I think the quote is from Farina, and his interest included room and hall acoustics, so much more than 1m, maybe 10s of metres and more for travel distance of reverberant echoes.
One possibility for rapid variation is simply air currents, a puff of cooler air from the air conditioner or a breeze if one does outdoor measurements.
I believe this is more than just a theoretical issue and does cause problems in practice.
But no personal experience yet.
Best wishes
David
Since Kelvin and Celcius are both linear scales (just shifted), temperature difference (not absolute values!) can just be done in celcius.
Although the smart reader would say that's the same thing.
The basic engineers toolbox formula is:
v=331 + 0.6*T
Where T is temperature in degrees Celsius.
v = velocity of sound
We can use much fancier formulas that actually keep pressure and humidity in mind as well, but the difference around room temperature is basically none existence.
So inside an house/room the speed of sound will vary between say 17 degrees and 27 degrees or so (hot summer day).
Which is between 341m/s and 347m/s
That's about -1.0% and +1.1%
In practice this will be a lot lower.
Between measurements this is of no significance obviously.
Noise floor of the mic is still a big problem when useing 1/4" microphones!
We had problems at precisely EQing the measurement speaker at 20Hz with continuous sweep cause there where noise artefacts influencing at these low frequencies - at 94dBSpl measurement level.
I think you mean SNR here, and not self noise (= noise floor)?
Most microphones have already a bit of a HP roll-off at the lower frequencies.
Obviously we can correct for this (calibration), but that won't change the poor SNR at those lower frequencies.
(since it boosts the noise as well)
The unfortunate part is that we basically don't get any or very little data on this with most microphone datasheets.
I personally don't see any reasons why you would want to measure those very low frequencies with a microphone?
As long as you know the Fs and Qt of a speaker system (as well as the drivers), those lower frequencies and response are super predictable from your favorite simulation software programs.
System Fs and Qt can be derived from impedance measurements.
Self noise of the microphone is the problem in the usecase I'm referring here.
As this is a measurement microphone thread ... there are MANY more things to do with a measurement microphone as measureing a speaker in a living room
The case I used just as example of noise influences which even surprised me was measureing microphones and doing a linearisation of the measurement speaker. But it's just an example to show that you always have to know your limits. The customer simply thought the measurement routine is bad and continued to use his old, way slower routine. I showed him that the self noise of the 1/4" was the issue.
It's very easy to get a very low corner frequency with a measurement microphone, it should be a few Hz (GRAS often uses 2Hz, some mics 0,2Hz) or buy a different mic. It's an omni mic (=closed back), the low corner frequency comes from the vent hole.
Noise over frequency ... really would be an interesting curve! But it doesn't look good compared to an A weightet number. And can differ with different impedance converter preamps. So no, you have to measure it yourself.
