From Gabon to 14 Countries Screening: How LALAL.AI Rescued a Documentary's International Run

When filmmakers and journalists Manon and Sanshey got their documentary When AI Meets Conservation selected for the Goethe Institute's Science Film Festival, they faced an unexpected technical bottleneck. The festival needed music and audio as separate files for translation but the original project files were on a hard drive somewhere in India, while the filmmakers were six months deep into a project in Liberia. With screenings across multiple countries at stake, they needed a solution fast.

What could have derailed the entire international release was solved for a few dollars. Using LALAL.AI, they separated the audio tracks in time for the festival to produce translations in Thai, Vietnamese, Burmese, and Portuguese. The film eventually went on to screen in 14 countries.

In this case study, we talk to the filmmakers about the intersection of technology and nature, their documentary, and how AI became the bridge to its international run.

The Film Concept: AI Beyond the Chatbot Hype

The documentary When AI Meets Conservation explores a very specific niche of technology. The team noticed that when ChatGPT came out, public discourse became dominated by LLMs and generative AI. But their curiosity pulled them in a different direction. Conservation was the space they wanted to explore, and camera traps quickly stood out as one of the most promising areas for practical AI application. They discovered an organization already using AI to sort thousands of photo and video clips by species and send real-time alerts whenever a specific animal passed by.

“The central question we wanted our audience to walk away with was: is the only valid implementation of AI in the world to generate images and text? I think the answer we are realizing is—it isn't,” Sanshey says. “AI for identifying species captured by camera traps is a really important use case to assist greatly in solving a problem that has been hard to navigate for a very long time. And I really wanted anyone watching the film to have a new perspective on this piece of the puzzle.”

Conservation researchers had long been working with DNA sampling and large datasets to identify patterns in ecosystem health. What AI brought to the table was speed and rigor, saving researchers' time and enabling far more robust real-time monitoring. What once might have sounded like science fiction had quietly become reality.

“One of the main benefits of using AI to classify images captured by camera traps is the incredible amount of time conservationists save. In the past, if a poacher was captured on a camera trap, it would take months before the conservationists collected their cameras from the field, and months or years to sort through all of the visuals before coming across the visual of the poacher. So we wanted the audience to understand the value that processing image data faster can bring,” Manon adds.

The documentary follows Okala, a Gabon-based company developing AI-powered camera traps for land management and conservation. Working in forest concessions near the country's capital, Okala's team spent months training machine learning models to identify species passing through the jungle, such as elephants, gorillas, guineafowl, duikers, and built a system that sends real-time alerts via satellite whenever an animal is detected. 

The cameras also flag when a unit goes offline or runs low on battery, reducing the guesswork that typically sends researchers back into difficult terrain. That terrain is no small consideration: reaching the cameras to swap SD cards and replace batteries can take days of trekking through humid, hazardous forest, with the risk of malaria, injury, and encounters with both poachers and wildlife.

The film's origins trace back further than a chance encounter. Manon and Sanshey had been reporting on camera traps and conservation for years, and had previously covered how the technology developed since its invention in the late 19th century, including the more recent shift toward AI-assisted species classification. That's how they first came across Okala's founder. An email exchange for one story eventually led to a ticket to Gabon, where they were already headed for a training project. They reached out to see if an interview was possible, and one thing led to another. A few months later, they had funding for post-production.

"We don't give enough credit to how stories find documentary makers as much as documentary makers find stories," Sanshey reflects. 

The film was published on YouTube, picked up by the Science Film Festival, and screened in 14 countries, none of which was part of the original plan. Since then, Okala's AI has been trained to recognize more species across more countries, a reminder of how quickly the field is moving.

“If we couldn’t separate the files, the Science Film Festival may not screen the documentary in so many countries”

The film was eventually selected for the Goethe Institute’s Science Film Festival, which sounds like a dream, but it came with a major technical requirement. The festival needed separate music and dialogue tracks for dubbing, and the team's original project files were on a hard drive back at their apartment in India. They were six months into a project in Liberia with no way to access them. For a moment, it looked like a logistical problem might cost them some screenings.

The festival organizers gave the team a little over a month; Manon and Sanshey even sent a friend to their apartment in India to look for the hard disk, but she wasn’t able to find it, so the team started to look for other solutions. 

Beyond their own work, the two have trained over a hundred journalists worldwide in using AI tools for audio cleanup and transcription. With expertise spanning long-form documentaries, short-form video, and podcasting, they make a habit of staying current with new tools as they emerge. But when the festival deadline hit, nothing in their existing toolkit was doing the job, which sent them looking for something they hadn't tried before.

“We were thrilled to get selected for the Goethe Institut Science Film Festival and that they wanted to translate and dub our documentary. When they sent their request for separate music and dialogue files, we were a little worried about how we would be able to pull that off because we had left the hard disk where we saved all of our documentary project files at our apartment in India, and we were in the middle of a six-month project in Liberia,” Manon says. “We couldn't access the files that we would need to have separate versions of each. We were worried that that was going to affect our ability to be a part of this film festival. We thought that if we couldn’t separate the files, they may not screen the documentary in as many countries, and we didn't want it to limit how many places this would be screened in.” 

Sanshey worked through multiple tools, testing each within their free tier limits. LALAL.AI stood out immediately, even the sample results were strong enough to justify a subscription. The final output held up across both French and English audio, something no other tool had managed to deliver.

“We tried different audio processing methods on different video editing and audio editing software, and nothing was working. I spent a few days trying to figure that out until we landed up on LALAL.AI.”

“Before uploading the film, I was cautiously excited because the test that I ran worked really well. I exported the audio track with the music, ambient sounds, and voices from all the different sources that were included in the doc, and the voiceover, everything in just one audio file,” Sanshey adds. “The most difficult part was separating the music from the voices and ambient sounds. Surprisingly, it did pretty well, and we gave it a listen through multiple times to catch any kind of artifact or any glitch that may have happened in the middle. But from what we could tell, it was done to a point which only a very skilled audio engineer would be able to do. It was really amazing that with just a tool, two people who are very limited in their ability to process audio through video editing software were able to pull off something like this.”

“I couldn't believe it had come out so well. To get the files separated felt unreal! I was so happy! It felt nice that we were going to be able to share the film with more people.”

Sanshey had braced for a messy process: isolating chunks of audio, stripping each one separately, then stitching the pieces back together. Instead, LALAL.AI took the full track in one pass and returned two clean, completely separated outputs. The files went straight to the dubbing agency the festival had arranged, and that was that.

“We have been receiving a lot of encouraging positive messaging on our news of the film getting screened. So it has been a really rewarding experience, and people from all over are really interested in this topic,” Manon says. “We don’t have the exact numbers for the audience that saw our film, but we know that the film festival was run in 20 countries, and we were screened in 14 of them. In total, the film festival had 953,000 attendees and reached more than 2,000 schools.”

More Documentaries to Come

Their latest documentary, Rewilding the Skies, is already out. This is a film following efforts to bring an endangered parrot species back to Liberia. As for what comes next, neither Manon nor Sanshey has a firm answer, and they seem comfortable with that. Exploratory travel is on the horizon, ideas are being scouted, and some of them may eventually become films. Their YouTube channel Neighbird Press, now over five years old, will keep growing in the meantime with hopes of bringing more contributors into the stories they tell. Where it all leads, they say, only time will tell.

Follow LALAL.AI on Instagram, Facebook, Twitter, TikTok, Reddit, LinkedIn, and YouTube to keep up with all our updates and special offers.