Intron, a Nigerian voice technology company, has launched Sahara v2.5, an update to its voice AI model focused on speech patterns that combine languages within the same conversation. The release adds bilingual code-switching speech recognition across 12 African languages and expands the model’s coverage to 31 languages.
The company said the update is designed to address situations in which speakers move between languages within the same sentence or interaction. Sahara v2.5 also adds support for Nupe, Kanuri, Nigerian Fulfulde, Tigrinya, Kikuyu, Dholuo and Somali. Its text-to-speech and voice-agent capabilities now cover 13 languages.
Intron reports that Sahara v2.5 recorded a 34.3% average word error rate across 12 languages, compared with 53.8% for Gemini 3.6, representing a 36% relative reduction in errors. In separate testing commissioned by the Gates Foundation and CLEAR Global, Sahara ranked ahead on five of seven Nigerian languages evaluated, including comparisons with Gemini and Meta’s Omnilingual model.
The company is also developing a trilingual speech-recognition model supporting Kinyarwanda, English and French. According to Intron, the model uses proprietary technology and is covered by U.S. patent filings.
Commercial deployments illustrate where the technology is being applied. Branch International, a financial technology company, used Sahara-powered collections agents to recover more than ₦1.2 million in delinquent loans in one week, including repayments made outside normal business hours and on weekends. Intron said the agents outperformed human agents on loans overdue by more than 356 days.
The Ogun State Judiciary has expanded its use of Intron’s transcription technology from one pilot court to nine. The company said proceedings that previously took about four hours can now be completed in roughly half that time, with plans to extend the technology to all 18 high courts in the state.
In Nairobi, Intron said a 14-minute Swahili-English doctor-patient consultation can be converted into a structured clinical note in less than 30 seconds. Sahara has also been deployed offline on Nvidia hardware at PAMO Clinics in Port Harcourt through a donor-funded project, targeting environments where connectivity or regulatory requirements limit the use of cloud-based AI tools.
"Code-switching was one of the biggest problems that consistently came up for clients deploying real-world voice AI," said Tobi Olatunji, CEO of Intron. "Africa needs AI built for how Africans really speak."
Sahara currently supports more than 40 enterprise customers across six countries: Nigeria, Kenya, South Africa, Uganda, Rwanda and Ghana. Since 2024, Intron has raised $1.6 million in pre-seed funding and built a training dataset containing more than 150,000 hours of African-language audio from over 53,000 speakers, covering 64 languages and more than 500 accents.
