INVESTIGATION Thanks to advances in artificial intelligence, new technological tools allow people who are visually impaired or hard of hearing to better perceive their environment and to communicate more easily.
It is a long kitchen, all that is most normal for an American apartment, with a table for breakfast or coffee. There is no big surprise, and there won’t be. At Christian Laine, a computer trainer adapted to the Federation of the blind in the USA, the house is not “all connected”. We are still a long way from Google’s demonstration spaces where everything from the popcorn machine to the living room light is connected to a single voice assistant – showcases that give the concept of “home automation” a lot of old fashion. But we are coming.
On the table is a discreet Google Home speaker which he calls to check the weather or listen to the radio. “OK Google: put USA info. “ The voice assistants are in the process of becoming an essential tool for Christian Laine: ” It opens up superb opportunities for the blind, we are on par with the visionaries in front of Google Home, “ he rejoices. He can buy train tickets through his enclosure, consult his calendar. He still hesitates on the brand for his connected thermostat, the American Nest or the French Netatmo, and is looking forward to new applications like the microwave with a voice assistant announced by Amazon. At home, he can also control his Samsung television with his voice, but it is not yet connected to Google Home, nor its induction hobs or its kitchen scale, which gives the oral level of heating or weight.
Today, 20% of mobile searches are carried out by voice.
Fernando Pinto Da Silva, digital manager of the Federation of the blind in the USA, hopes to be able to connect light bulbs to his voice assistant: “My partner and I are blind, but I would like to know when the lights are on or off so that not leave our daughter, who can see well, in the dark. » Only, he finds their cost still too high: 20 dollars minimum per bulb, and not all of them are compatible with his system. He opted for Alexa from Amazon.
Table of Contents
The voice assistant boom
The voice becomes a new interface for accessing everyday tools and the Internet. “Today, 20% of research on mobile is done by voice,” notes Cecile Pruvost, strategy and development manager at Google, during a conference at the National Institute for Young Blind People (Inja), in Paris, organized early December. Sales of connected speakers are estimated at 1.7 million in the usa for their first year of marketing, according to NPR. In 2018, the figure would be between 62 million and 75 million worldwide, at least double compared to the previous year. A market dominated by Amazon.
Most often, each brand of the connected speaker or smartphone has an assistant: Siri from Apple for HomePod and iPhone, Google Assistant for Google Home and Pixel, and Alexa from Amazon for Echo. But soon the voice assistants may be like browsers, separated from the hardware. Orange said that on its live box will be its own, Djingo, as well as Alexa. The Mozilla Free Software Foundation, which started Firefox, is developing its own voice assistant. What facilitates what is designated by the barbaric concept of “interoperability of systems” or more simply “all-connected”.
These tools are not without raising questions. During the conference at Inja in which Google participates, a person blind to the public worries about the fate of his personal data, which must be sent to the servers of the internet giant to be processed: “I have don’t want Google to know everything about me. “ Response from Cécile Pruvost: ” Each user can share or delete their history. “ The manufacturers also insist that only the moments where he was ordained to the speaker to wake are treated.
In order to short-circuit the problem, the French start-up Snips presented, at the CES (Consumer Electronics Show) in Las Vegas, smart chips to board on household appliances: ovens, vacuum cleaners, etc. Its credo is the processing of information locally: the data does not leave the hardware. A technology that responds to fairly simple commands, however. We must, therefore, learn to use it while the giants of the Web seek interaction with the most intuitive tool possible: “There are 5,000 ways to program your alarm clock,” says Cécile Pruvost.
Visual recognition systems
Advances in artificial intelligence (AI) have been made possible thanks to the processing of mass data, including that of Google: the more data, the more the machine can train to learn to recognize patterns. The techniques developed today apply as much to sound as to the image. Since 2015, visual recognition systems have been more relevant than humans for identifying objects or animals in photos.
Microsoft was the first to make such an announcement. The same year, the Israeli company OrCam, which was also working on this technology, launched the marketing of a so-called intelligent camera for the visually impaired: MyEye. Mounted on glasses and fitted with a small loudspeaker, it allows you to read the text, recognize banknotes or people and announce them out loud. With this device, the data remains local. Its cost, which can be partially covered by the MDPH (Departmental House for the Disabled), amounts from 3,750 dollars to 4,750 dollars .
