Large or small?

[lovinthehd]

The principle holds at all frequencies, although auditory bands tend to be wider at lower frequencies, so subwoofers are especially good at masking higher frequencies.

I get the general idea, what I'm having trouble with is why would bass be so elevated during dialog in a movie?

 

[shadyJ]

I get the general idea, what I'm having trouble with is why would bass be so elevated during dialog in a movie?

That's a good question for Christopher Nolan or Micheal bay.

 

[JerryLove]

The masker is a sound that 'hides' all other signals around it by dominating that auditory band in human hearing. In that graph, it is the tone that masks the darker bar at around 400 Hz. Every frequency sound with an amplitude that lay below the solid line would be rendered inaudible by the masker.
Here is a cruder graph that also illustrates this effect:

We have this phenomenon to thank for us humans being so insensitive to harmonic distortion. As you can see, most masking occurs at higher frequencies than the masker although lower frequencies can be masked to if the masker has sufficient amplitude.

My first curiosity is whether our FQ sensitivity should be applied? I would think so.

I assume that infrasound won't mask audible sound; and if that assumption is correct, then it's likely that we need to apply a FR curve.

Also: I notice we are talking (assuming that graph is real values and not an abstraction) about a -10db to -15db per octave.

The voiceband starts at 300hz (if you look at the old POTS system for phones, they only reproduce 300-3300hz). So I'm thinking the sub would have to be very too loud and one of the give-aways that this was going on is that low voices would be far more illegible than high (men vs women, for example)

 

[shadyJ]

My first curiosity is whether our FQ sensitivity should be applied? I would think so.

I assume that infrasound won't mask audible sound; and if that assumption is correct, then it's likely that we need to apply a FR curve.

Also: I notice we are talking (assuming that graph is real values and not an abstraction) about a -10db to -15db per octave.

The voiceband starts at 300hz (if you look at the old POTS system for phones, they only reproduce 300-3300hz). So I'm thinking the sub would have to be very too loud and one of the give-aways that this was going on is that low voices would be far more illegible than high (men vs women, for example)

The graph in the post that you quoted is an abstraction that was created to show how auditory masking can obscure some harmonics of a fundamental but not others. It doesn't purport to indicate actual masking levels. Masking levels change based on amplitude of the masker, among other factors such as frequency and hearing health of the subject. If you want to see figures taken from actual research, here is a very good article on the matter.