It sounds like a less wordy, more vague version (as you can't visibly tell who exactly is humming) of the "aye"/"nay" voice voting.
Often in the sessions I've been in, there may be two hums taken - one for those in support of a question, and one for those opposed. Often there can be a clear distinction - but sometimes the results can be inconclusive if both sides seem to have equal volume. (But that, too, is helpful to the chairs as they can get a sense that the people in the room are divided.)
It's also possible that humming has little to do with sight and more to do with using the volume of humming to help indicate the strength of the sense of the room -- that actually is a very useful aspect of humming. However, humming became a thing concommitantly with participation by people with limited sight, which is probably why I remember that as a motivation, but in fact it might not have been! It's been a long time now that the IETF uses humming...
Or for people with a different range of natural and comfortable humming volumes.