A 16 second seamless loop of a busy room: a low wash of many voices talking at once, with the occasional cup meeting a saucer. No words are distinguishable, which is the point. Use it under a work scene, a podcast intro, or any video that needs to sound like other people are nearby and doing fine.
voices: white noise, band-pass 420 Hz, Q 0.9, level 0.9 (the body of many voices, no intelligible words) air: white noise, band-pass 2.2 kHz, Q 0.7, level 0.16 (the room, not the voices) swell: level wobbles 1/8 Hz (exactly 2 cycles in 16 s) by +/- 12%, so the room breathes two cup clinks per loop at 3.1 s and 11.4 s, each a 2.9 kHz / 4.3 kHz pair of sines, 60 ms, level 0.055, added AFTER the crossfade with ((i % N) + N) % N so they stay periodic seam: 0.6 s rendered past the end and equal-power crossfaded (sin/cos) back onto the head noise seeds: 101 (voices), 102 (air)