Voice tech startup Intron launches new AI models amid broader sector expansion

Intron, a Nigerian AI company that provides speech-to-text transcription tools for healthcare workers, is introducing new speech recognition and text-to-speech AI models as it expands its services beyond the health sector.

The launch comes as African startups develop multiple high-demand tools across various sectors to create additional revenue streams amid tighter funding and rising investor expectations.  Similar to how PaidHR expanded its product offerings last year to diversify its revenue streams, this strategy continues to gain traction across Africa’s tech sector. Intron’s move reflects a similar approach.

Intron is introducing three core models: Sahara-Optimus, a speech recognition model optimised for African accents; Sahara-TTS, a text-to-speech system featuring over 80 voices across more than 40 accents; and Sahara Voice-Lock, an intelligent voice authentication tool trained to combat fraud and deepfakes.

“Intron was born in the busiest hospital wards, where background noise and scarce resources made accurate speech recognition a daily battle,” said Tobi Olatunji, Intron CEO. “We built for the hardest environment first, and now our technology scales effortlessly to courts, call centres, and content creators.”

The startup, formerly Intron Health, piloted its clinical speech platforms in 2022 for hospitals and other key players in Africa’s healthcare sector, including health ministries. Since then, it has expanded its capabilities to offer voice-AI products to the Ogun State Judiciary in Nigeria to alleviate the burdens of manual note-taking and provide human-like conversational voice agents for call centres in the digital finance space.

Intron aims to position itself as the key voice-infrastructure layer for startups, enterprises, and government institutions by offering voice technologies tailored to local needs in Africa, where platform giants like OpenAI may not meet these unique needs.

“Rather than rail against Big Tech model bias, why not build better models?” Olatunji stated.

The startup raised $1.6 million in a pre-seed round in July 2024. Intron is now training a new-generation Sahara-Titan model, which the company hopes will be able to understand, transcribe, and translate twenty major African languages like Swahili, Hausa, and Zulu.  Alongside it is the Sahara-Primus model, which Intron hopes will generate fluent and natural-sounding speech in these twenty languages.

The question is whether more African startups will continue to adopt an industry-agnostic approach and build for broader audiences to keep revenue flowing and stay financially sustainable in an increasingly challenging funding landscape.