India – 2025 – For millions of people, speaking is effortless, something done without a second thought. But for individuals living with speech disorders, even a simple conversation can feel like a barrier. The emotional cost of not being understood can be profound. It affects relationships, self-expression, and even the sense of identity. For Pranet Khetan, the founder of Paraspeak, this understanding didn’t come from textbooks or research papers. It began during a school-organised field trip to the Indian Stroke and Paralysis Foundation in Sukhbir Nagar, Delhi in April 2024. There he met patients whose voices were intact but unintelligible, whose speech was present yet inaccessible to those around them. Their struggle was not about forming thoughts, but transmitting them. The visit stayed with Pranet. Curious and concerned, he began studying impaired speech recognition technologies and quickly discovered how limited the field was. Existing tools relied heavily on speaker-dependent models or English-language datasets and were rarely usable in real-world conditions. For low-resource Indian languages, especially Hindi, the research was sparse. The lack of accessible, scalable solutions meant millions of Indians living with speech impairments had little technological support. This insight led to Paraspeak, a project that sets out to create a system that enhances slurred or unclear speech. It has started with dysarthria—or when speech muscles are too weak to control—in Hindi, with the aim of expanding to other disorders and languages across India.

Transforming Slurred Speech with Deep Learning and Real-Time Accessibility

The journey toward a functional solution began with data, which had not existed yet. There was no substantial dataset of Hindi dysarthric speech, so Pranet created one himself. Using a custom toolkit, he collected over 1,400 voice samples from 28 patients, amounting to nearly 42 minutes of raw speech. It became the largest dataset of its kind for Hindi speakers and a critical foundation for building a system that could understand highly variable speech patterns. From there he designed a patent-pending framework for accurate speech enhancement using limited data. Rather than relying on massive, English-focused datasets, Paraspeak’s algorithms are optimised to understand the nuances of impaired Hindi speech and reconstruct it into clear audio. But accuracy wasn’t enough. For the technology to be useful in daily communication, it needed to be seamless. That’s why Pranet developed a portable IoT device that connects users directly to a cloud-based AI engine. With a single action, a patient can activate the device, speak naturally, and hear their clarified voice played back almost instantly. The device was intentionally created to be affordable, portable, and intuitive, ensuring accessibility even for users with limited technical experience.

Scaling the idea through Samsung Solve for Tomorrow

When Pranet entered the Samsung Solve for Tomorrow program, Paraspeak was already a proven, functional prototype but still evolving. The program accelerated its progress significantly. Over the course of multiple stages—including the Top 40 and Top 20 rounds—he completed two major prototyping cycles. These refinements focused not only on hardware design but on strengthening the underlying software. This included expanding the system to support Presbyphonia, a common voice disorder associated with ageing. Adding this capability broadened the scope of Paraspeak beyond its initial focus on dysarthria and demonstrated its potential to serve a wider population. Mentors and industry experts guided him through the transition from student project to market-ready product. They helped refine Pranet’s pitch, sharpen his business model, identify the most impactful user groups, and understand the practical considerations of scaling medical technology. Grant funding will further support development needs and infrastructure. Samsung Solve for Tomorrow both improved the system and provided the confidence and roadmap needed to prepare Paraspeak for real-world deployment.

Feedback that fuels innovation and purpose

While technical milestones matter, the moments that resonate most deeply with Pranet come from user testing. Each time a patient hears their enhanced voice—often clearer than they imagined possible—the emotional response is immediate. Some express relief; others joy; and many feel newfound hope that communication may no longer be a daily struggle. Among all feedback, the request that stands out most is the one to reduce latency. Users want their enhanced speech to return even faster, ideally within two or three seconds. Latency directly affects natural conversation, so improving it has become one of Pranet’s top priorities. Currently, the model runs on a server hosted on his personal computer, which allows for rapid iteration but limits speed. His next goal is to transition to a cloud provider, which will bring latency down to a level that will make real-time conversations more fluid and intuitive. Patient feedback continues to guide the technological roadmap, ensuring that every improvement has clear human impact.

Mentorship that strengthened leadership and vision

Though confident in his technical direction, Pranet acknowledges how much Samsung’s mentorship shaped his understanding of the entrepreneurial landscape. The workshops and consultations exposed him to topics he had never encountered before and helped him develop lean canvas models, structure pricing and distribution strategies, explore funding options, and identify key customer segments. This exposure will help transform Paraspeak from an ambitious research prototype into a structured, scalable solution. It gave Pranet a clearer sense of how to navigate the ecosystem of accessibility technology in India and prepared him for the complexities of future deployment, partnerships, and investment.

Changing How India Understands Speech Disabilities

Beyond its functionality, Paraspeak aims to shift societal perceptions. Speech impairments are often misunderstood, leading to assumptions about a person’s cognitive or emotional abilities. Pranet believes technology can play a transformative role in challenging that stigma. When someone uses Paraspeak and suddenly sounds clear, listeners realise that nothing was ever wrong with the person’s mind; they were only missing a tool that could interpret and transmit their voice accurately. The clarity reveals the thought, intention, and personality that were always present. This shift is central to Paraspeak’s mission of ensuring that disability does not overshadow identity. By enabling clearer communication, the technology encourages society to listen more generously—and to judge less quickly. The project’s guiding statement captures this philosophy with simple power: “Every Voice Deserves Clarity.” In these four words lies the spirit of Paraspeak. It is one that restores agency, dignity, and confidence to people whose voices deserve to be heard exactly as they intend.

About Samsung Solve for Tomorrow

Samsung Solve for Tomorrow is a global competition that empowers students to address real-world challenges in their communities through STEM. By fostering creativity and critical thinking, the program encourages students to innovate and develop impactful solutions to local and global issues.

Through hands-on projects, participants gain the skills and confidence to tackle complex problems, transforming their ideas into actionable solutions. The program nurtures the next generation of leaders while strengthening the global community by connecting young minds across borders to share knowledge and inspire collective progress.

To learn more, visit https://csr.samsung.com/en/program/samsung-solve-for-tomorrow