Kasa Noma Tech Solutions: Building AI Communication Tools for People Living with Speech Impairments in Ghana

At a Glance 

Kasa Noma Tech Solutions is a Ghanaian assistive technology startup building AI-powered communication tools for people living with speech impairments — in their own languages. Their flagship product, the Kasa Noma Speech Assistant, enables people with dysarthric or non-standard speech to send voice messages via WhatsApp and Telegram that are accurately transcribed and converted into clear, synthesized audio. A second product, GSL Connect, provides real-time Ghanaian Sign Language interpretation using computer vision. Both tools are designed to run on ordinary Android smartphones — no expensive dedicated hardware required. 

Kasa Noma was formed during the Tɛkyerɛma Pa Innovation Sprint (February–May 2025), Ghana’s first inclusive AI innovation sprint focused on speech technology and disability supported by the Centre for Digital Language Inclusion (CDLI). The team won first place and subsequently registered as a company — making them the standout post-sprint venture from Ghana and one of the clearest examples of an innovation sprint producing a venture with real continuity beyond the programme. 

Partner: University of Ghana • Sprint duration: 4 months • Placement: 1st Place Winner 

The Problem 

In Ghana and across West Africa, people living with speech impairments face a compounded exclusion. Mainstream speech AI tools — voice assistants, dictation software, speech-to-text applications — are built on standard speech datasets, dominated by non-African accents and unimpaired speakers. They routinely fail for people with dysarthria, Autism Spectrum Disorder, Cerebral Palsy, ALS, cleft palate, or stroke-related speech differences. 

The failure is compounded when those users also speak Akan, Ewe, or other Ghanaian languages. The assistive technology products that do exist — tools like Tobii Dynavox or Proloquo2Go — are priced at $200 or more, require devices that are rarely available in Ghana, and offer no support for local languages. The result is a communication gap that technology could close, but has not. 

Kasa Noma was built on the premise that this is not a technical impossibility — it is an investment choice. The data exists to change it, if someone decides to collect it. The devices to run it are already in people’s hands. The question was whether anyone would build for this community, in this context, in these languages. 

The Innovation Sprint: What Happened 

The Tɛkyerɛma Pa Innovation Sprint — the name means “good speech” in Akan — ran from February to May 2025 as the first CDLI-supported inclusive AI innovation sprint in Ghana. Implemented through University of Ghana, the sprint brought together developers, researchers, and people living with speech impairments to design and build practical AI solutions grounded in Ghanaian realities.  

The sprint was itself made possible by prior work: CDLI in partnership with University of Ghana had already undertaken data collection and developed ASR models for Ghanaian languages, creating the technical foundation that innovation sprint participants could build on. Kasa Noma was therefore not starting from zero — they were working within an ecosystem where the groundwork for Ghanaian speech AI had already been laid. 

The sprint was structured around application development, with teams supported through onboarding, technical training, mentoring, co-creation sessions, a midpoint review, and a final showcase. 

The Kasa Noma team, from the University of Environment and Sustainable Development, identified the gap that defined their eventual company: not only were mainstream voice tools inaccessible to people with non-standard speech, but the entire assistive technology market in Ghana was built around imported, expensive, English-only products. Their response was to ask a different question — what if you built the same capability on a platform people already use and trust? 

At the final showcase, Kasa Noma placed first. Within weeks of the sprint ending, the team had registered as a company: Kasa Noma Tech Solutions. 

The Solution 

Kasa Noma Tech Solutions is developing two core products, both designed explicitly for Ghanaian contexts, Ghanaian languages, and ordinary Android devices. 

Kasa Noma Speech Assistant 

A voice assistant bot integrated directly into communication tools such as WhatsApp and Telegram — platforms already used daily by millions of Ghanaians. For a person with dysarthric or non-standard speech, sending a voice message is often unusable: recipients cannot understand what was said, and the sender has no way to clarify. The Kasa Noma Speech Assistant listens to the voice message, transcribes it into text, and converts it into a clear, synthesized voice — effectively acting as a real-time communication bridge. In contexts where literacy levels are lower and verbal communication is the dominant mode of daily life, voice messaging is already the natural default — making it a far more accessible interface than text, and meeting users where they already are. The AI model is trained on local, non-standard speech patterns, including Akan, making it functional for users that standard assistants cannot serve. 

GSL Connect 

A real-time Ghanaian Sign Language interpreter using computer vision via a smartphone camera. GSL Connect translates sign language gestures into spoken audio and text, and converts spoken Akan, English, or Ewe back into sign language avatars. The tool is designed for use in hospitals, schools, and workplaces — everyday settings where the absence of interpretation creates concrete barriers to access. 

Both products operate on a Bring Your Own Device model — no specialist hardware, no expensive imports. A user needs only the smartphone they already own. 

