A YouTube Two Minute Papers episode (July 30, 2017) by Károly Zsolnai-Fehér that reviews a research paper on simulating the sound of virtual elastic rods made from materials such as aluminum, steel, oak, and rubber. The video explains challenges (time-varying frequencies, motion and deformation, many contact events), presents a dipole-based sound synthesis model that is computationally efficient and avoids expensive precomputation, compares results to prior work and to real-world recordings (frequency plots), and demonstrates examples including a Slinky.
Chapters
Dear fellow scholars, this is two minute papers with in this series, we talk a lot about photorealistic rendering and making sure that the appearance of our virtual objects is simulated properly. A lot of works on how things look. However, in order to create a more complete sensorial experience, we also have to simulate how these things sound. And today we are going to have a look at a really cool piece of work that simulates the sound of virtual elastic rods made of aluminum and steel, oak tree and rubber. And of course, before you ask, this also means that there will be sound simulations of everyone's favorite toy, the Walking Slinky. As for all papers that have anything to do with sound synthesis, I recommend using a pair of headphones for this episode. The sound emerging from these elastic rods is particularly difficult to simulate because of the fact that sound frequencies vary quite a bit over time and the objects themselves are also in motion and subject to deformations during the simulation. And as you will see with the Slinky, we potentially have tens of thousands of contact events. In the meantime, let's have a look at some results. For the fellow scholars who are worried about the validity of these Star wars sounds. I know you're out there. Make sure to watch the video until the end. The authors of the paper proposed a dipole model to create these simulations. Dipoles are typically used to approximate electric and magnetic fields in physics, and in this case it is really amazing to see an application of it for sound synthesis. For instance, in most cases these sound waves are typically symmetric around 2D cross sections of these objects, which can be described by a dipole model quite well. Also, it is computationally quite effective and can eliminate these lengthy pre computation steps that are typically present in previous techniques. There are also comparisons against the state of the art and we can hear how much richer the sound of this new technique is. And as you know all too well, I love all papers that have something to do with the real world around us. And the reason for this is that we can try the very best kind of validation for these algorithms, and this is when we let reality be our judge. Some frequency plots are also available to validate the output of the algorithm against the real world sound samples from the lab. It is really amazing to see that we can use science to breathe more life in our virtual worlds. Thanks for watching and for your generous support and I'll see you next time.
