Improving the FIR Filter of My 4-Way System

[omousis]

Dear All,

I would appreciate your thoughts on the way the FIR filters in my system are currently being generated. From my assessment, the Group Delay and Impulse Response appear to be optimized correctly. However, I’m experiencing an issue with the sound imaging—it doesn’t seem as well focused as I would expect. The midrange and treble are somewhat aggressive, and the tonal balance could be improved.

I have included some screenshots of my filter design below for your reference. I would be grateful for any feedback or suggestions you might have.

Thank you,
Olivier

 

[omousis]

The treble issue visible on one channel has been resolved, but the sound imaging has not improved. The TTD correction was halted at 4000 Hz for the affected channel. However, the Group Delay still seems unusual, as the curve behavior below 100 Hz appears to be inverted.

 

[juicehifi]

Inverted gd in the below 100 Hz isn't that uncommon. It typically can happen at a strong cancellation. I can't say whether that's the case here, though.

About the focus. You should examine the measurement close if you haven't.

About aggressive top: That may have something to do with your target. Your target is globally very flat, but somewhat concave. This may emphasis the midrange or perhaps higher up. You could try a target that is straight between 100 and 6 kHz, and with 2-6 dB downtilt.

Furthermore, the measurement and correction window should always be set bigger than the TTD window.

And it may also be worth to try some of the following:
A 6-4 TTD window. For some reason this sometimes tends to "see" the acoustic challenges better.
Compare with and without TTD per driver. I have the imporression that without is the better option more than 50% of the time.

WIth and without selective preringing prevention. Sometimes without works better.

There is also an option to use a very short TTD window...

 

[omousis]

@juicehifi Thanks, I'm going to explore your suggestions with the CPD... However, the agressive top is noticed with all target curves I have implemented. I attach below a few screenshots with different target curves. The GD and IR are identical to the posts at the top...

 

[juicehifi]

Yes, there will be very small differences on gd and ir for a different target.

 

[Ofer]

Why did you choose to use a mid frequency? Also try to increase the TTD window like Bernt suggested, leaving 2 cycles difference between the 24khz and 10 hz. Look at the TTD default settings.