Sonilo与fal发布视频/文本生成音效模型Sound Effects 1.0

Sonilo and fal Launch Sound Effects 1.0 for Realistic Sound Effects from Video and Text
Exclusive API co-launch brings Video-to-Sound Effects and Text-to-Sound Effects generation to developers through fal
SAN FRANCISCO, July 21, 2026 /PRNewswire/ -- Sonilo, a generative audio company building video-native sound and music models, and fal, the generative media platform for developers and enterprises, today announced the launch of Sonilo Sound Effects 1.0, a new model that generates highly realistic sound effects from video or text.
With video input, Sound Effects 1.0 analyzes what is happening on screen and generates one finished audio track synced to the motion, timing and scene. With text input, developers and creators can describe a specific sound effect and generate it directly.
fal will serve as the model's exclusive API launch partner during its initial launch period, providing developers with day zero access through fal's production-ready infrastructure.
Sound Effects 1.0 is designed to address one of the most persistent gaps in AI video production: footage can look complete while still requiring significant manual work before it sounds complete.
When given a video, the model analyzes on-screen motion, scene context, environments, and timing before generating audio that follows what is happening on screen.
Instead of returning a collection of disconnected audio assets that still need to be placed and aligned one by one, Sound Effects 1.0 can produce a synchronized audio track that is ready to review, refine and add to the edit.
"Sound effects only work when they feel like they belong in the scene," said Trista Hong, Co-Founder of Sonilo. "Sound Effects 1.0 was built around that complete problem: understanding the footage, generating realistic audio, and synchronizing it automatically. We're excited to launch it together with fal and bring video-native sound into real production workflows."
A Sound Model Built Around the Video
Traditional sound-design workflows typically begin outside the footage. Editors search sound libraries, preview multiple assets, place them on a timeline, align each effect to the appropriate frame, adjust levels and repeat the process across every action in the scene.
Sound Effects 1.0 begins with the video itself.
The model uses the footage as both a source of semantic information and the timing foundation for the generated audio. It determines what is happening in the scene, what sounds are appropriate for those events and when those sounds should occur.
This video-native approach is particularly useful for scenes containing multiple actions, transitions, impacts and environmental details. Rather than requiring creators to build the sound layer one asset at a time, the model can generate audio around the structure of the footage as a whole.
Sound Effects 1.0 supports video inputs of up to three minutes, making it suitable for short-form content, advertisements, gaming footage, product videos and longer narrative scenes.
Automatic Generation When Speed Matters, Prompt Control When Direction Matters
Sound Effects 1.0 supports two complementary generation workflows.
Video-to-Sound-Effects analyzes uploaded footage and generates sound effects matched to its visible actions, environments and timing.
Text-to-Sound-Effects generates specific standalone sounds from written descriptions, giving creators and developers direct control when they need a particular audio asset.
Prompts are optional in the video workflow. Users can allow the model to interpret footage automatically or provide a prompt requesting a particular sound, emphasis or creative direction.
The prompt helps shape what the model generates, while the video continues to determine when the sound should occur.
This gives users two practical modes of working: automatic sound generation when speed and coverage are the priority, and prompt-guided generation when a scene requires more precise creative control.
Bringing Video-Native Sound Generation to Developers through fal
The co-launch gives developers access to Sound Effects 1.0 through fal's generative media infrastructure, allowing video-conditioned sound generation to be incorporated directly into products and production workflows.
Developers can use the model to build synchronized sound generation into:
- AI video editors and generation platforms;
- Short-form and social video tools;
- Advertising and branded-content workflows;
- Game prototypes, gameplay videos and cinematics;
- Film and narrative-production pipelines; and
- Multimodal creator products that combine video, music and sound.
"We're entering a new era where AI applications don't just generate assets, they produce complete experiences," said Tina Sang, Head of Marketing at fal. "Sound is fundamental to making those experiences believable. Sonilo Sound Effects 1.0 helps developers generate context-aware, synchronized audio that matches what's happening on screen, and we're very excited to bring it to fal, day zero."
The integration is designed to let teams move from initial testing to product deployment without building and operating a separate model-serving stack. Developers can access the model through fal's API and developer tooling while keeping sound generation inside the same environment as their broader generative media workflows.
Expanding the Sonilo and fal Partnership
The launch expands an existing relationship between Sonilo and fal.
Sonilo Music v1.1 is already available through fal, giving developers access to both Video-to-Music and Text-to-Music generation. Sound Effects 1.0 extends that integration from generated music into highly realistic, video-conditioned sound effects.
Using the same source footage, creators and developers can generate sound effects around visible actions and environments, then generate music informed by the video's pacing, scene changes, mood and timing.
This creates a broader video-first audio workflow in which a single video can serve as the timing foundation for both sound design and music. Sound effects can follow what happens on screen, while music can follow the emotional and structural movement of the edit.
By connecting both layers around the source footage, Sonilo aims to reduce manual synchronization, repetitive asset placement and unnecessary switching between separate audio tools.
Built for Real Production Workflows
For AI video creators, Sound Effects 1.0 can add action cues, environmental details, movement and transitions to generated footage that otherwise arrives without usable audio.
For high-volume creators and gaming channels, the model can reduce repetitive timeline work across content requiring dense sound design, including impacts, interface sounds, room tone and movement.
For filmmakers and narrative teams, it can generate scene-level elements such as footsteps, doors, physical interactions and ambience directly from an edit.
For brands and advertising teams, it can produce precisely timed audio around product interactions, camera transitions, packaging moments and visual reveals.
For platforms and API products, it provides a way to add video-conditioned sound generation without requiring users to leave the product and assemble audio in a separate editing workflow.
About Sonilo
Sonilo builds video-native generative audio models for creators, developers and media platforms. Its technology generates music and sound effects directly from footage or text, helping teams bring audio into the video-creation workflow and reduce manual timeline work. Sonilo is headquartered in San Francisco and backed by B Capital.
Learn more at https://sonilo.com/.
About fal
fal is a generative media platform that provides developers with access to the world's best generative image, video, and audio models through a unified API. Trusted by over 2.5 million developers and leading companies, fal offers the fastest inference engine for diffusion models, on-demand serverless GPUs, and dedicated compute clusters for fro