In a world where voice technology is becoming increasingly integral to our daily lives, the limitations of existing speech-recognition systems are glaring—especially for non-Western accents. Most voice-activated technologies, from virtual assistants to customer service bots, struggle with accurately recognizing and responding to accents that deviate from the American or British norm.
Recognizing this gap, Intron Health, a rising tech startup, has developed a speech-recognition tool specifically designed to recognize and understand African accents. Recently, the company secured significant backing to advance this innovative solution, signaling a major step forward for inclusivity in voice technology.
Intron Health’s innovation has not gone unnoticed. The company recently secured substantial backing from a consortium of investors, including prominent venture capital firms and tech industry leaders. The funding will be used to further refine the technology, expand the company’s reach across the continent, and integrate the tool into various applications, from healthcare to education.
Tobi Olatunji, the founder and CEO of Intron Health, a clinical speech-recognition startup, aims to address this issue. Intron Health is positioned as Africa’s largest clinical speech database, boasting an algorithm trained on 3.5 million audio clips (16,000 hours) from over 18,000 contributors, primarily healthcare professionals.
These contributors represent 29 countries and 288 distinct accents. Olatunji emphasizes that having a majority of contributors from the healthcare sector ensures that medical terms are accurately pronounced and captured, which is crucial for his target markets.
“Because we’ve already trained on many African accents, it’s very likely that the baseline performance of their access will be much better than any other service they use,” he said, adding that data from Ghana, Uganda and South Africa is growing and that the startup is confident about deploying the model there.
As the world becomes more digital, speech-recognition technology is becoming a crucial interface between humans and machines.According to a report titled “Speech and Voice Recognition Market by Deployment Mode, Technology, Vertical, and Geography – Global Forecast to 2027” by MarketsandMarkets, the global speech and voice recognition market is expected to expand from USD 9.4 billion in 2022 to USD 28.1 billion by 2027. The market is projected to grow at a compound annual growth rate (CAGR) of 24.4% during the period from 2022 to 2027.
The impact of this speech-recognition tool on the African market could be profound. In healthcare, for instance, the tool could be used to streamline doctor-patient interactions, particularly in telemedicine, where clear communication is crucial.
In education, it could aid in language learning and literacy programs by providing more accurate feedback to students with African accents. The technology also has potential applications in customer service, enabling companies to better serve their African customers by reducing miscommunication and improving user experience.
Intron Health’s speech-recognition tool is unique in its focus on African accents, a feature that sets it apart in the crowded field of voice technology. While most existing systems are trained primarily on American or British English, Intron Health’s tool has been developed using vast datasets of African accents. These datasets encompass a wide range of linguistic nuances from across the continent, allowing the tool to recognize and interpret various African accents with a level of accuracy that was previously unattainable.
Intron Health, which was launched in 2020, initially aimed to digitize hospital operations in Africa by implementing an electronic medical record (EMR) system. However, adoption proved difficult as physicians preferred writing by hand over typing, according to Olatunji. This challenge prompted him to focus on improving the fundamental issue of making data entry easier for doctors. Initially, the company explored third-party solutions to automate tasks like note-taking and integrated existing speech-to-text technologies into their EMR system.
However, these tools struggled with frequent mis-transcriptions, largely due to thick African accents and the pronunciation of complex medical terms and names. This made it clear to Olatunji that existing foreign transcription tools were not suitable for the African context.
The company’s technology leverages machine learning and artificial intelligence to continuously improve its accuracy. As more data is collected, the system becomes increasingly adept at understanding the subtleties of different accents, making it more reliable for users across Africa
The development of Intron Health’s tool also has broader implications for digital inclusion. In many African countries, access to technology is growing rapidly, but the usability of that technology often lags due to language barriers and accent recognition issues.
There have been some immediate positive results from Intron Health’s efforts. For example, Olatunji noted that the company has significantly reduced the waiting time for radiology results at one of West Africa’s largest hospitals, from 48 hours to just 20 minutes. Such improvements are crucial in healthcare, particularly in Africa, where the doctor-to-patient ratio is among the lowest globally.
“Hospitals have already invested heavily in equipment and technology. It’s essential to ensure these technologies are effectively utilized. We are providing value by facilitating better adoption of the EMR system,” Olatunji explained.
By creating a tool that is tailored to the linguistic diversity of Africa, Intron Health is helping to bridge this gap, making technology more accessible and useful for millions of people across the continent. Moreover, the success of Intron Health could serve as a catalyst for other tech companies to prioritize inclusivity in their product development. As the world becomes more connected, the importance of creating technologies that cater to diverse populations cannot be overstated. Intron Health’s speech-recognition tool is a step in the right direction, demonstrating that innovation can—and should—be inclusive.
In addition to developing voice technologies, Intron Health is also making significant contributions to speech research in Africa. The company has recently teamed up with Google Research, the Bill & Melinda Gates Foundation, and Digital Square at PATH to assess major large language models (LLMs) like OpenAI’s GPT-4, Google’s Gemini, and Anthropic’s Claude across 15 countries. This research aims to identify the strengths, weaknesses, and potential biases or risks associated with these models. The goal is to ensure that culturally relevant models are available for clinics and hospitals in Africa.