Thanks I'm working with them at the moment.

I also got the latest SynthEdit upodate this morning.

Does it do multicore?

The Frequency Analysis object is not accurate enough in the lo freq to make it worth using. I tried it and took it out and shrunk the gui:



Unfortunately it seemed to have a noticeable impact on my old laptop test system. It was fun to make in SynthEdit but I'd prefer something similar written by someone who knew how to streamline everything and keep any side effects in mind etc.

I've done several simple object patches in SynthEdit and they always seem to run the CPU more than store bought stuff.

best,
mike

Nice demonstrations, Mike!

ORIGINAL: mike_mccue

Something I read on here some months ago inspired me to try the approach that your Sonitus illustration shows. I believe I used it on bass and an electric rhythm gtr bus. I quite liked the results. Just chopping off below the fundamentals that I wanted. In the mix, everything became more transparent. Perhaps "thinner," which can be both good and bad, depending on, well, lots of things. But it worked.

I believe this accounts for the overall lower level of the mix... I had to boost it (I used Boost 11) to get it to where it was comparable to other mixes at that stage of production. What was interesting was, it seemed easier to get it to sound "loud" without sacrificing the transparency and overly squashing it. As though the compressor wasn't responding to energy that it could detect but that wasn't actually contributing anything musically/aesthetically to the mix.

What's nice about the Sonitus for this kind of thing is that they're built in to the channel, though you only have direct access to the first 4 bands through the channel interface; you have to open the instance to get to the other bands if you need them. Can't say for sure if I took it to all 6 bands (I'm in self-imposed exile from DAWdom atm.) But the concept worked.

I can't say if there were any specific artifacts up in the harmonics... the "domino effect" you mentioned. But I figure that if I like what it sounds like, then any artifacts that are there are just part of the sound. Some of my favorite sounds were probably originally artifacts, if not outright "mistakes," so who knows. (Prior to this thread, I thought Butterworth was a maple syrup! So grains of salt are perhaps advised. )

Hi Tom,

I disagree mildly with your idea that low cutting aggressively may cause some sort of unnatural result.

Here's the situation that got me thinking about lo-cuts recently. I'm currently working with a Traditional Irish ballad ensemble. I have an acoustic guitar, mandolin, flute, hammer dulcimer and 3 vocals.

I forgot to add lo-cuts on my template and so found myself adding 12 instances of lo-cut to each .cwp file. We had 14 song/projects in this two day session. That's 168 instances.
As I was adding instances of Sonitus FX EQ to each track I was thinking to myself... "there has to be a specialized lo-cut that's super lite weight and actually CUTS the lows rather than just turn them down."

I got to dwell on this while adding each and every instance, all while realizing that the Sonitus FX EQ hi-pass slope wasn't steep enough for my confidence that the bass uglies would disapear.

Now here's the most important point to consider...

Recording on a nice analog system would have provided a built in automatic low cut. OK, I've never tracked on 2" tape and maybe that does go that low but I've also never recorded to tape and had sub 30 Hz artifacts swimming around for the rest of the production process. this is a circumstance that I'm now realizing is common in digital workflow.
For example, when I record to a Beta camera (yes we still send Beta tape to New York and L.A. with frequency) I never use lo-cut on my mic or mixer... I let the actual tape recording mechanism act as, for all practical purposes, a lo-cut filter. It going to do it anyway... so any cut I make in front of the tape deck is redundant (of course there are other compelling reasons to use lo-cut at the mic or mixer)

So, anyways, here I am with 12 tracks of acoustic musicians recorded with SONAR... none are playing anything near the bass range, but because the sound is made up of 12 tracks (which ISN"T natural) the bass builds up. I'm attracted to the idea of eliminating all the sub frequencies that don't really exist in that music. I WANT to filter out any proximity effects, the resonance of the house we recorded in, the jet plane 10 miles away, etc etc.

If it's true that there are no side effects than I'm definitely going to pursue using these steep cut-offs.

I've heard of stacked filters before... I know I'm just illustrating and old concept... I'm glad it's sparked some conversation.