Microsoft has also entered this market with the Seeing AI application, available in The usa since December 2017, currently only in English, the english version expected to arrive in 2019. The tool is free but collects the data for processing in the cloud. It also allows us to describe scenes (a perfectible aspect) and people or to indicate the level of brightness. It is only available on the Apple AppStore at this time.
I don’t know any blind Braillist who doesn’t have an iPhone as a smartphone .
It is no coincidence: “I do not know any blind Braillist who does not have an iPhone as a smartphone”, explains Fernando Pinto Da Silva. The reason is simple, Apple is the first manufacturer to have included voice synthesis in its devices to read the screen, VoiceOver, in 2009, on the iPhone 3GS. “It was to access the American education market, ” said Fernando Pinto Da Silva. Accessibility is a prerequisite for services to the public in the United States, and the powerful federations of the blind take care to uphold the law.
Phone app for the deaf and hard of hearing
In the USA, a law of 2005, supplemented by the law for the Digital Republic of 2016, guarantees, in theory, the accessibility of the websites of communities and of delegates of public services, but Fernando Pinto Da Silva regrets that the texts are not sufficiently applied. No sanction has fallen, but a decree planned for the first half of 2019 must improve monitoring and extend the obligation to large companies. Today in the USA there are 1.7 million blind and partially sighted people. There are approximately 250 million worldwide, a figure that could be multiplied by two or three by 2050due to population growth and an aging population. Enough to constitute a market likely to arouse the interest of the private sector.
But regulation seems to remain the best incentive. This is in any case what Olivier Jeannel found with his RogerVoice application intended for deaf people (like him) and hard of hearing, who are 4 million in the USA: “The law has given a boost. ” Since October 7, large companies are obliged to have telephone service available to the public. RogerVoice makes it possible to transcribe in writing dialogue by telephone and to respond by typing on the keyboard (a speech synthesis then takes care of reading the text aloud). In addition, it is possible to hire a sign language interpreter via video. One hour is offered per month, then the plans range from 5.99 dollars to 29.99 dollars.
RogerVoice, the app for the deaf and hard of hearing
The idea came to him when he was in the United States, precisely because accessibility was more developed there, but he specifies that his tool is second to none. His business is also a telephone operator, which allows the application to be used even when the interlocutor does not have it. Recently, Olivier Jeannel signed no less than twenty partnerships with companies like OuiSNCF, Allianz, GRDF or Aéroport de Paris. Similarly, it helps other operators such as Orange, Bouygues and SFR.
But he does not intend to compete with them: “We complete what they do, we have no network, no SIM card. “ With 18 people, more than respectable for an effective start-up created in 2013, it would still be difficult. It doesn’t stop him from thinking big. Its application is available (transcription and text-to-speech only) in the United Kingdom, Germany, Mexico and Japan. It is now targeting the United States and Canada. His core business is an application, not AI, provided by a service provider, but he nevertheless declares that he is working on an emotion recognition system: “When I call my father, I would like to know if he is happy or angry. “
RogerVoice collaborates with Ava, a start-up for the same public co-founded in particular by Frenchman Thibault Duchemin. Like Olivier Jeannel, he passed through Berkeley, in the United States, but a few years later. Born in 1991, Thibault Duchemin is the only one who can hear well from his family. Having used to play the interpreter during his childhood, he had the intuition to develop a tool to facilitate dialogues between deaf people and those who do not master sign language: “The problem is not hearing, it’s communication, ” he defends.
Ava allows you to transcribe discussions between several people on a smartphone screen, but each must have downloaded the application – a free service for five hours per month, then at 29 dollars unlimited. It was launched in 2016 in the United States and this summer in the USA, Thibault Duchemin claims 100,000 monthly users in these two countries. He does not want “to wait for Google or Apple to develop this software and for the crumbs to be recovered”. For him, its application should not be reserved for deaf and hard of hearing people, it should also be useful for those who want to catch up with a meeting or follow it remotely.
Accessibility tools are generally used for other purposes; “Seeing AI is a tool for promoting the cloud service, “ explains Philippe Trotin, accessibility director at Microsoft. The OrCam company will launch a connected personal development camera in March for everyone. “The Google assistant was not created for the primary purpose of helping the visually impaired, but today it is a key element, “ says Cécile Pruvost. Accessibility has become a field of choice for the development of artificial intelligence systems, both in terms of image, for technical interest and to perfect tools that must be addressed to all. Some have understood,