Talov, accessible tech for everyone.

Across the world, 760 million people with disabilities (470M deaf & 290M blind people) face big challenges, no correct access to education and decent work, affecting their INDEPENDENCE.

Less than 5% goes to school, some technical aids can cost up to USD 100K and, according to the UN, the combined global annual cost of unaddressed deafness & blindness is USD 1400 Billion.

Our solution is two apps that combine the power of AI with the daily use of the smartphone to give people with hearing and visual impairment more independence for their daily life activities. Both apps work in real-time, without the Internet, and in up to 48 languages:

SpeakLiz for deaf people: camera-based sign language detection, surrounding sounds identification, human voices transcription.

Vision for blind people: surrounding world audio description, helps in maps navigation, identify objects, distances, money bills of many currencies, read multilingual texts, identify colors, and more.

Across the world, 760 million people with disabilities (470M deaf & 290M blind), face big challenges, no correct access to education and decent work, affecting their INDEPENDENCE. Less than 5% goes to school, some hearing & visual aids can cost up to USD 100K and the combined global annual cost of unaddressed deafness & blindness is USD 1400 Billion.

Since our technology is already being used in 86 countries, our approach is definitely global, we currently have more than 24K downloads, and from them 1,7K are paying users. Despite the fact that the pandemic affected mainly in an economic way to many people we have grown, slowly but still growing in our B2C structure. However, we believe in the power of organizations to create a major culture change when it comes to accessibility, that’s why we are now trying to reach more users through a B2B structure. In that way, we can globally reach more people, often in territories where we currently are, but we need to strengthen our presence due to their better conditions of respect and inclusion for people with disabilities, like Europe, North America, Asia-Pacific.

Our apps both have as the main technical core, pattern recognition through artificial intelligence. First, we consider all the urgent needs of the community of people with disabilities to build our own datasets according to those needs. Once collected, that data is processed and labeled to enter neural network structures.

Primarily we use languages like C, Python, for the training process, and Swift, Kotlin or Java for mobile deployment. The testing stage it’s done from many perspectives, first comes the performance and technical viability, and then comes the main part: testing by real users to determine when an Artificial Intelligence model works and when it doesn't. After their feedback, if there’s a problem we return to the beginning (dataset quality) and start the process again.

But, when models are working well, it’s time to include them through updates. Models frequently used/planned to use in our apps are:

SpeakLiz (for deaf people): sound patterns recognition, multilingual speech recognition, emotion estimation from speech, camera-based sign language movements recognition.

Vision (for blind people): general objects detection, distance estimation, money bills recognition (specific & highly detailed object recognition), GPS pattern recognition to help in maps navigation, color identification, places recognition (context prediction).

The first step is to visit the federations or associations of people with disabilities. We usually make workshops where they can teach us about their unsolved challenges in daily life.

In the case of deaf people, the urgent needs are around better communication levels with hearing people and understanding what's happening around them in terms of sounds and conversations. In the case of blind people, the challenges are related to identifying many objects, text reading, and more about their surroundings.

It's usual to have tons of information, and we start classifying the most urgent or high priority. With that new structure, we usually start prototyping possible solutions. In this stage, it's crucial the help from people with hearing or visual impairment to assist us in the hard & real testing of those prototypes. The participation of deaf & blind people in this process is a key factor to learn and understand their needs and concerns, and it's usual that many suggestions, corrections, and improvements are generated in this stage. That’s why in our small core team (5 people), two of its members are actually people with disabilities. Diana is a blind person that manages our general communications and social media due to her background in Social Communication, but also she manages the R&D and testing of products for people with visual impairment. And Erick is a deaf person who manages our visual and design aspects due to his background in photography and design, but also he manages the R&D and testing of products for people with hearing impairment

Once the prototypes are validated by the deaf & blind people communities, we have a better level of understanding of their needs and we can deliver better updates or improvements to our technology platforms.

