The Republic: Building AI for African Voices
Project: Minim
Newsroom size: 10 - 20
Solution: An AI-powered platform that enables newsrooms and creators to translate content into indigenous African languages and audio formats.
When Nigerian digital media platform The Republic struggled to find text-to-speech tools that could pronounce African names correctly or capture the cadence of local languages, they decided to build their own solution. The result is Minim, an AI-powered platform that promises to revolutionise how African stories are told – and heard.
"If you're telling a Nigerian story, you don't want to be listening to it in an American accent," explains Wale Lawal, Founder of The Republic. "You want to listen to it in a Nigerian accent."
This simple observation sparked a journey that would see a media company transform into a tech innovator, tackling one of the most pressing challenges in African digital media: the underrepresentation of indigenous languages and accents in AI tools.
From student journal to tech pioneer
Founded as a student publication in 2018, The Republic gained prominence during Nigeria's 2020 #EndSARS protests, earning coverage from Al Jazeera. But as a subscription-based platform, they discovered that basic infrastructure didn't exist in Nigeria. "We couldn't accept subscriptions in Naira because nothing existed for that," Lawal recalls. “It could take us up to 60 days to find and license images, which slowed our ability to tell the compelling stories needed to drive subscriptions in the first place.”
The Republic turned obstacles into opportunities, winning a 2022 Google News Initiative award to build Atlas, a media licensing platform. Now, they're applying the same innovative approach to audio with Minim.
The problem: A billion voices unheard
The numbers tell a stark story. While millions of Africans speak at least one indigenous language, these languages remain dramatically underrepresented in digital media. Existing text-to-speech tools like ElevenLabs offer high-quality audio but lack African accents, creating a disconnect between content and audience.
"Media is something that people get very attached to,” Lawal notes. "Can we get a regional accent? Can we get a regional female voice telling this story? Can we tell it in English? Can we tell it in Swahili?"
The challenge extends beyond accents. Reader engagement across digital news platforms is declining, with newsrooms struggling to create immersive content. For long-form journalism – The Republic's specialty – readers increasingly want audio options for consuming content during commutes or other activities.
Building the solution: Four languages, multiple voices
Minim's MVP aims to support four languages: English in Nigerian accent, Pidgin English, Hausa, and Swahili. The platform offers two pre-loaded voice profiles (male and female) but also allows users to train the model on their own voices.
The technical stack reflects the complexity of the challenge:
Python for AI model refinement
Hugging Face for hosting
The Urroman toolkit for text processing
Google Cloud for computing power
Node.js and Next.js for backend and frontend development
The team structure reveals how media companies must evolve to embrace AI. Beyond traditional editorial roles, The Republic assembled software engineers, an external AI specialist, UI/UX designers, and a project manager to coordinate between technical and editorial teams.
Navigating the human-AI bridge
Perhaps the most significant challenge wasn't technical but human. "When we started focusing on AI, we got resistance within the team,” Lawal admits. "People asked, 'Is this going to replace us? Are there going to be cuts?'"
The Republic addressed these concerns by positioning AI as a productivity tool rather than a replacement for creativity. However, coordinating between journalists and AI engineers proved complex. The team appointed two internal staff members – a project manager and a full-stack engineer – to bridge this gap.
"The AI engineer is very concerned with just building,” Lawal explains. "But that's not where the project ends. We need to get feedback from audiences, organise focus group sessions to sense-check Minim’s features."
The scarcity of AI talent in Nigeria added another layer of complexity. With limited local expertise available, sourcing and verifying local AI consultants cost The Republic up to three months of “building time”, and, once resolved, created dependencies that highlighted the urgent need to develop in-house AI capabilities.
The opportunities: Ethical AI leadership from the Global South
Minim is more than a technical solution – it's ethical AI development in practice. By integrating with Atlas, The Republic's licensing system, it creates a marketplace where voice creators monetise their work while newsrooms access authentic voices.
"This will affect any Africa-focused newsroom that wants to create more enriching media," Lawal says.
The opportunities: new revenue streams, enhanced reader engagement, and linguistic diversity preservation. Minim shows how Global South media can lead ethical AI development. This Nigerian newsroom proves that technology can amplify rather than replace authentic voices.
Lessons for newsrooms
The Republic's journey offers crucial insights for media organisations venturing into AI development:
Bridge the communication gap early: Successful AI integration requires dedicated liaisons between technical and editorial teams. The Republic learned this lesson after struggling with coordination: having translators who understand both worlds is essential from day one.
Address AI anxiety transparently: When staff worried about job losses, The Republic positioned AI as a productivity tool, not a creativity replacement. Clear communication about AI's role helps maintain team morale.
Plan for in-house expertise: Relying on external AI consultants proved challenging. "If we're going to scale this platform, we're going to need internal AI skills," Lawal reflects.
Lead with trust: Newsrooms have an advantage: journalistic ethics. "We don't think 'move fast and break things.' The kinds of questions we have around ethics are not questions a tech company would ask,” Lawal observes. "We're thinking about sensitivity checks, focus groups, and ethical considerations because it's embedded in journalistic practice.
Explore Previous Grantees Journeys
Find our 2024 Innovation Challenge grantees, their journeys and the outcomes here. This grantmaking programme enabled 35 news organisations around the world to experiment and implement solutions to enhance and improve journalistic systems and processes using AI technologies.
The JournalismAI Innovation Challenge, supported by the Google News Initiative, is organised by the JournalismAI team at Polis – the journalism think-tank at the London School of Economics and Political Science, and it is powered by the Google News Initiative.
