A sound feels alive only when its parts refuse to agree exactly
Two tones at exactly the same pitch sum to one louder, static tone. Move them a few cycles apart and the sum starts to pulse, the amplitude rising and falling at the rate of the difference. That pulse is what the ear reads as motion, breath, a thing that is happening rather than a thing that just is. Perfect agreement is inert. The life is in the residual mismatch.
Balinese gamelan forges this into the metal. Metallophones are made in pairs, a lower pengumbang and a higher pengisep, deliberately tuned a few hertz apart so that striking them together produces ombak, "wave": a shimmer nobody could coax out of a single perfectly tuned bar. This is not a flaw the smith failed to grind away. It is the entire point. The slight wrongness is the sound.
The West keeps rediscovering the same trick whenever it tries to make electronics sound warm. Stack two oscillators at one identical frequency and you get something thin and dead; detune one a hair and it "fattens" into the effect every synth patch files under "chorus." A string section in unison is never mathematically in unison, which is exactly why sixteen violins breathe where one violin multitracked sixteen times sounds like a machine.
The general move: when a standard treats every deviation as error to be stamped out, check whether the deviation was quietly carrying the thing you actually wanted. Sometimes the noise is the signal.
Two tones f₁ and f₂ sounded together beat at a rate of |f₁ − f₂| per second. A few hertz reads as a slow, living shimmer. Push them too far apart and it stops being one note breathing and becomes two notes clashing.
Source claim: Identical signals sum to a static result while near-identical ones beat, so a small deliberate mismatch, not perfect agreement, is what makes combined sound register as alive.