EmeraldSnorlax
commented
8 months ago few prelim readings to inform future tests:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8831717/pdf/fnins-15-730744.pdf (seems to help most with low quality audio, or when multiple people are speaking over each other, and may actually require more cognitive load in ideal [good quality, one speaker at a time] conditions?)
https://dl.acm.org/doi/10.1145/365024.365092 (audiences subjectively prefer spatial audio and are indifferent to visualisations; does this carry over to active participation in voice calls?)
https://www.researchgate.net/publication/228746589_Listening_to_unfamiliar_voices_in_spatial_audio_Does_visualization_of_spatial_position_enhance_voice_identification (providing a visualization of the position in addition to spatial audio seems to provide most benefit in large calls, and with unfamiliar voices)
https://www.researchgate.net/publication/370156987_Spatialized_Audio_and_Hybrid_Video_Conferencing_Where_Should_Voices_be_Positioned_for_People_in_the_Room_and_Remote_Headset_Users (waiting for the authors to send me a copy)
purely anecdotally, speaking to multiple neurodivergent people that experience overstimulation mention that spatial audio helps to reduce the effects of it (they drew a comparison between vrchat and discord voice calls)
no need for this to be a protocol level thing, just on the client is fine, but having speakers in a voice call come from different directions may help with overstimulation / recognising speakers