Well, of course, working with an already-good signal to noise ratio is definitely recommended.
But, using reasonably long sweeps, the ESS method allows for increasing the S/N ratio up to +90 dB !!!
This means that, in some circumstances, you can operate with a sweep signal which is WEAKER than the background noise, even 20 dB below it, so entirely inaudible.
We did perform impulse response measurements through the furnace of a power plant, which is 20 m wide, and the sound goes through the flames.
Inside the combustion chamber the SPL was more than 110 dB, and the sweep signal was around 100 dB. Despite this, we managed to measure impulse responses good enough for getting the time-of-flight, hence the speed-of-sound, hence the average gas temperature...
This is reported in this paper:
www.angelofarina.it/Public/Papers/201-AcTechn.pdf