If you are thinking of an L-pad you will have to get to your tweeter. The L-pad usually goes after the last crossover component, on the tweeter circuit.
A series resistor and a parallel resistor to the tweeter. The aim is to reduce the db of the tweeter without adding in too many ohms. Usually calculators will try to get the L-pad rating around the rating of the tweeter. But the tweeter ohms will range from 3 or 4 to 16 or more over their working range, so it is not an exact science.
I just can’t live with accentuated treble. I think it’s possible to use a resistor on the speaker terminals to reduce treble output, though not too sure how to rig this up.
The ANJ’s we use sound lovely in the treble, though we still use these home made tweeter covers, variations of which we’ve used for 20 years on AN and other makes of speakers. It’s just a preference and probably a reflection on my hearing!
The idea behind this ‘Fred Flintstone’ approach is to reduce the dB output of the tweeter; really not too sure re the science behind this . Anyway, seems to work ok.
Personal view is that if the treble is audibly accentuated it is from the crossover point upwards .. so in my experience adding a 1 ohm or so resistor to the tweeter reduces the output and can resolve the problem. The advantage of an L Pad is the ability to have variable resistance and so alter the level and have more flexibility in fine tuning.
I am not sure how correct this is but it is from another forum where it suggested 3 ways of wiring in an L Pad. If it was me I would start with the simplest and if that didn;t do the trick try the next and finally the 3rd.
It more or less coincides with the BBC dip but it's unrelated.
The dip resulted from psychoacoustics research regarding the perceived distance of the phantom sources and was found beneficial for orchestral music (see Harwood "Factors in loudspeaker quality" in Wireless World 05/1976).
Some of the issues described in this topic relate to the range where the human ear is most sensitive as per the Loudness Contour Curves below:
What the graph above shows is that our ears are most sensitive between 700Hz and 5kHz with a broad peak centered at 2kHz.
If you look at the table below which illustrates how we perceive peaks and dips in particular parts of the spectrum you can read adjectives like forward, bright, harsh, edgy or sharp:
And this means that any of the following issues happening in that 2kHz to 5kHz will have audible consequences that may be difficult to tolerate:
1. Frequency respose anomalies (bump in the frequency domain) - depending on the cause, it might be possible to EQ a bump in the frequency respose.
2. Cone break-up resonances (bump in the time domain) - addressing cone break-up resonances will require a complete rethink of the crossover frequency and possibly a different tweeter that can be crossed lower or mid-woofer which break-up outside of the required range.
3. Off-axis flare (narrow directivity mid-woofer transitioning to a wide directivity tweeter) - this can be corrected by adding a waveguide, or possibly through the absorption of early reflections.
All of the above are visible in Stereophile's set of measurements.
Edit: Some (poorly designed) ports in (usually small) speakers produce resonances as high 1kHz that can be at the same level as the remainder of the speakers response.