According to data from the World Health Organization (WHO), more than one billion people are living with some significant disability today. Moreover, with the market for AI-related technologies set to grow to a cumulative valuation of over $2 trillion in the next seven years, it is reasonable to suggest that the marriage of these spaces can help introduce a new era of accessibility.
Transforming the lives of people with speech impediments
A key area where AI is making its presence felt is when it comes to supporting people with non-standard speech. Voiceitt is an accessible speech recognition technology company that uses AI and machine learning to assist people with speech impairments.
The tech is designed to recognize and adapt to non-standard speech patterns, thereby enabling clearer communication. The technology is particularly beneficial for individuals who have cerebral palsy, Parkinson’s disease and Down syndrome, wherein producing clear speech can be challenging.
As the realm of artificial intelligence (AI) has grown, this still-emerging technology has exhibited an ability to help improve the quality of life for people living with different kinds of disabilities.
Dr. Rachel Levy, speech-language pathologist and customer success manager at Voiceitt, told EdaFace, “The way our technology works is that people input their speech data into our system, and we have a huge database of non-standard speech. So we have held all of this speech data plus the individuals’ speech data that affects their own model.”
“This means that the technology learns from the individual’s unique speech patterns and uses this information to translate their speech into a form that is easily understood by others,” she added.
Magazine: Should we ban ransomware payments? It’s an attractive but dangerous idea
Levy further explained how the technology adapts to changing speech patterns, particularly for individuals with degenerative disorders. Therefore, as these individuals use the tool, Voiceitt continues to record their speech while human annotators transcribe the data to increase recognition accuracy. So if there’s deterioration in their speech intelligibility, the platform can adapt accordingly and retrain its data models to incorporate the new speech patterns.
Voiceitt also has a live captioning capability. This feature allows for real-time speech transcription during video conference calls or live interactions, making conversations more accessible for individuals with speech impediments. Levy demonstrated this feature to EdaFace, showing how the technology can transcribe speech into text and even share it on social media or via email.
Enhancing vision
According to a 2023 study by the WHO, more than 2.2 billion people have some sort of vision impairment, and at least one billion of these cases are easily treatable.
AI-powered imaging tools now have the potential to assist by converting visual data into various kinds of interpretable formats. For instance, tools like OCR.best and Image2TxT are designed to automatically decipher visual cues and convert them into text and audio-based responses.
Similarly, advanced AI models like ChatGPT-4 and Claude 2 have introduced plugins that are capable of decoding extremely complex info (such as scientific data) contained in images, and interpret them with optical character recognition tools.
Lastly, AI-based image tools can increase and decrease contrast and optimize the resolution quality of images in real time. As a result, individuals with conditions like myopia and hyperopia can alter the resolution of images to suit their visual abilities.
Redefining hearing
As of Q1 2023, the WHO estimates that approximately 430 million people currently have “severe disabling hearing loss,” which accounts for nearly 5% of the global population. Moreover, the research body has indicated that by 2050, over 700 million people — or one in every 10 people — will have disabling hearing loss.
Recent AI-assisted hearing tools have allowed individuals with compromised hearing to obtain live captions, audio and video content transcripts. For example, Ava is a transcription app providing the text of any conversation happening around its periphery. Similarly, Google’s Live Transcribe provides a similar service, making everyday conversations more accessible for people with hearing impairments.
Another platform called Whisper harnesses sound separation technology to enhance the quality of incoming speech while reducing background noise to deliver sharper audio signals. The platform also uses algorithms to learn and adapt to a user’s listening preferences over time.
AI-enabled mobility
The Centers for Disease Control and Prevention notes that a little over 12% of Americans experience mobility issues.
Recent innovations in AI-enabled mobility assistants have aimed to build upon already existing mobility aids like wheelchairs.
For example, there are now AI-powered wheelchairs that can take audio cues from the user, thus opening up a new dimension of freedom and mobility. Firms like UPnRIDE and WHILL have created products that offer autonomous navigation and movement capabilities.
AI also appears in mobility-focused exoskeletons and prosthetic limbs, improving the autonomy of finer movements in prosthetic arms and boosting the power of electromyography-controlled nerve interfaces for electronic prosthetics.
AI-based systems can actuate and read different nerve inputs simultaneously, improving the overall function and dexterity of the devices.
The University of Stanford has also developed an exoskeleton prototype that uses AI to improve energy expenditure and provide a more natural gait for users.
Challenges for AI-enabled devices
AI requires the processing of massive data sets to be able to deliver high-quality results. However, in the context of disability, this involves collecting and storing sensitive personal information regarding an individual’s health and physical or cognitive abilities, raising significant privacy concerns.
In this regard, Voiceitt’s Levy stressed that the platform complies with various data privacy regulatory regimes, like the United States Health Insurance Portability and Accountability Act and the European Union’s General Data Protection Regulation.
She also said it is standard practice to “de-identify all of the speech data separating personal data from audio recordings. Everything is locked in a secure database. We don’t share or sell any of our voice data with anyone unless expressly given permission by the user.”
Secondly, because AI tech is expensive to devise, the development of personalized tools for people with specific ailments can be costly and time-consuming. Moreover, the cost of maintaining and updating these systems is also significant.
Recent: PayPal’s stablecoin opens door for crypto adoption in traditional finance
To this point, Jagdeep Sidhu, CEO of Syslabs — the firm behind SuperDapp, an AI-enhanced platform supporting multilingual voice translation and recognition — told EdaFace:
“When it comes to people with visual, auditory, or mobility-related impairments, there is no denying that AI-driven technologies hold incredible potential. That said, one of the most significant hurdles in integrating AI for accessibility lies in the realm of cost. It’s an unfortunate reality that people with disabilities often face steeper costs and challenges to perform everyday tasks compared to those without disabilities.”
As AI and its associated technologies see increased adoption, there is reason to believe that people with disabilities will increasingly explore this space to enhance their lives.
Even recent legislation across Europe and North America is being tailored to improve accessibility and inclusivity, suggesting that AI will play a crucial role within this realm.
