Intron's Sahara v2.5 Enhances African Language AI with Mid-Sentence Code-Switching and Igbo/Hausa Voice Generation

Nigerian AI company Intron has launched Sahara v2.5, a significant update to its voice AI platform. The new version introduces advanced code-switching capabilities, allowing the system to seamlessly handle conversations where speakers switch between English and approximately 20 African languages mid-sentence. This feature was developed in response to real-world deployment challenges in African courtrooms, call centers, and hospitals, where users naturally blend languages.

Beyond improved speech recognition, Sahara v2.5 now offers synthetic voice generation for languages including Igbo and Hausa. This text-to-speech functionality addresses a critical gap in African language technology, enabling more effective voice bots and public service applications. The update is already being implemented at Meridian Hospital in Enugu, where it streamlines doctor consultations conducted primarily in Igbo, allowing for direct dictation rather than post-consultation note-taking.

Intron's business model leverages an API for embedding its technology, alongside direct sales to government and financial services. The company's models are trained on a combination of human recordings and proprietary synthetic data, with a focus on data residency options crucial for navigating tightening African data protection laws. This strategic approach highlights Intron's commitment to building AI solutions tailored to the unique linguistic and regulatory landscape of Africa.

The release positions Intron within a rapidly expanding African voice AI sector, which includes other players like Spitch, Botlhale AI, and N-ATLAS. Intron aims to maintain a distinct competitive edge through its expertise in code-switching, a prevalent feature of actual African speech patterns. While performance benchmarks for v2.5 are pending, the company has a history of publishing comparisons against global AI leaders, underscoring its commitment to rigorous development.