I would like to learn more about the technicalities.


Hi Junski, Yes I saw a 4-6% CPU increase as well with the SynthEdit EQ I made. I know that would translate favorably to my new DAW but still it would not compare well to other EQ choices. :-(

"by the graphs, stacking five filters in GlissEQ gives result of a 20th order filter = 120dB/octave, though the roll-off is for 4th order filter ... not very 'sharp'"

It took me while to understand what you were saying... now I think I do. One thing, are you comparing the slope just using the graph? Or some practical knowledge about filter theory? Is it a 20th order filter? Or just similar to one?

Are you saying that the stock Sonitus filter is a 4th order?

best regards,
mike

Thanks Junski,

I grew up thinking in terms of Q and using my ears while working with hardware EQ... dB slopes and Hz values are less familiar to me when working with EQ's so it took me a while to follow your train of thought.

That all makes sense now.

best regards,
mike

ORIGINAL: mike_mccue

This is what I'm speaking/thinking about... I don't know enough about filters but it seems like there should be some easy way to ERADICATE any data below a certain frequency with a one click plugin while feeling confident you were not building a resonant peak near the cutoff.

Unfortunately your frequency analyser doesn't tell you the whole story. It doesn't show you the phase shifts and the group delays. Whatever type filter you use, there is always a trade-off. Butterworth have the aforementioned phase shifts and group delays. Linear Phase filters have pre/post ringing. FFT filters have imprecisions in the time domain (or the frequency domain) and possibly other artefacts. Etc.

Instead of trying to eradicate everything below a certain frequency, in my experience, it makes more sense to try and tame any unwanted frequency to an acceptable level while keeping the rest of the signal's integrity as well as possible. Staking five filters doesn't usually achieve that.

I'm usually happy with the Sonitus EQ's hipass filter with a Q of 0.7 for most material. For more critical stuff and working on lower frequencies like bass and kick drum I tend to use PSP Neon HR. For extreme stuff I might use the Sony Oxford EQ set to 36 dB/Oct. But that is very very rare.

UnderTow

ORIGINAL: UnderTow

...

Unfortunately your frequency analyser doesn't tell you the whole story. It doesn't show you the phase shifts and the group delays. Whatever type filter you use, there is always a trade-off. Butterworth have the aforementioned phase shifts and group delays. Linear Phase filters have pre/post ringing. FFT filters have imprecisions in the time domain (or the frequency domain) and possibly other artefacts. Etc.

Instead of trying to eradicate everything below a certain frequency, in my experience, it makes more sense to try and tame any unwanted frequency to an acceptable level while keeping the rest of the signal's integrity as well as possible. Staking five filters doesn't usually achieve that.

...

UnderTow

Here are some results I got from measures done for GlissEQ and for my test HP_Filter_VST using C-W. Budde's VST Plugin Analyzer (measured: Magnitude, Harmonic Distortion, Phase A/B, Group Delay, Time and Power). Settings are based on HP settings used in examples of Mike's animations.

Sorry 'bout the size of pictures in tables ... by right_mouse_clicking the image and then selecting "Show picture" you can see pictures in original size.


Junski

ORIGINAL: Junski


Here are some results I got from measures done for GlissEQ and for my test HP_Filter_VST using C-W. Budde's VST Plugin Analyzer (measured: Magnitude, Harmonic Distortion, Phase A/B, Group Delay, Time and Power). Settings are based on HP settings used in examples of Mike's animations.

As you can see, GlissEQ already affects the signal when no filter is engaged. My guess is that that is due to the oversampling. (High Quality mode). You can also see that as you add nodes, the phase shift plots get more and more jagged.

Mike, in the end it is really about listening. The graphs might tell you that an EQ causes a phase shift but it won't tell you how it sounds... My point was really that looking at frequency analysers might lead you to the wrong approach. Remove enough sub/bass so that you don't hear it any more and it doesn't affect the mix. But don't try and remove everything simply because it shows up on a frequency plot. That's all.

UnderTow