A San Francisco startup run by people who built TikTok's AI and music operations has raised $11 million to make software that scores video automatically, generating music and sound effects timed to what happens on screen.
Sonilo announced the round on Oct. 8. B Capital led it and Redpoint Ventures also invested, according to Variety, which first reported the deal. The company did not disclose its valuation.
What the product does
Sonilo calls its core technology a Sound World Model. Instead of producing a song and leaving an editor to cut it to fit, the system looks at the footage itself, reading motion, pacing, scene changes and shifts in mood, and then builds audio for that specific clip. Users can upload a video, type a description of what they want, or do both, and get back a soundtrack meant for commercial use.
CEO and co-founder Shawn Song told Variety the goal is for the software to understand the visual story, including what's happening frame by frame, and produce music and effects that fit the scene. Song led multimodal AI work at TikTok. Co-founder and CTO Alex Yin also came from TikTok, and COO Keli Li ran the app's global music operation.
According to Founderland, Sonilo released its first model in March and added a video-to-sound-effects tool in July.
Betting on licensed music
Copyright is the sore spot for AI music, and Sonilo is pitching itself as the rights-cleared option. Shutterstock is a licensing partner for the company's training music, and the AI creator tool TapNow and the open-source image and video tool ComfyUI are early distribution partners. Sonilo plans to launch a consumer app in the fourth quarter.
The new money will mostly go to computing power to scale its models, along with expanding its music licensing business and its reach on creator and developer platforms.
"We believe generative audio will become an essential part of the AI-native video stack," B Capital general partner Daisy Cai said in the Variety report.
Why it matters here
Sonilo joins a crowded field of Bay Area AI startups racing to automate pieces of video production, from generating footage to dubbing it. Its argument to investors is that the soundtrack is one of the most obviously broken parts of AI video today, where music often drifts out of sync or ignores a cut entirely. Whether creators will pay for a fix is the next test, and the consumer app later this year will be the first real look at that.