.

1. What is STTI?

Speech-to-Text Interpreting is a service designed primarily to help deaf and hard of hearing people by offering them equal access to spoken content.

2. How does STTI work?

Speech-to-Text Interpreting can be provided with two techniques. The first one is Retyping. In this case, the interpreter listens to the spoken content, they interpret it in their brain in easy and clear language with correct syntax and grammar that respects the rules of written discourse, then they write it down, apply edits with the help of their assistants and present the final result to the audience.

The other technique is called Respeaking. In this case, the interpreter listens to the spoken content, they interpret it in their brain in easy and clear language, with correct syntax and grammar that respects the rules of written discourse, then they vocalise (respeak) their interpreting in a speech recognition software that they have trained with relevant vocabulary and their voice. The software displays the text, which is subsequently edited and displayed to the audience.

3. How is it different from surtitling and subtitling?

Surtitling involves preparation of the content. For example, we have surtitling in a theatrical performance which is based on a script. Surtitling can be combined with STTI when the performances have changes and improvisation, i.e. when the text is not standard. An example of this are stand-up comedy performances and performances with unscripted text.

STTI is a method used in live subtitling, but not every STTI service is subtitling. In order for the output of STTI to be considered live subtitling, it must still adhere to the rules of subtitling even in a live setting. Whatever does not adhere to the rules of subtitling, is not considered professional subtitling. Likewise, respeaking and retyping are techniques used in live subtitling, as well as in STTI. However, they can also be used outside subtitling.

4. How does it differ from speech-to-text software?

Speech recognition software is a tool for the interpreter, not a solution, unless the content is of a standard, repetitive, technical nature, aimed at providing more immediate and quicker information, rather than dependability and understanding content of great volume. The output of such a tool is by no means considered a safe solution for simultaneous education and information provision. Moreover, the fact that it is not edited by a human usually makes it unreadable not only for the deaf and the hard of hearing audience, but also for the hearing. Finally, the imprudent use of software without an interpreter is a violation of the rights to equal communication, i.e. communication that allows access taking into account parameters related to accessibility, such as reading speed, correct writing of the language, reliable transfer of information, reproduction of sound elements, etc.

5. Why is STTI important?

For many years, there have been discussions about the usefulness of STTI in Greece and the absence of the service. The need has been expressed and the demand has been asserted from time to time by institutions such as the Hellenic Federation of the Deaf, the Greek Institute for the Deaf and the National Confederation of Disabled People. The first official pilot implementation of the service in a conference setting in Greece was actually during COVID at an event organised by the Greek Institute for the Deaf.

Today, the service is used in the theatre, at concerts (depicting sound live), at conferences, in educational programmes of major institutions employing deaf workers, and even in political speeches by trained professionals of ATLAS E.P.

It is absolutely necessary to use the service for live television programmes, such as news bulletins or other live/news programmes, as also in cases in which the content cannot be rendred easily and clearly in the Greek Sign Language, for example in highly specialised scientific fields.

Finally, research, as well as our experience, shows that this service can be useful to even more people as the written version of the content helps people to better understand what is being said, as for example by coveying it to deafblind people. Moreover, it provides an alternative way of language learning, e.g. for people whose mother tongue is not Greek. Finally, people with learning and even cognitive difficulties can benefit from this service.

COPYRIGHT NOTICE: Any use or republication of the information provided above must explicitly acknowledge its source, as the aforementioned material is protected under copyright law. As copyright holders are identified ATLAS E.P. and Prosvasis.