FIR correction

[Wylandright206]

So recently I've become obsessed with the idea of making presets for my otops. We spent a lot of time working with FIR this year so I wanted to dive into that and see what sort of gains could be had from fancy linear phase processing. I downloaded Rephase and have just been going back and forth between that and REW.

Well up front I'll say that most of the gains to be had are from traditional IIR, parametric EQ. I found the most effective way to process these speakers is to flatten them out with parametric EQ first, then flatten the phase with phase eq. The IIR filters seem to smooth out some kinks in the phase that would require extra steps with FIR. Two birds one stone. I don't think the fir processing "sounds better" than iir alone, But I do feel that starting from a flat baseline before compensating for the room offers several benefits over simply integrating it into the room.

The first thing I noticed is the additional bass response from the otops. Yeah they always kick like a mule but there is no question the low end is more robust. I only chopped a few db off the top but even that made a big difference. The second thing is that you can recognize trends easier with your measurements in room. You know the lines should be straight so if there is a deviation, chances are it's probably the room or a boundary interaction. This makes reading measurements easier. Thirdly they sound better without any work. Lol in free air, it is a big difference in quality when you first plug them in. Tuning it in the context of the room results in (in my opinion) significantly higher quality results than just eqing in the context of the room without the preset. We watched John Wick on 4 corner titans and a pair of these processed otops and it absolutely blew my mind how realistic the otops sounded. I mean we have had these tops for years. That was abnormal and a total treat.

The main benefit to flattening these speakers out in the time domain would be in aligning the to other, unlike speakers. For example, you've got a Yamaha linear phase pa speaker that you want to use as a delay, it's much easier to phase align them when the phase of both speakers is flat down to 800hz. Obviously it's not perfect but it's better than a bunch of phase wrap and wrangling with all pass filters blah blah blah. In practice it is much easier. All in all I'd say that was absolutely worth the effort. It makes me like these tops more and I already loved them a lot lol

[Seth]

Yooooooo... Wyland! This is completely new to me. Never heard a thing about it prior to now. I'm not sure I completely understand it, but it sounds like the DSP can compensate in the realm of timing to manipulate/correct/compensate for the native phase response of a speaker cabinet. Am I understanding correctly? If so, that's super interesting! If not, would you mind giving me a brief rundown?

[Wylandright206]

Yes that is exactly what is happening! This DSP I have is FIR capable. Meaning you can introduce a set amount of latency, and use that pocket to correct for time discrepancies. You can eq phase and magnitude separately. Normally with IIR when you make any sort of change to the frequency domain it effects the time domain, as in minimum phase systems, phase and magnitude are linked. So even EQ bell curves will effect the phase. FIR is not the end all be all solution however when it does work it works very well. In this case I eq'd the phase separately from the magnitude. And I didn't use fir magnitude EQ either just IIR parametric. As I mentioned above it worked out some kinks in the phase to EQ phase and magnitude together. Then once I had a flat magnitude response I was able to flatten out the phase the rest of the way using FIR phase EQ. I could have used FIR eq to flatten out the higher frequencies but it didn't feel right in practice. IIR and fir can be used interchangeably at times but often one tool is better than the other in a given situation, as I discovered here.

Now there are definitely some drawbacks to FIR. If you are not careful you can introduce pre-ringing which is sort of a bizarre phenomenon. And the lowest frequency you can eq is totally dependant on both the Q of the filters you are using and the amount of latency you have introduced. Fir is practically useless for subwoofers because of this. Typically manufacturers will target 500hz and above with FIR. A subwoofer would have to have a significant amount of latency to use even FIR crossover filters. Something like +25ms or more I'd have to do some math. Simply too much latency to be practical. With a typical high frequency enclosure corrected down to about 500hz would introduce around 4ms

[Seth]

This is all fascinating stuff. It's funny to listen to my thought process around it all, figuring it out.

Well, won't it sound funny Seth? No. No it won't. It's not doing anything that's not already going on. Matter of fact, it's undoing cone to signal inaccuracies, potentially restoring the measured audio signal closer to that of the electrical signal... but with a slight delay I assume.

I'm really glad you brought this up because I love this stuff. And I'm really glad I'm shifting the main focus of my life away from audio, because this sounds like the sort of thing I could spend weeks down a rabbit hole researching and figuring all out. I'll leave that to you now and just live my audio life vicariously through you, if you don't mind.