Spatial audio - Perception, recording, and synthesis of sound scenes

In acoustics and music, sounds are generally characterized by their properties of duration, pitch, timbre and intensity. In addition to these attributes, sounds are also organized in space. Spatialization is the study of sound in its spatial dimension. More precisely, sound spatialization is concerned with the analysis, synthesis and transformation of the properties of sound that contribute to the impression of space and the sensation of immersion. This field of study draws on fundamental research and technological advances from several scientific disciplines: acoustics, signal processing, computer science, psychoacoustics and cognition. Sound spatialization, sometimes referred to as "3D audio", "spatial audio", "multichannel sound", or "virtual acoustics", is a fast-growing field of research with industrial and consumer applications in a wide range of sectors: audiovisual production (music, cinema, radio, 360° video), live performance (sound reinforcement), video games, virtual reality, sound installations, acoustic simulations, etc. For example, in music mixing, spatialization is used to spatially distribute the various instruments to enhance their intelligibility and the clarity of the soundstage. In virtual environments (video games, virtual reality), it simulates sound sources in the space around the listener, contributing to the sensation of immersion.

Depending on the field of application and use case, the aim of sound spatialization is sometimes to capture and restore existing sound scenes or acoustic spaces, and sometimes to virtually create auditory experiences with an immersive sound dimension. To achieve this, the processes implemented exploit networks of electroacoustic transducers (loudspeakers, headphones, microphones) and audio signal processing algorithms, and aim to produce either a realistic rendering ("objectively precise") or a more artistic illusion ("perceptually plausible").

Spatial sound diffusion systems seek, on the one hand, to simulate spatial localization cues for virtual sound sources arbitrarily placed in space and, on the other, to produce artificial reverberation effects that contribute to the creation of an immersive sound space. The techniques available can be classified into three main families: so-called "perceptive" approaches, which rely on models of human hearing to create the illusion of an acoustic phenomenon (positioning a "phantom" source in 3D space, or a reverberation effect that appears "natural"); sampling approaches, which involve measuring an acoustic field at a discrete set of positions and then restoring it using convolution techniques; and physical model approaches, which aim to synthesize, in a certain area of space, a precisely controlled acoustic field that conforms to the laws of wave propagation. Each of...