Real-Time sound synthesis of audience applause

Abstract

We investigate a procedural model for synthesizing applause sounds that contains novel aspects to ensure high quality and usability. Synthesis of a single clap is generated as a result of filtering a noise source and applying an envelope with exponential decay, based on prior art and existing experimental data. An ensemble approach is introduced to simulate many clappers in a spatially distributed environment. This renders how applause interacts with the space in which it is hosted, including the room impulse response, and where each clap is situated relative to the listener's position. The applause features realistic build-up and fadeout based on natural audience response. The implementation contains meaningful parameters that allow a user to configure and change the sound to achieve a multitude of different types of applause, such as an “enthusiasm parameter” to simulate the greater perceived intensity from an enthusiastic audience. Subjective evaluation was performed to compare our method against recorded samples and four other popular sound synthesis techniques. It showed that the proposed implementation produced significantly more realistic results than other forms of applause synthesis, and it was almost indistinguishable from real-life recordings.