Woosh: A Sound Effects Foundation Model

2026-04-02

AI Summary

This work addresses the absence of open-source foundation models in sound effect generation, a gap that has hindered both research and practical applications. We present the first open-source foundation model tailored for sound effect synthesis, comprising a high-fidelity neural audio codec, a text–audio alignment module, and a multimodal generative architecture capable of text-to-audio and video-to-audio synthesis. To enable efficient inference, we further introduce a lightweight variant via knowledge distillation. Comprehensive experiments demonstrate that our model matches or exceeds the performance of existing open-source alternatives—such as Stable Audio Open and TangoFlux—on both public and proprietary datasets. The full codebase and model weights are publicly released, effectively filling a critical void in the field.

Abstract

The audio research community depends on open generative models as foundational tools for building novel approaches and establishing baselines. In this report, we present Woosh, Sony AI's publicly released sound effect foundation model, detailing its architecture, training process, and an evaluation against other popular open models. Being optimized for sound effects, we provide (1) a high-quality audio encoder/decoder model and (2) a text-audio alignment model for conditioning, together with (3) text-to-audio and (4) video-to-audio generative models. Distilled text-to-audio and video-to-audio models are also included in the release, allowing for low-resource operation and fast inference. Our evaluation on both public and private data shows competitive or better performance for each module when compared to existing open alternatives like StableAudio-Open and TangoFlux. Inference code and model weights are available at https://github.com/SonyResearch/Woosh. Demo samples can be found at https://sonyresearch.github.io/Woosh/.

Problem

• sound effects
• foundation model
• generative audio
• open models
• audio generation

Innovation

• sound effects generation
• foundation model
• text-to-audio
• video-to-audio
• model distillation