@audio/spatial

Spatial & channel tools — stereo imaging, panning, headphone crossfeed.

Package	Status	What
@audio/spatial-panner	✔	equal-power stereo panner
@audio/spatial-widener	✔	mid/side stereo widener
@audio/spatial-haas	✔	Haas precedence effect
@audio/spatial-autopan	✔	LFO-driven auto-panner
@audio/spatial-crossfeed	✔	headphone crossfeed
@audio/spatial-binaural	✔	structural binaural panner (Brown-Duda ITD + head-shadow ILD, no HRIR)
@audio/spatial-midside	planned	M/S encode-decode
@audio/spatial-channelsplit	planned	channel split/combine (FFmpeg channelsplit)
@audio/spatial-delay	planned	per-channel delay (FFmpeg adelay)
@audio/spatial-surround	planned	stereo → 5.1 upmix

Extracted from effect (spatial/) and filter (eq/crossfeed). FFmpeg-parity targets: stereotools, stereowiden, extrastereo, bs2b.

spatial-binaural is azimuth-only (ITD + head-shadow ILD, per Brown & Duda 1998 eq. 2-5). Elevation is not implemented: the paper's pinna-echo model (eq. 8, Table I) has a per-listener-fitted scaling factor, not a general physical constant, and the paper itself flags it as unvalidated near 90° elevation — using someone's individual calibration as a fake default would misrepresent it as physics, so this atom omits it rather than invent generic coefficients.