I understand what you mean. When a loudspeaker is unenergised, then clearly the cone(s) don't move, so no noise, but when the cones are moving, then there has to be some mechanical noise from the cones not due to the amplifier.
Particle Atom.
I guess the answer to your question is that noise in electronics is usually thought of as something everpresent even with no audio, like random thermal noise or crosstalk or power supply harmonics leaking into the signal. Changes to the audio are generally thought of as distortion.
I like the concept of "box boof" put forward by Hi-Fi World:
DECAY SPECTRUM
WHAT IT TELLS US
The idea behind this measurement is to fire a short signal at a loudspeaker and see what emerges when it stops. Ideally, there should be nothing. In practice, energy bouncing around inside the cabinet and stored in mechanical reactance within resonant systems dissipates, out through the cone(s) and cabinet walls and port, producing a decaying signal. This isn’t wanted, as it muddies the sound and colours it.
Measurement of spectral decay (waterfall plot) produces a large amount of complex data in a pretty picture that looks impressive but is difficult to interpret. Unlike frequency response there is no agreed and commonly used methodology behind this measurement. We use this test to look at perceptibly obvious colouration, such as box ‘boof’, rather than the intrinsic character of a drive unit, for example. Consequently we use a decay time outside the 20-30mS boundary between intrinsic colour and a divorced echo or colouration, such as box 'boof’. At present, after experiment across a large sample of loudspeakers under test for review, we use 200mS for a useful picture. For similar reasons International Audio Group (Quad, Mission, Wharfedale, Castle and Leak brands) use 500mS (half a second) decay time. Many published waterfall decay plots, however, show a short time window of 20mS or less, inside the Intrinsic / Divorced boundary. These attempt to reveal intrinsic drive unit colouration and are not comparable to our data.
Ideally, there will be no time delayed information but in practice long decays, seen as ranges of descending hills, exist and are evidence of colouration. The undamped glass cabinet of Waterfall's Victoria Evo loudspeaker (above) clearly illustrates this. Often, a colouration seen in this plot can be linked to a small perturbation in our green frequency response plot, and greater disturbance in our red port plot,indicating a strong internal box mode exists. This not uncommon phenomenon, revealed by a spectral decay plot, can be heard as slight chestiness and boxy colour to deep male speech on the radio, where live talk direct to the microphone provides a stringent real life test signal. So spectral decay analysis can be useful, especially when it helps identify a problem seen less obviously within other measurements.
HOW WE MEASURE IT
We fire a test signal known as a ‘log chirp’ (short, fast gliding sine wave burst) through the loudspeaker, as this gives strong low frequency excitation and good signal-to-noise ratio (unlike mls noise). Decay over 200mS (0.2sec) afterward is analysed, depicted as a waterfall plot and as a contour-coloured map. The latter looks at the waterfall from above and uses colour to identify different levels, much like an Ordnance Survey map. Correlation between the two display methods is shown clearly, where the 'hills' in the waterfall are seen as streaks of colour in the contoured map. Highest levels are Red (hot) and Lowest are Blue (cold), running through the colour spectrum between. This is visually easily assimilated and quite powerful once understood. The data is referred to, but not published in the magazine, to save space.
http://www.hi-fiworld.co.uk/index.php/loudspeakers/69-tests/95-decay-spectrum-waterfall.html
Graham Audio BBC LS5/9 (review source)
(thin-wall resonant cabinet construction)
--------------------------------------------------------------------
Usher Dancer Mini-X Diamond (review source)
(thick 50mm baffle, 25mm curved walls cabinet construction)
Yeah, it's interesting. The thin resonant walls are typically thought of as inferior to huge thick ones-but maybe the Graham "gets it over with" faster. I'd like to see more measurements like this. And maybe it's more about bracing and/or damping than sheer mass/thickness. (Shhh, don't tell Wilson and their customers, they'll drop dead of a heart attack ha ha)
