The Berlin startup Mirelospecializing in the automated generation of synchronized sound effects for video, has just raised approximately 35 million euros in seed during a tour co-led by Index Ventures And Andreessen Horowitzwith the participation ofAtlantic.vc And TriplePoint Capital. A significant fundraising for a company founded in 2023, in a still unstructured segment of generative AI.
If text, images and video have experienced rapid acceleration under the effect of artificial intelligence, the sound remains largely behind. In most workflows, adding sound effects or music still relies on existing content libraries and manual synchronization work, which is often time-consuming and expensive.
Automate a step that is still largely artisanal
The startup is developing its own audio foundational models capable of analyzing video and automatically generate sound effects consistent with what is happening on screen. In practice, a user can import a video and obtain, in a few seconds, a synchronized soundtrack, without manual intervention.
This type of technology responds to an increasingly visible constraint as content multiplies. Videos generated by AI, short formats, adaptive video games or interactive environments require rapid sound production capabilities, sometimes faster than real time, which are difficult to match with traditional methods.
An audio-native approach, distinct from general models
Unlike large language or vision models, the models developed by Mirelo are designed specifically for audio. The startup highlights lighter architectures, requiring significantly fewer computing resources, while remaining adapted to the constraints specific to sound, in particular temporality and synchronization with the image.
This specialization reflects a different approach to generative AI, more oriented towards specific uses than towards general models capable of doing everything. Mirelo now offers its technologies via an API as well as a web application, Mirelo Studiointended for both independent creators and audiovisual professionals.
A team from research and musical practice
Mirelo was founded by CJ Simon-Gabriel And Florian Wenzelboth researchers in artificial intelligence, passed through AWS Labs And Google Brain. Their academic career is accompanied by a strong musical practice,This dual scientific and artistic culture structures Mirelo’s approach, which seeks to automate technical tasks related to sound, without replacing the creative work of professionals, but by allowing them to concentrate on artistic choices rather than on synchronization operations.
