Linear modes sound strange when in mid/side mode
Linux Studio Plugins Project
Brought to you by:
sadko4u
The linear modes in the parametric eq sound a bit strange to me. In the attached files I have boosted a high shelf to 8db at 450hz using RLC BT filter type on the sides only. I compared to izotope ozone's linear eq and set that up with the same settings (as much as possible) and the result is very different. It sounds like ozone is boosting much more than LSP is but apart from that the difference is quite big. LSP eq sounds like it is introducing quite a big delay/latency between the mid and the boosted sides, if you listen to the high hat it sounds like it is flamming almost whereas the ozone eq keeps it much more under control and you don't hear the delay.
Yes, LSP equalizer adds high latency to the output in linear mode because implements digital filter with finite impulse response. That means that impulse response of the FIR filter with linear phase is symmetric in the timeline with peak in the center. This yields to two mutually exclusive things:
1. Frequency characteristics of FIR filter are much better when the impulse response is large.
2. The longer impulse response is the longer latency FIR filter adds to the input signal.
So currently I know no other way to implement FIR filter with ideal linear phase.
This may be fixed by latency compensation but all equalizers do report latency to the host.
Also I don't completely understand your setup. Do I right understand that you mix processed sound with the unprocessed sound?
Last edit: SadKo 2016-09-23
Also, here is a good FAQ about FIR filters:
http://dspguru.com/dsp/faqs/fir/properties
No, the files above are completely wet/processed, ust one track running through the EQ.
That's why I was a bit confused with this behaviour. I maybe misunderstanding linear phase EQ (you definitely know more about this than me) but shouldn't the bands inside the EQ instance remain relative/in time? I understand that the unit itself would introduce latency relative to other tracks but I don't think I would expect to hear a noticable delay when using a filter on one part of the signal, eg the side shouldn't be delayed compared to the mids when 100% wet signal is going through one instance of EQ, should it?
Last edit: Alex 2016-09-24
Yes, it's right, in the same EQ both lines should not be delayed relative to each other.
Alright, now I'm sure that you do not mix dry and wet sound, and it's good. I'll try to reproduce the problem in the nearest future and fix it.
Okay, reproduced the problem and already fixed it.
You may download debug version from this folder that already contains fixes:
https://sourceforge.net/projects/lsp-plugins/files/dbg/
Also I'm waiting for your feedback that all is OK now.
That definitely seems to be working as expected now, thank you for the quick fix.
Nice, will close after new release.
Fixed in release 1.0.14.