Apotieliorwa Owoade has a problem he can’t stop thinking about.
He had worked at Aforevo, a local streaming and dubbing company in Lagos, Nigeria, and the experience stuck with him. During his time with the company from 2022 to 2023, he witnessed first-hand how cost-prohibitive the dubbing industry can be.
According to Owoade, translating a film into another language costs more than $500,000 for a complete production.
But what frustrated him even more than the cost was the lack of nuance that most translators failed to capture in the local language. Voice actors were overworked and underpaid, flattening the emotional texture of the scenes they were recording. Existing software tools weren’t great either.
He explained that he had seen the Yoruba language expressed so badly that the phrase “I’m pregnant” sounded flatly “I’ve got balls,” and he winced visibly as he watched our video call.
He wanted to fix it. He called his friend David McAssoir, a computer engineering undergraduate and software developer at the time.
Owoade and Makasoa have known each other for years, and their friendship was partially cemented through their collaboration at Living Faith Church Worldwide International, one of Nigeria’s largest churches. Since 2022, the two men have been collaborating on a project to bridge the language gap between English- and French-speaking congregations at the church’s headquarters in Ota, Ogun State, southwestern Nigeria, Makasole said.
Makasoa was on board when Owoade pitched the idea, and they agreed they needed someone with machine learning expertise.
assemble a team
Makasoa contacted two former classmates of Covenant University, a private Christian university in Ota Ward, Marian Nagy and Emmanuel Ibian. They both graduated in 2024.
Four people answered the phone. At the time, they didn’t even have a name for what they were building. According to Owoade, they called it “The Hagen Project,” and that name made me laugh.
The “Hagen Project” eventually evolved into Reedapt in 2025.
Nnaji brings much-needed machine learning depth to the team.
Before joining, she built a sign language speech-to-text and text model as part of her undergraduate thesis, working on the entire pipeline from data collection to training, deployment, and testing.
She recognized the friction that the Hearing Impaired (HOH) community faced in their daily interactions and wanted to use technology to address it.
She had to put her work on hold. One of the reasons was the data gap that I felt very nostalgic about later when I started working on Reedapt.
According to Nnaji, most of the research on sign language recognition is built on Western contexts rather than Nigerian or African contexts.
“When I started the project, it was just a way to really see how technology could serve as a way to bridge this gap,” Nagy said. “To prove to myself that this is the case.”
Ibiang, a product engineer at Reedapt, arrived with a different but equally important instinct: an obsession with ease of use. The team’s AI engineers sought precision, and Ibiang optimized for the user.
“Can the average Joe use the product without a walkthrough?” Ivian asked rhetorically, as if expecting a response. “The ease of use of the product is what I optimize for.”
For a technically complex product like Reedapt, that perspective has been the team’s internal check against building a great product that no one can use.
In a small team of four, with Mac-Asore and Nnaji as technical engineers and Ibiang as a product specialist, Owoade was described by his teammates as the person with ideas.
Four people, all fresh out of school and under the age of 25, decided to build a house together.
From translation tool to dubbing platform
I first met Owoade and Makasoa at the Builders Summit by Founders Connect, a networking event for early-stage technology founders, held in May 2025 in Lagos, Nigeria.
At the time, they told me they wanted to build a translation tool that would allow users to move between different languages via text and voice without having to learn a new language.
I remember joking that their ambition would eventually put Duolingo out of business.
More seriously, I asked them why we need this in a world that already has DeepL.
They didn’t have a clear answer. It felt like they were still finding the edge.
Since that conversation, the vision has become much clearer.
Reedapt is currently focused on becoming the go-to dubbing and real-time multilingual streaming platform for Nollywood filmmakers, churches, and African content creators looking to spread their work beyond what English can accommodate.
Owode said the startup has signed two enterprise dubbing deals with Nollywood gospel producers and these projects are expected to be completed by the end of 2026.
Currently, Reedapt’s paying customers include a mix of Nollywood producers and churches. We currently serve over 200 active users.
About 94% are individual creators in the consumer base, Owoade said, and the remaining 6% are corporate customers who generate the bulk of the revenue.
Readapt makes money by charging subscription fees.
Prices are structured in dollar terms. There’s a free plan that offers up to 60 minutes of usage, a Creator plan that costs $11 per month, and higher tiers that cost $39 and $99.
Owoade said the decision to price in dollars was intentional.
The team experimented with naira pricing early on and found that it undermined credibility for both users and potential investors.
“Most of our expenses are not in naira,” he said. “Therefore, charging in with naira would be a huge disadvantage.”
The team is currently aiming to reach 50,000 users by the end of 2026.