When it comes to education access, especially in this pandemic situation, the accessibility of the tools that teachers and educators have is a critical issue. This aspect causes many people with disabilities to experience an even lower rate of learning or access to quality education, in this situation our technology can build bridges to close those gaps while using it not only for educational purposes for better accessing information but also for daily life purposes in general. We strongly believe in the deep & positive impact Talov can have on worldwide education with the correct support of organizations like MIT.

Pilot stage was selected because we already have +24K free users and approximately 2K paying users (our business model is a combination of freemium + subscription), they are distributed in 86 countries right now, due to our up to 48 languages compatibility.

Some technical reasons because our solution is innovative are:

A team composed of great professionals that have hearing or visual impairment and know exactly what their communities need, and these people work with us using their own academic skills and making a high-quality product design & test process.

Great mixing of a smartphone and AI power.

Artificial Intelligence is exclusively designed to improve the quality of life of people with disabilities.

No dependence on an Internet connection.

World coverage in up to 48 languages (Android) and 35 languages (iOS).

Adaptive AI that learns every second to improve pattern prediction.

But the most important aspect is that Talov's work has shown to be a strong game-changer when it comes to improving people’s lives. That’s why we strongly believe now is the right time to take the next step, going massive with the support of organizations like MIT, in order to reach more people and generate an even powerful positive impact on communities.

In general, our technology has Mobile devices and Artificial Intelligence techniques such as Deep Learning as the core, and they can be grouped in branches like:

Computer vision: for object detection, camera-based sign languages identification, distance measurement, text reading (OCR).

Computer listening: for detailed sound identification and voice to text transcription.

Voice synthesis: for text to speech generation (for deaf people) and audio description of surroundings (for blind people).

Primarily we use languages like C, Python, for the training process, and Swift, Kotlin, or Java for mobile deployment. The testing stage it’s done from many perspectives, first comes the performance and technical viability, and then comes the main part: testing by real users to determine when an Artificial Intelligence model works and when it doesn't. After their feedback, if there’s a problem we return to the beginning (dataset quality) and start the process again. In that sense, our technology also goes public after validation of real users, not only by developers, we strongly believe that our technology first has to work in real life with real testers, and then go to the app stores.

There’s plenty of information about computer pattern recognition techniques, like the following (some papers are written by myself (Hugo Jacome, Talov’s co-founder & President):

Object detection:

https://arxiv.org/pdf/1807.05511.pdf

Sound identification with Neural Networks:

https://bit.ly/2MhLc1P

Advanced relation between image & sound, images generation from sound parameters 

https://doi-org.ezproxyberklee.flo.org/10.1121/1.3508471

Check also the video about this work of sound & image relations (it’s in Spanish, I suggest turning on automatic translated captions):

https://youtu.be/18tr8tHeScE

But the most important aspect here is to watch how it works, and much better if is a testimonial of real users:

https://youtu.be/KJfrJZ38hAU

Risk can be considered from many perspectives, on one hand, there’s the perspective that considers how risky the public deployment of this technology can be among the general population, in that sense we don’t see any risk of prohibitions from governments or similar for this technology usage. On the other hand, creating technology that intends to become sensory support for people with disabilities always have risks from the fact that Talov’s doesn’t intend to be the equivalent of having a human assistant (like sign language interpreters or blind-man guidance), so it will never become so and will never replace human work.

About the privacy risk, that’s a major concern for us and is the main reason why all of our AI tasks and processes happen inside the device, avoiding user’s data traveling to and from remote servers with privacy risks in the middle of that road.

In all the world today, the total number of people with hearing impairment is 470 million, and 290 million with visual impairment. That is a total of 760 million people. According to the International Telecommunications Union (ITU), 30% of them have a smartphone, this number is due to the modern technologies built in modern smartphones are very useful for them, for example, they highlight especially voice messages for blind people or video call for deaf people. In that sense, 30% of the world's total deaf+blind people, represent 230 million. 

Now we have 2,3 thousand users (0,001% of the addressable market).

