Issue link: https://digital.copcomm.com/i/1521235
72 M PS E . O R G Psychoacoustics and Technology Ambisonics is a technology based on an expandable, mathematics-based approach to spatial audio capture and reproduction. It encompasses the encoding, storage, and rendering of directional auditory data by formulating the spatial sampling of an infinitesimal sound field such that it may be resynthesized by a finite number of point sources. Unlike channel-based audio, Ambisonics is agnostic to the number of speakers required for playback, but there is a strict paradigm of configuration and quality considerations. It still comprises a number of constituent monaural-audio channels—called components, but these are encoded and must be decoded for playback on any given system. The number of Ambisonics channels depends upon the order of Ambisonics being used. Higher orders require a greater number of channels and generally provide an increased spatial resolution at the cost of data storage, computation, and increased complexity. For example, 1st Order Ambisonics B-Format consists of four channels, 2nd Order of nine channels, 3rd Order of 16 channels, and so on. Ambisonics can be decoded for headphone playback, and this mode lends itself to head-tracking whereby a sound source can be locked to a visual location of an object or an event-emitter and maintain this relationship independently of head movements. Ambisonics has been a prevailing distribution method for head-tracked audio for 3DoF experiences, 180° to 360° films because of its high suitability for head rotation. For example, major XR content platforms Meta and YouTube and the new-generation audio codecs such as MPEG-H and Immersive Audio Model and Formats (IAMF) support content with Ambisonics. Ambisonics can be recorded natively, and stereo and mono signals can be panned and integrated into an Ambisonics signal flow, a hybrid approach is often deployed for more complex scenes that contain dialogue, environmental audio, and other SFX. Another pervading technology is binaural audio. This is generally played via headphones and it endeavors to present spatial sound in the manner that humans hear. Binaural audio can be typically obtained either by recording the sound scene at the eardrum or the ear canal entrance of a listener, such as a dummy head or by synthesising the virtual audio using head-related transfer function (HRTF) filters. A dummy head is a recording device in the shape of a human head (like a mannequin), that hosts two microphones in the ears. This morphology causes the audio magnitude and phase that reaches each of the ear- mics to be filtered by its shape. Alternatively, this filtration can be emulated—synthesised, where HRTF's are applied to monaural signals. This is a dummy-head- equivalent transfer function that describes the acoustic propagation between the same source and the listener's ears. There are a number of binaural capture devices, including wearable devices where you would wear a pair of microphones on your head, thus being able to record audio from the first person's perspective (POV). Similarly to Ambisonics, binaural content can be made without native recording, monaural sources can be panned with binaural panners in DAW and rendered binaurally with a binaural decoder and reverberation setting of choice. Objects can be rendered binaurally in game engines like Unity or Unreal with a native or a third-party binaural decoder. Binaural native recordings and binaurally mixed audio can be part of the hybrid workflow. Ambisonics and binaural audio are not mutually exclusive technologies, as Ambisonics can be decoded binaurally via the HRTF algorithm when content is rendered and played back in real time. In this instance, binaural audio is rendered according to the user's field of view determined by head rotation. With the modern cross talk cancellation algorithms combined with sensors for head- tracking, it is possible to deliver binaural audio over speakers. This technology is available in soundbars, laptops, or automotive, however, the audio production conventions for playback systems largely remain the same. It is worth mentioning the recent emergence of personalised HRTF (P-HRTF) technology for the betterment of end-user experience. This can be done in a controlled anechoic environment, which can be more scientifically accurate but costly and time-consuming. A SOFA (Spatially Oriented Format for Acoustics) file is extrapolated and can be used in audio production workflow within the decoder or assigned to a binaural renderer