What Difference It Made 

The most direct impact of the sprint was institutional: a team that entered as students left as founders. Kasa Noma Tech Solutions is now a registered Ghanaian company with a developed business plan, a clear product roadmap, and active partnerships with the Global Disability Innovation Hub (GDI Hub). 

From a disability inclusion perspective, the sprint helped the team develop several critical commitments: 

• Co-design with the disability community at every stage — not as validation at the end, but as a shaping force throughout product development. 

• Explicit focus on Ghanaian languages — the first known effort to build AI speech models for non-standard speech in Akan. 

• A data sovereignty approach — building and owning the datasets of impaired Akan speech that currently do not exist, with the intention to contribute these to the global research community. 

• A financial model designed for affordability — including a cross-subsidization approach where higher-end institutional services fund accessible tools for low-income users. 

Challenges and How They Were Addressed 

Building inclusive AI in a low-resource language context surfaces challenges that are not well understood by the mainstream AI industry: 

• Technology literacy barriers. Even where the tools work, caregivers, teachers, and family members may not know how to use them. Kasa Noma has building a Train the Trainer model into their operational plan, reaching end users through the people closest to them. 

• Investor perception of disability markets. The disability technology sector is often seen as niche or unscalable by commercial investors. Kasa Noma’s response is to demonstrate both the scale of the underserved population and the transferability of their technical approach to other language groups and markets beyond Ghana. 

• Sustainability in a low-ability-to-pay market. Many of their target users cannot afford to pay directly for services. Their cross-subsidization model — institutional licensing and data licensing subsidizing free or low-cost access for individuals — is designed to address this, though it requires patient capital and strong institutional partnerships to work. 

Where Kasa Noma is Heading 

In the short term, the focus is on completing the MVP of the Kasa Noma Speech Assistant and piloting it with 50 users through WhatsApp and Telegram. In parallel, they are formalizing partnerships with the Ghana National Association of the Deaf and university research partners for clinical validation of their tools. 

Over the medium term (six to eighteen months), the team plans to officially launch the Speech Assistant and pilot GSL Connect in selected hospitals. Their longer-term ambition is regional expansion into Nigeria and Kenya, supporting Yoruba and Swahili speakers, and establishing Kasa Noma as a key contributor to African speech impairment research. 

The company is currently seeking GH₵100,000 in initial capital to fund prototyping, user testing, and clinical validation.  

Six Lessons for Others Working in This Space 

Based on the Kasa Noma experience, and the broader Ghana Innovation Sprint, the following approaches may be useful for others designing inclusive AI programmes or supporting disability technology ventures in African contexts: 

1. Start with language, not just technology.The most significant technical gap in Ghana was not capability — it was data. Without speech data in local languages from people with non-standard speech, no AI model can serve these users. Funders and programme designers should support data collection as a first-class activity, not a precondition someone else has already solved.
2. Build on what people already use.Kasa Noma’s decision to deploy via WhatsApp and Telegram rather than a standalone app was a deliberate and significant design choice. In contexts where smartphone adoption ishigh but data costs are real, integrating with familiar platforms reduces friction dramatically. Inclusive technology does not always require new infrastructure. 
3. Co-design is not a consultation — it is an ongoing relationship.Kasa Nomais building co-design with the disability community into their operating model, not just their development process. Involving people living with speech impairments in needs assessment, product testing, and training delivery produces different outcomes than bringing them in at the end to validate decisions already made. 
4. Pair academic partnerships with community access.The Ghana sprintdemonstrated that academic partners bring technical depth and institutional credibility, but may not have the disability-community networks needed for sustained user engagement. Future programmes and ventures should seek both — or build community access deliberately into the programme design from the start. 
5. Sustainability must be designed in, not added later.Kasa Noma’s cross-subsidization model — where institutional revenue funds affordable access for individuals — was part of their thinking from early in the sprint, not retrofitted afterward. Disability technology ventures in low-income markets need business model support alongside technical development.
6. Post-sprint supportdetermineswhether ventures survive. Winning a sprint is not the same as building a company. The transition from prototype to pilot, and from pilot to business, requires mentoring, networks, funding pathways, and sustained institutional relationships. Without structured post-programme support, even the strongest sprint teams can disengage. Kasa Noma’s continued momentum reflects both the quality of the team and the importance of maintaining those connections after the showcase ends. 

Why This Case Matters 

Kasa Noma illustrates a critical shift in how innovation programs are evaluated. 

The success of the initiative is not defined solely by the creation of a prototype. It is defined by the emergence of a sustained pathway from: 

• Idea
• Prototype
• Company
• Real-world implementation 

The formation of a registered company and development of a scalable business model demonstrates that innovation programs can create long-term capacity, not just short-term outputs. 

This case study was produced as part of the CDLI Innovation Sprint programme. Kasa Noma Tech Solutions is based in Ghana and supported by the GDI Hub.