How four alumni built Reedapt
Building voice technology products for African languages is not the same as building voice technology products for English or French.
According to Owoade, the tools that exist on the market are not built with Africa in mind.
“Big tech companies are not building Africa for Africa as a priority,” he said.
“Apple released AirPods with live translation, but you can see the focus on languages like English, French, and Spanish.
In fact, they seem to prioritize Spanish even more as they cater to Latin American users. So if I have to say, Nigeria is off topic. ”
Reedapt is trying to change that, but the foundation of any good model is data, and the data problem for African languages is serious.
According to Nnaji, training data for the languages Reedapt requires is either very scarce or of poor quality to be useful.
The machine learning engineer compared the problem of training a model to raising a child.
Nnaji explained the difficulty visually, saying that if children are shown only one type of dog, they will have a hard time recognizing other breeds.
“If you’re not giving enough,” she says. “Your model’s performance is still poor.”
The team is currently working with open source licensed data while building a pipeline to collect their own high-quality training sets.
Model architectures are built on existing foundations rather than starting from scratch.
Reedapt was previously credited with Eleven Labs, one of the largest text-to-speech and dubbing companies, but has since moved on to become independent.
By the end of Q2 2026, the team plans to release the first in-house model designed to address the unique challenges of African speech, including code-switching between English and Yoruba in sentences, spells, spiritual languages that should not be translated, and names that existing models routinely confuse.
“We aim to release to the world the first model that can handle a variety of tasks, limited to niche cases,” Owoade said.
“that [will be] Quite independent. But still [today]we do not provide our customers’ data to any third parties and are also aware of the fact that it is a competition. ”
The infrastructure built around the model is just as important as the model itself.
Mac-Asore discussed one layer of the pipeline that is often overlooked. This means that before the audio reaches the model, the team performs a diarization and cleaning process on the model. This allows you to isolate different speakers and remove background noise from your audio.
“We do proper audio engineering before arriving at the model,” says Mac-Asore. “It’s to improve my first speech.”
Accuracy is tracked using word error rate, a metric that measures how far the model’s output deviates from the correct transcript, supplemented by human ratings where team members listen back and flag anything that sounds incorrect to native ears.
Owoade said Reedapt’s dubbing already performs at the same level of accuracy as Eleven Labs for general content, and exceeds it for African-specific nuances.
The platform includes an editor that allows clients to correct mispronunciations, and allows Nollywood producers to upload scripts before a dubbing job to give additional context to their models. Owoade said voice cloning is handled entirely in-house.
All of this is expensive to build. The computational cost of training machine learning models is “prohibitively expensive,” Nnaji says.
Collecting data at the quality the team requires requires a controlled environment and specialized recording equipment.
Owoade said that throughout the development process, the four graduates spent more than $50,000 on Reedapt, subsidizing a significant portion with cloud credits from service providers and AI model providers, and covering the rest out of pocket.
The startup is currently seeking to raise $500,000 to accelerate product development.
The weight of building something personal
For Owoade, building Reedapt is an expression of something he’s had for years.
I studied foreign languages and literature in college and worked in the dubbing industry before starting my own company.
He said he built Reedapt out of a deeply held belief that African creators deserve to tell their stories in their native languages and have them heard around the world without having to route everything in English.
He remembers growing up in a school environment where speaking his native language, Yoruba, was treated as a violation.
“In most schools, they prescribed the fact that when you speak Yoruba, it is the local language,” he says. “It’s our mother tongue.”
What Liedapt tells us is a counter-narrative to that, and his ambition for scale in Africa and storytelling for Africans is undeniably passionate.
He and his team aim to make Reedapt the top dubbing and real-time multilingual streaming tool in Africa by 2026.
Within five years, the company plans to expand into India and the Philippines, which are fast-growing markets for content creation, but where creators face the same revenue disadvantages as in African countries.
The team aims to support 500 languages by 2030.
Owoade asked his teammates to call him “cockroach,” a half-serious reference to the insect’s famous resilience.
He says the journey has been grueling enough that it’s hard to tell enough, and there were days when he wasn’t sure where his next resources would come from. But he hasn’t taken a step back.
Still, expanding Reedapt and betting that African-made dubbing tools are what people want is a big gamble for a team still counting credits and paying for itself.
Owoade said the only thing that would stop them is if 50,000 customers told them the product wasn’t worth making.
So far, no one has done so, he added.
And if, despite everything, Reedapt did not survive, Owoade says he also thought about it.
He plans to leave an open source version for someone else to take over.
“That’s how passionate we are about the problems we’re trying to solve,” he said. “[Reedapt] It’s more than just a tool. It’s very personal to us and that’s how we approach our mission at Reedapt. ”
