Sounding Liquids:

Automatic Sound Synthesis from Fluid Simulation

William Moss, Hengchin Yeh, Jeong-Mo Hong, Ming C. Lin and Dinesh Manocha

We present a novel approach for synthesizing liquid sounds directly from visual simulations of fluid dynamics. The sound generated by liquid is mainly due to the vibration of resonating bubbles in the medium. Our approach couples physically-based equations for bubble resonance with a real-time shallow-water fluid simulator as well as an hybrid SPH-grid-based simulator to perform automatic sound synthesis. Our system has been effectively demonstrated on several benchmarks.

Video

Full Video (86mb encoded with XviD)
UNC Tech Report: TR09-012. (3mb pdf). To appear in ACM Trans. on Graphics and ACM SIGGRAPH 2010.

Results

	In this scenario, water is poured from a spigot above the surface. The initial impact creates a large bubble as well as many small bubbles. The large bubble disperses into smaller bubbles as it is bombarded with water from above. The generated sound takes into account the larger bubbles as well as all the smaller ones, generating the broad spectrum of sound heard in the supplementary video. Up to 1,530 bubbles were processed in one simulation frame to generate the sounds.
	In this benchmark five objects are dropped into a tank of water in rapid succession, creating many small bubbles and one large bubble as each one plunges beneath the water surface. The video shows the animation and the sound resulting from the initial impacts as well as the subsequent bubbles and sound generated by the sloshing of the water around the tank. We used ten spherical harmonic modes and processed up to 15,000 bubbles.
	In this benchmark we simulate the "dam break" scenario that has been used before in fluid simulation, however, we generate the associated audio automatically. We processed up to 15,000 bubbles using five spherical harmonic modes. This benchmark also demonstrates the creation of a tube-shaped bubble when the wave breaks as the water sloshes back to the left, highlighting the importance of our spherical harmonic decomposition to handle scenarios such as this, where highly non-spherical bubbles are generated (please see the video).
	As a user interactively moves a duck around a bathtub, our algorithm automatically generates the associated audio. The waves created by the duck produces regions of high curvature and velocity, creating resonating bubbles.
	Here we simulate the sound of water as it flows in a small brook. We demonstrate the interactive nature of our method by increasing the flow of water half way through the demo, resulting in higher velocities and curvatures of the water surface and therefore, louder and more turbulent sound.