With how great speech recognition is becoming, it seems like this is something remote workers could easily discreetly do since our conversations tend to be stationary, through a computer, and with only a small part of our body visible. Just wire up some electrodes to zap you every time the computer detects filler. I'm now seriously considering doing it myself.
Neat! Without the electrodes I don't think it would be effective for me for "uhh" / "uhm". Considering how unconscious filler words are, I think I'd need the immediate unignorable feedback. But you've got all the logic there, it just needs to be made more violent.
It would be about as easy, and certainly less painful, to just have a video processor remove and smooth over filler words in real time.
If the filler words are excessive it would slow down the apparent rate of speech, but obviously not the real rate of speech, by definition, since we're only removing words with zero semantic value.
I don't think it's an argument of efficiency but rather the avoidance of noise.
The "ums" isn't redundant, it's not repeating or decorating the conversation. It's filler like static. Stops people from filling the gaps with their own thoughts
