Should the accents be “deleted” by artificial intelligence?

June 2022, thecompany SANAS announces that it has raised 32 million dollars for the creation of technology based onartificial intelligence whose purpose is to remove accents. September 2022, the platform was born not without creating interest, curiosity and excitement around the world English speaking to French speaking.

Such software throws us into a dystopia where technology comes to make a difference, markers of identity and individuals’ cultures disappear. However, this idea is not new: the movie Excuse me for interrupting published in 2018 already addressed the issue of the accent of African-American populations in a satire about call centers.

“Sorry to Bother You” Movie Trailer – Universal Pictures UK

So how can you actually remove an accent? Between utopia and dystopia, why might developing an artificial intelligence capable of “removing” accents be a problem more than a solution? What do you remove more than a sound mark by neutralizing an accent?

How artificial intelligence can mute an accent

Accent can be defined as a bundle of often oral clues (vowels, consonants, intonation, etc.) which participate in the more or less conscious development of hypotheses about geographical, social or linguistic origin. This accent can be said, among other things, ” regional » or «foreigner» by referring to different notions.

The relevance of identifying an accent lies in the fact that a number of sound characteristics seem homogeneous among speakers of a language, geographical area or social group, as Philippe Boula de Mareüil points out.

These startup technologies are often a black box, and there is little concrete information about the tools used to “remove” the accent. However, the means are several and they mainly aim to partially transform the structure of the sound wave in order to bring certain acoustic signals against a perceptually determined norm.

We can thus play on the timbre of certain vowels, the realization of consonants or even transform parameters such as rhythm, intonation or accentuation according to expected perceptual goals.

At the same time, we will maintain a maximum of vocal parameters that make it possible to identify the voice of the first speaker in the image of ” voice cloning “which may lead to fraud on” voice profoundly false “. These technologies make it possible to separate what is in the speech sequence from what is related to the voice.

These technologies make it possible to separate what is in the speech sequence from what is related to the voice – Pixnio CC0

Real-time automatic speech processing presents technological difficulties, the most important of which is the quality of the audio signal to be processed. Nevertheless, there are different solutions based on deep learning and neural networksas well as large speech corpuswhich makes it possible to better manage the uncertainty in the signal.

In the case of foreign languages, Sylvain Detey, Lionel Fontan and Thomas Pellegrini identify a few challenges inherent in the development of these technologiesnamely, what standard should be used to make a comparison with what is expected, or indeed the role that the corporations can have in determining these objectives – without any particularly promising answers emerging at the moment.

The myth of the neutral accent

But identifying an accent is not limited to acoustic cues alone. Donald L. Rubin was able to demonstrate that listeners can reproduce the impression of a perceived accent simply by associating faces of supposedly different origins with voices.

Likewise, in the absence of these other cues, speakers are not not so good in their ability to recognize accents that they do not hear regularly or that they represent to themselves in a stereotypical way, for example the idea that there are many consonants in German.

Want to remove accents to counter the social effects of a accent discrimination is equivalent to asking the question of what a “neutral” accent is. Now all pronunciation variations involve representations.

Médéric Gasquet-Cyrus, “Marseille specialist” according to the media, recalls that himself the so-called “Parisian” accent is an accent. In French, the accent called “standard” has evolved from sociologically dominant groups : Parisian upper middle class, media (radio, TV), favored middle classes for example.

Tour de France of regional accents and linguistic discrimination – France24

For several years, researchers gathered in a collective attempt to determine the contours of a French reference based on the similarities that exist between all the dialects of the Francophonie. The project ” Phonology of Modern French has thus made it possible to give the general public accents to hear.

It should also be noted that the value attached to an accent (strong, soft, romantic, hard) largely depends on individuals, periods and social groups. But Iván Fónagy, philologist hHungarianshowed that people tend to attribute the same qualities to sounds in his work The living voice : Essays in psychophonetics: /r/ a feisty sound, /i/ as small, /u/ (spelling “ou”) as opulent, etc.

Delete or keep, the chicken or the egg?

In sociology, Wayne Brekhus raises the question of the need to look at the invisible and at the same time deal with the marked and the unmarked – the accent and what is considered a non-accent. This leads to a review of the power relations that exist between individuals and the way in which we homogenize the marked: the one who has (according to others) an accent.

We are also led to question how new technologies can make us more “actor” or “actress” than “automaton”according to Catherine Pascale, by participating in the creation of an eco-ethical framework.

Eliminating an accent means appreciating a dominant type of accent while ignoring the fact that other cofactors will participate in the perception of this accent as much as the emergence of language discrimination. Removing the accent does not remove the discriminations. On the contrary, the accent doesidentify thus participating in phenomena such as humanization, group membership, and even empathy: the accent is quite alteritarian.

If the development of technologies using artificial intelligence and deep learning offer society a still unexplored potential, they can also lead to a dystopia in which dehumanization leads to shifting the political and social role, however great, in that live together and the diversity that is repeated by UNESCO’s World Declaration on Cultural Diversity.

Instead of hiding them, it seems necessary to make recruiters aware of how accents can contribute to customer satisfaction and for politicians to address this issue. Whosenational assembly had taken a strong step by voting in 2020 for a text prohibiting discrimination based on accent, Provence Remember to The Senate don’t seem to get it since it’s still not on the agenda two years later.

This article is produced by The Conversation and hosted by 20 Minutes.

Leave a Comment