In one year we'll be reaching 30 thousand users (0,013% of the addressable market)

In five years we'll be reaching 0,6 million users (0,26% of the addressable market)

Our goals are closely related to growing as a startup and being able to offer more and better technology to our public, people with disabilities. We consider will never be enough actions in the way to build a more inclusive world in terms of discrimination and lack of opportunities for people with disabilities.

That's why as a tech startup, we invest a lot of resources in the R&D process to continuously improve and prototype new solutions that are in fit with our users' requirements.

That dream applies to all the world, that’s why we are not focused on just one region, but worldwide.

In one year we plan to serve 30.000 users (0,013% of the addressable market) with our technology, and with that volume of operations start our expansion for our five years goal, serving 600.000 users (0,26% of the addressable market). Could sound like small percentages, but we are realistic that the hardest challenge it's not selling technology, the hardest is changing the general society's mindset towards real inclusion and non-discrimination. That will take us harder effort in the early years.

We are currently measuring our impact mainly with the number of downloads and their geographical location, then we measure the amount of paying users that engage with the subscription plans and their period of permanence. In that sense, those are just numbers, we try to be in contact with the users that allow that or want to be in touch with us, to gather information about how our technologies are actually contributing to their lifestyles, and even more specifically to their education and job access conditions.

We work 5 people as full-time staff. Additionally, sometimes in the year, we receive a few students from Universities to do research with us.

Hugo Jacome, Co-founder & President: Classical Musician, Acoustics Engineer, and MBA with International Business specialization. Leads Business Growth, PR, Digital Signal Processing, and Artificial Intelligence models development.

Carlos Obando, Co-founder & CEO: Sportsman, Mechatronics Engineer, and ICT Postgraduate. Leads mobile software development and technology deployment.  

Diana Ceron, Communications: Diana has had a total visual impairment since she was born, in charge of General Communications and community management. Research, development, and validation of technology for people with visual impairment.

Erick Ponce, Corporate Identity: Erik has had a total hearing impairment since he was born, in charge of Corporate Identity and Image of the webpage. Research, development, and validation of technology for people with hearing impairment.

Gustavo Jácome, General Counseling: Experienced Entrepreneur. General Counsel and Legal Affairs, Investor.

The team is expert in artificial intelligence, software development, and accessibility. In a small 5 people team, two members are representatives of hearing and visual impairment communities that contribute with their different academic backgrounds but also with their extremely valuable experience regarding their own disabilities. All this knowledge combined is a powerful way to research and develop new products/services for our users. The most important fact is the full team commitment to work towards real inclusion, and technology is one of our tools for achieving that dream.

We are totally committed to building a diverse, equitable, and inclusive leadership team from our beginning. That is totally evident in our today’s team, which is composed of both genders, and especially (due to our accessibility orientation) our team is composed of people with disabilities working with us and contributing with their valuable experience in the accessibility field.

The main barrier TPrize can help us to overcome is funding to grow our team and accelerate our processes of improving our technology and developing new tools and features for our users, and of course to increase our revenue streams to make it sustainable. We consider TPrize will be useful in specific goals:

15% to strengthen our today's mission of giving free access to our technology by issuing free lifetime licenses to the cases of people that cannot afford it, especially among vulnerable communities, in coordination with national and international disability agencies that have accurate data of the people really worthy of that benefit.

45% to improve our Artificial Intelligence capacities (tech devices, cloud services, and talented people hiring), which will boost usefulness for the subscriptions and means better sales numbers because of product improvements.

40% to increase sales by improving multilingual digital marketing strategies for this special niche, especially directed to the app store search behavior of people with disabilities and their beloved.

In summary, we need support in increasing the user base (that includes final users and organizations), that's the endpoint. But for that to happen, it is necessary to make a lot of processes like improving technology, boosting marketing and brand awareness, among many other tasks. That's why we think our support needs are more like a sequence of aspects instead of one thing.

Tech & cloud computing giants and their teams to expand in a huge way our technological capacity.

Media and awareness specialists for this niche marketing segment, people with disabilities. 

International sales agents that generate increased awareness and sales in far territories from us.

Funds providing capital for specific tech for disabilities.