Introduction to Spatial Audio

Today: from hearing space to reproducing it

  • Identify the cues that reveal width, height, and distance.
  • Explain how distance, directivity, and reflections change a recording.
  • Compare channel-based, object-based, and scene-based audio.

The Spatial Dimension in Natural Sound

  • Width: left to right placement
  • Height: perceived vertical position
  • Depth: perceived distance from the listener

Outdoor sound fields

  • Diffuse ambience arrives from many directions.
  • Discrete events, such as birds or footsteps, remain localizable.
  • With few nearby boundaries, reflections are usually weak or delayed.

Indoor sound fields

  • Direct sound carries the clearest location cue.
  • Early reflections reveal nearby boundaries.
  • Late reverberation suggests the room’s size and absorption.

Listening 1: What space is this?

Listening 2: What space is this?

Listening 3: What space is this?

Listening 4: What space is this?

Listening 5: What space is this?

Level loss in a free field

  • Each doubling of distance reduces sound-pressure level by about 6 dB.
  • 1 m to 2 m to 4 m is two doublings, for a total drop of about 12 dB.
Inverse-square spreading from an omnidirectional source

Sound directivity

Reading a loudspeaker directivity plot

Adam Audio Speaker Directivity Chart

Tuba directivity

Measuring directivity

  • Q = on-axis intensity / average intensity
  • DI = 10 log10(Q)
  • An omnidirectional source has Q = 1 and DI = 0 dB. Larger values indicate narrower radiation.

Sources in reflective spaces

Critical distance

  • Direct sound falls about 6 dB per distance doubling. The reverberant field varies much less with source distance.
  • At critical distance, direct and reverberant sound have equal levels.
  • Dc ≈ 0.057√(QV / RT60) meters

Critical distance and microphone placement

  • Critical distance changes with frequency because source directivity and room absorption also change.
  • In one tracking room, it measured 1.73 m at 63 Hz and 3.2 m at 8 kHz.
  • A microphone at 2 m can be inside the direct field for treble but outside it for bass.

Boundary loading

  • At low frequencies, one nearby boundary can add about 3 dB of output.
  • Two boundaries can add about 6 dB. A three-boundary corner can add about 9 dB.
  • Subwoofer placement uses this gain, but the room’s modes and frequency response still require measurement.

Three ways to represent spatial audio

  • Channel-based: signals are assigned to a fixed loudspeaker layout.
  • Object-based: audio carries position metadata for a renderer.
  • Scene-based: a sound field is encoded independently of the playback layout.

Binaural recording

Binaural Head Diagram

Ambisonics

  • Scene-based representation of a full-sphere sound field
  • Spherical-harmonic channels describe the field, not individual speakers
  • Decoded for a loudspeaker array or binaural playback

The ITU 5.1 reference layout

  • Center: 0°
  • Left and right: ±30°
  • Surrounds: 100° to 120°

Exit ticket: trace one spatial choice

  • Physics: what changed in the sound field?
  • Perception: which cue tells the listener about that change?
  • Reproduction: which recording or playback method preserves it?

Appendix: early spatial-audio milestones

Early sound reproduction

The Théâtrophone: early stereo transmission

Théâtrophone poster by Jules Chéret

Bell Labs in the 1930s