Ava : subtitles for all

The transcription of videos is an increasingly important subject in the current context of accelerating digitisation. Accessibility (in the widest sense of the word) is also a multifaceted issue. Accessibility for people with disabilities (particularly the deaf or hard of hearing, and even the visually impaired), the accessibility of those that do not understand the language spoken in the video, multi-device accessibility, and synchronous or non-synchronous real time accessibility, etc. We had the pleasure of meeting the Ava team, represented by Antoine. Ava is firstly working on the transcription (speech to text) of conversations (and therefore videos as well). Their objective is to give deaf and hard of hearing people their independence back.

Example of video transcription on smartphone

Lixin (Tradutec Group): In a few words, what was the initial idea that led to Ava being launched?

Antoine(Ava.me): The initial idea was the development of vocal recognition. Our founder, Thibault Duchemin comes from a deaf family. His initial goal was tobreak down communication barriers for these people. There are an estimated 450million people that are “deaf or hard of hearing” in the world.

‍

Lixin (Tradutec Group): What does Ava do? How does it work?

Antoine(Ava.me): Ava is an Artificial Intelligence solution with tone of voice recognition for written transcriptions, in real time, that is available on mobile, tablet or computer. The “real time” feature is what enables our solution to stand out from any competition. Our technology also enables the transcription of so-called “identified” conversations (i.e., knowing “who said what”). All this is paired with Human Intelligence to correct errors, which can be seen as a “less expensive alternative to velotyping”. We therefore work in the “speech to text” tools sector. This real time automatic transcription with speaker recognition is a genuine technological challenge (recognition of accents, hesitations and tones of voices, etc.).

Lixin (Tradutec Group): Would you describe Ava’s solution as providing real time and two-way transcription? Or is that overly simplistic?

Antoine(Ava.me): Yes, we provide written transcriptions, in real time, of conversations that are then available in more than 15 languages. We have combined a translation function to the solution for more than 5 languages. Our application is available on mobile phones with the possibility of inviting other people to join conversations, both remotely and in person.

Lixin (Tradutec Group): The R&D efforts are significant. What does this involve? Funding, recruitment, etc.

Antoine(Ava.me): This firstly involves the desire to help, and the feeling of contributing to the resolution of a serious societal issue. The Ava teams want to change the world of accessibility and autonomy and work towards a more inclusive society. The management of the application’s development also involves a close relationship with our users. We receive a lot of information that enables us to evolve our product every week thanks to this “user”feedback. We signed a fund raiser in December 2020. The funds raised enabled us to accelerate our recruitment and R&D.

Lixin (Tradutec Group): The organisation enables people with disabilities to take back their freedom. Enabling their independence is an excellent objective. Is there a genuine market? Who are your specifiers? Are there any hurdles to overcome before the solution can be adopted?

Antoine (Ava.me):The market is vast. It represents almost 10% of the French population. With the health crisis, the challenges are no longer the same in terms of autonomy and accessibility for deaf and hard of hearing people that study or work in private or public businesses/organisations. This is due to generalised videoconferences, sign language interpreters that are not always available, increased cost of other solutions, and the wearing of masks preventing people from lip reading, etc.

Deaf and hard of hearing people are often the specifiers of the solution through their employers or support organisations.

Lixin (Tradutec Group): For our part, we respond to requests for the transcription of videos and podcasts, that we then have to translate and reintegrate, sometimes in the form of subtitles (subtitles for a video, voice over for a podcast). Are you going to provide an automatic translation of the transcribed conversations? If yes, what difficulties do you anticipate?

Antoine(Ava.me): We are currently developing a roadmap for this as demand is increasing. It is a potential new area of development. It should also be noted that our solution can also transcribe videos and podcasts in real time. This is another possible area of development.

Lixin (Tradutec Group): What is your 3-year plan? What is next for Ava?

Antoine(Ava.me): Our future plans are very clear. To provide a maximum number of deaf and hard of hearing people with the Ava solution: including students, workers, retirees, etc. It will firstly be provided to private individuals or professionals in the USA/Canada, France, Belgium, the Netherlands, Germany,Italy and Spain, and then those is many more countries after that.