Sonilo Raises $11M Seed to Put AI Audio Inside Video Workflows

Onstage at Visual 1st, Trista Taylor let an audience wait while Sonilo generated sound for a silent clip. The company's account of its CMO's demonstration describes a product making its case in the time it takes to produce the work, a demanding introduction to people who build tools for visual creators.

Sonilo announced an $11M Seed round on October 8, 2026, led by B Capital with Redpoint participating. The San Francisco-based startup develops AI music and sound effects conditioned on video, and plans to put the financing into computing capacity, creator and developer distribution, and relationships with music rights holders. Its investment proposition reaches from understanding a scene to getting usable audio into a production.

CEO and co-founder Shawn Song and CTO Alex Yin bring experience from TikTok's multimodal AI work. COO Keli Li previously ran TikTok's global music operation, while Taylor handles product and market launch, according to Variety's original reporting. That combination puts technical, creative and licensing decisions close together inside the company.

What Sonilo's $11M Seed round will fund

B Capital is backing a business whose product asks the model to interpret footage before supplying its audio. Redpoint joined the financing, and Li's direct announcement identifies the round as Seed. An exact legal closing date and valuation were not established in the reviewed source record.

Management's spending plans run along three connected paths. More computing infrastructure supports model development and scaling; broader distribution brings Sonilo into creator and developer platforms; deeper label, publisher and distributor relationships support the music side of the business. These are intended uses of the capital, with delivery still ahead of the team.

The financing therefore puts several different kinds of work on the same budget. A developer integrating audio generation needs a reliable service and useful controls. A creator needs a result worth keeping. A professional production needs permissions that match its intended use. Improving one of those relationships can bring attention to weaknesses in another, which makes their coordination part of the operating job Sonilo is funding.

A video gives the model an assignment

Sonilo calls its underlying technology the Sound World Model. The company describes a multimodal system that reads movement, pacing and emotional changes in a scene, then produces music and sound effects suited to that footage. Users can start with a video, a text description or both. The ambition is to make the visual edit carry more of the instruction that would otherwise have to be translated into a search or prompt.

That changes the creator's starting point. An editor can arrive with the sequence already cut and ask for sound that follows it, then listen and decide what belongs. Creative judgment stays with the person making the video. Sonilo is trying to reduce the preparation and adjustment required to reach that listening decision, an interpretation of its workflow rather than a measured claim about hours saved.

The company has an existing web product and API, with early distribution through TapNow and ComfyUI described in the funding report. Its product history includes music-model releases, an API announcement and a sound-effects release. The consumer app discussed with the financing is planned for the fourth quarter of 2026, a future distribution step alongside those existing ways to use the technology.

For a platform builder, putting that function inside an established creative workflow can be more useful than sending the user elsewhere. The integration has to preserve the moment at which someone wants to hear an alternative, revise the direction and keep working on the picture.

Licensed training and permission to deliver the work

Shutterstock's public confirmation supports Sonilo's music-catalog partnership. Sonilo's May 11 announcement says it licensed that catalog to train its music model. The relationship gives the company a documented source for training music; it should not be expanded into a claim that the same catalog supplies every kind of sound-effect data.

The permission attached to generated audio is a separate part of the product. Sonilo's current licensing page ties commercial use to the applicable subscription tier or API agreement. Free and Creator outputs are for personal, noncommercial use. Pro covers general commercial uses, while Premium adds categories including film, television, broadcast and studio games; Enterprise uses can carry custom terms.

A production team consequently has two questions to answer: where the training material came from, and whether the generated work can be used for the planned delivery. A licensing partnership helps explain the first. The output's agreement, generation record and intended channel answer the second. That distinction matters when the person enjoying the demo hands the resulting file to somebody responsible for the client's release.

Li's music-industry background belongs directly to this part of Sonilo's strategy. The funding announcement describes plans to deepen rights-holder relationships, while Variety reports that additional publishing and master-recording agreements were still being completed. Those prospective agreements remain plans in this record, rather than finished deals credited to the round.

Bringing sound into the tools creators return to

The team is building for creators, developer platforms and enterprise production workflows, markets with different routes to a purchase. A creator can evaluate the sound immediately. A platform builder also needs to understand how generation fits into a service, and a production buyer has to account for approvals and repeat use. Sonilo's combination of distribution and licensing work acknowledges that the audience for a convincing audio result extends beyond the person who pressed generate.

The surrounding creative-software market includes companies focused on AI video workflows and music royalty infrastructure. Sonilo enters through the relationship between the picture and its sound, then has to carry that relationship through the tools and permissions a production uses.

B Capital General Partner Daisy Cai described generative audio as part of the emerging AI-native video stack in the original funding report. The firm's investment and Redpoint's participation give Sonilo resources to pursue that position. The source record does not establish customer retention, revenue or the cost of serving repeated generations, so a commercial advantage cannot be measured from the funding amount alone.

The useful market observation is the proximity Sonilo wants to create: picture, sound and release permissions available within the work of making a video. Taylor's public demonstration made the first encounter visible. With the Seed financing in place, the company can work on the less visible continuation, as a generated track moves into an editor's project, a developer's product and a production team's delivery record.