The lab the NAO works out of has a Framery O phone booth, for taking calls. It's very good at sound isolation, but being so small (~1.5 m3, 54 ft3) you might be worried about CO2 building up:

I decided to test it. Here are two rounds of sitting in the booth until the meter (Temptop M2000) shows CO2 levels plateauing, then exiting and leaving the door cracked until CO2 has returned to baseline:

The booth is advertised as having 21.5 L/s (45 CFM) of mechanical ventilation, and you can see it works well. With the ventilation disabled, however, it reaches 2k PPM CO2 in 12min, 2.5k in 25min, and plateaus at ~2.7k.

Ours was unplugged, I think because someone assumed the ventilation sound was a white noise generator. If it was just for noise masking it wouldn't be needed: the sound isolation is very good. But since it's for ventilation we should be keeping it plugged in. It's still very quiet: I measure -25db on my Mac in Audacity with the mic volume all the way up vs -39dB with the ventilation disabled. Even with the ventilation running it's still slightly quieter than the office with no one talking, just light noise from the building HVAC (-22db). The only case I'd see unplugging it was if you were using the booth for sound recording, in which case I'd probably make a habit of opening the door to let it air out between takes.