Convolution Equalizer - linphase EQ for SAC?

[Julius Drescher]

Hi Guys,

I´ve got a question for signalprocessing pro´s.
After searching in the internet, taking measurements and asking fellow students, I got no satisfactory answer.

I´m developing presets for some speakers of the concert series of HK Audio (CN112, CT108) for next weeks production and came to the point with linear phase eq´s. Naturally, they have got latency which is too large for livegigs and isn´t taken by SAC.

If I build the preset e.g. with waves lineq and create an impulse-response out of it, load this into SIR2, it sounds equal. The amplitude over frequency is equal, too.
My question is: Is the result coming out of SIR2 with this IR linear in phase? SIR2 has a linear phase eq, and if you change somethin in it, the IR is modified. So in my opinion the end result should be linear in phase but I don´t know for sure :confused:.

(I know, it works fine with an IIR eq, too, but the diference in sound is interesting and I would like to test this live and not only in the recording studio ;-) )

Thanks a lot for reading and thinkin´ !

Julius

[brent]

Why would you want EQ presets for any speaker. The sound from that speaker will be effected by each space differently, and even the same space with different placement/locations. Your combing, flutter echo, room modes and nodes will also be different and will affect the sound. Just curious.

[Julius Drescher]

Hi Brent,

HK Audio has 2 different controllers for their speakers. One is IIR (DSM2060) and the other is an FIR(FIRNET) one. If you want so, call them "loudspeaker-management-system". You´ve got presets for different setups, so crossover frequencies and filters are set individually.(e.g. if you fly a line-array with 2 elements, the crossover frequency from bass to top is higher than flying 8 elements per side.) It´s the same with d&b and other professional loudspeaker systems. Sometimes integrated in the system-amplifier, sometimes as 19" controller.

You´re right if you say it´s senceless because every vanue has its own "character" but the presets I´m writing of are developed from HK to balance out the disabilities and resonances of the speakers and cabinets. And last but not least to fit different speakers of the same series (contour-series in this case) so you can use them mixed e.g. as frontfill and main pa without a difference in sound. Further more they are aligned to the lab gruppen fp 10000 so you can drive them full active without using the internal passive crossover and be sure the speaker is protected carefully by the internal limiter.

This so far.

My aim is, to save soundquality and latency (DSM and FIR have about 4ms; A Lavry DA11 sounds really nicer than the D/A´s of DSM and FIR!) and rebuild these presets in plugins I can use in SAC.
And yes, why not take a IIR for example reaeq or another latency-free one?
Would really be easier to do so because you can change settings live...
But after comparing up to 15 EQ´s both by measuring and hearing (all with same setting and same material of input) the phase linear ones are my first choice. First of all the lineq.(yes I know it has a bit of coloration:p).
The latency this one needs to calculate is 2047 samples....not more to say about it.
So my idea was to create an impulse response and convolute it over the audiosignal. I´m satisfied with what I hear, but I haven´t got the chance to measure phase and -as written- don´t know for sure what SIR does.

But perhaps one of you knows ;-)

I hope I could bring a little light in this strange plan.


Thank you!

[dasbin]

It seems to me impossible that the audio coming out of SIR2 would be both linear-phase and latency-free.

Note the very wording of "linear phase." It doesn't say "NO phase." The idea is that all frequencies are delayed by the same amount, and my understanding is that it is physically impossible to digitally affect frequency response (no matter how you do it... SIR convolution included) without affecting the time domain in some way.

Also beware pre-ringing associated with linear phase in general; it can be an even worse audible effect than a gradual phase smear.

[TomyN]

Hi,

well, the tradeoff of FIR and convolution is latency, something you can not avoid, because you need some input to 'convolute' the result.

Tomy