AI has made video creation faster and cheaper than ever — but many AI-generated videos still ship without something essential: sound. Berlin-based startup Mirelo is tackling that gap head-on, building AI models designed to automatically generate sound effects that match on-screen action.
The two-year-old company has raised a $41 million seed round, led by Index Ventures and Andreessen Horowitz (a16z), as investors bet on a new layer of generative AI focused on audio for video, gaming, and creative tools.
Earlier this year, Mirelo released Mirelo SFX v1.5, an AI model that analyzes video content and generates synchronized sound effects in real time. While large players such as Sony, Tencent, ElevenLabs, and Kuaishou-owned Kling AI have launched video-to-audio models of their own, Mirelo is taking a more focused approach — zeroing in on sound effects rather than broad, general-purpose media generation.
That specialization is central to Mirelo’s strategy. According to CEO and co-founder CJ Simon-Gabriel, the company believes sound effects remain an underexplored area of AI research compared to video or music generation, making it easier to build defensible technology. Mirelo was founded by Simon-Gabriel and Florian Wenzel, both AI researchers and musicians, giving the team technical depth and creative insight.
The new funding will allow Mirelo to scale beyond its current 10-person team, with plans to double or even triple headcount by the end of next year. New hires will support research and development, product expansion, and go-to-market efforts as the startup moves from early experimentation toward broader adoption.
Today, Mirelo distributes its models through platforms like Fal.ai and Replicate, with API usage expected to drive most near-term revenue. At the same time, the company is investing in Mirelo Studio, a creator workspace that could evolve into a professional-grade tool for video producers and game developers.
As generative AI faces increasing scrutiny around training data, Mirelo says it has taken a more cautious path. Georgia Stevenson, who led Index Ventures’ investment, said the startup trained its models using public and licensed sound libraries and is establishing revenue-sharing partnerships to respect creators’ rights.
For now, Mirelo is targeting amateurs, prosumers, and indie creators through a freemium model, with a recommended creator plan priced at €20 per month. The goal is simple: make AI-generated video feel complete.
“Sound shapes how people experience video,” Simon-Gabriel said, echoing the long-held view in film that audio accounts for roughly half of the viewer’s emotional response. The same visuals, paired with different sound effects, can completely change the mood.
While Mirelo plans to explore AI music generation in the future, demand for sound effects is currently stronger. The startup believes that owning this layer could become a long-term advantage as AI video tools mature.
Mirelo did not disclose its latest valuation but said it increased significantly from its earlier pre-seed round, which was led by Atlantic. The company has now raised $44 million in total and is backed by notable angels, including Arthur Mensch (CEO of Mistral), Thomas Wolf (CSO at Hugging Face), and Burkay Gur, co-founder of Fal.ai.
As major platforms begin adding audio to AI video generation — including Google DeepMind’s Veo models — Mirelo sees validation rather than competition. The shift reinforces its core belief: AI video without sound is incomplete.