Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[llminsk]

Is more technical info regarding Parametric EQ functions available? Like "what type of filter (algorythm) is used for low(high)-pass filters?"

Most of terminology (parameters) in this field are pretty standard but not all. For example it took me a while to finally find out low(high)-pass filter functionality under _PARAMETRIC_ EQ.

I am considering to switch to JRiver from other player and I need x-over functionality.

THANKS

[Matt]

Welcome.

All processing is 64-bit.  The relevant parameters like frequency, decibels / octave, etc. should all be configurable.

If you ask something more specific, I might be able to provide a more specific answer.

Thanks.

[llminsk]

Matt, thanks for prompt reply.

Particularly I would like to ask

1) what type of filter (Butterworth, LR, etc) is used for low-pass and high-pass filters. I guess this is not user selectable?

2) How is Q defined for low-shelf and high-shelf filters. Again, what is the type of slopes?

By now I finally have managed to get x-over working and sound is EXCELLENT. But I would like to make sure that my implementation is what you meant it to be. Could I put questions in this thread or please advise better place.

thanks!

[Matt]

1) what type of filter (Butterworth, LR, etc) is used for low-pass and high-pass filters. I guess this is not user selectable?

High and low-passes are Butterworth filters.  The order varies with the decibels per octave.

Lots more on Butterworth filters here:
http://en.wikipedia.org/wiki/Butterworth_filter

2) How is Q defined for low-shelf and high-shelf filters. Again, what is the type of slopes?

The Q is user-configurable for the shelf filters.  You could use DSP Studio > Analyzer while playing pink noise to see exactly what it's doing.  You can generate pink noise test files using Tools > Advanced Tools > Create Test Clips...

[llminsk]

ok, clear with filter type for low&high passes.

for low&high shelf filters let me re-phrase the question : how Bandwidth(Q) is mathematically defined for processing algorythm. In other words : if Q = (say) 3.5 (no measuring units defined in UI)  - what does it mean?

This is how I implemeted x-over:

1) In Win Control Panel set the device as 4-channel
2) In JR DSP Studio - Output Format - Channels  == 4   and Mixing == Clone side channels to rear speakers
3) In JR DSP Studio - Parametric EQ I added low and high pass filters applying them to front and rear channels. Corresponding device outputs are used for bi-amping.

Great sound!

Please approve this methodology :-) or advise another one.

And just for curiocity - what Parametric EQ 2 is intended for? Processing chain flexibility? No?

thanks

[Scolex]

Q is a defined amount which is just a representation of whether it is narrow or wide band (octaves). Here is a calculator that converts octaves to Q. Just in case you don't know an octave is the equivalent of doubling frequency for example 20-40Hz is an octave the next octave is 40-80Hz and so on.

[llminsk]

Scolex, thanks. So Q in low&high shelf filters is the same Q (Quality) as in bell-shaped Parametric Equalizers. Well, I have to confess that my question appeared to be rather trivial and that is probably why it took a while to get answer :-)

Now I am just curious - what for it could be used in audio _listening_ (not mixing/mastering) situation? Just trying to learn more from you guys :-)

And still for Matt : could you please comment on my approach to 2-way x-over as described in my previous post?

[mojave]

Parametric EQ 2 is if you need more flexibility. Some examples are you might want to mix channels before Room Correction, but then add some PEQ filters after room correction. You also might want to process a limiter at full volume (see option at top right of DSP), but have the PEQ done at regular volume. This would also require two Parametric EQ DSP's. You could also set up each DSP with different filters and then switch back and forth and decide which one you like best. This would be helpful for testing various crossovers for active speakers. You could set up one Parametric EQ with 12 dB/octave filters and the other one with 24 dB/octave filters and easily switch back and forth for a listening test.

The low shelf filter can be used to do what is called a house curve. According to the Fletcher-Munson curves, the lower you go in frequency the louder it needs to be for the perceived loudness to be the same. By adding a low shelf filter with frequency of 100 Hz and gain of 3-5 dB you can make a slope that increases the level as the bass gets lower. You can read more about the house curve at HomeTheaterShack.com. There is some disagreement about house curves and those against it argue that the music is mastered with the curve built in and no extra boost in the bass region is necessary.

[llminsk]

Clear with Parametric EQ 2 and shelf filters! Thanks, Mojave.

The only "theoretical" question left : why not to implemet Linkwitz-Riley filter for low&high pass filters. It seems that it is most common way for x-overs?

And again "practical" question - how it is supposed for 2-way crossover to be implemented (using 4 output channel sound device)?

[Matt]

And again "practical" question - how it is supposed for 2-way crossover to be implemented (using 4 output channel sound device)?

This gets a little tricky, because you really want JRSS to mix to 2 channels and use Parametric Equalizer to split the signal for bi-amping.

I think your best bet would be to select 2.1 output (which uses 6 physical lines, with three lines always silent).  Then, use Parametric Equalizer to copy L/R to C/LFE.  Then add the high/low pass filters on the L/R/C/LFE channels.

Does that make sense?

[mojave]

You can cascade two Butterworth filters to create a Linkwitz-Riley filter by adding two identical high or low pass filters. I'm not completely sure, but I think the problems L-R filters are trying to fix are more relevant to analog crossovers.   

[llminsk]

mojave, LR filter has flat amplitude response for the sum of low and high parts - as stated in the article you referenced. This is relevant for both analog and digital domains. Indeed, cascading two Butt 2nd order should give LR 4th order - Matt, do you think it is ok?

Matt, will try 2.1 option to construct my two-way amplification instead of current 4.0 and report the results in few days.

So far I can compare these two options:

1) JR                                      -> myAudio IEEE1394->AES (plain stereo)   -> one myDAC  -> myAnalogXover -> Bi-amping
2) JR with implemented Xover  -> myAudio IEEE1394->AES (four channels) -> two myDACs                             -> Bi-amping

While both options sound quite satisfying for my ears and Opt 1 could be also used for non-computer sources it is Opt 2 which gives more clarity/precision to the sound.

[llminsk]

2.1 option tested. In this case I have to add channel copying action inside Parametric EQ.

With 4 channel option selected in Output Format "tab" this action could be accomplished by setting on the same tab.

I havn't noticed any difference in sound and I suppose it shouldn't be.

By the way, adding second zone and cloning output from main zone with low and high pass filtering could be yet another way to build digital crossover. Isn't it?