ORIGINAL: Razorwit

FWIW I was just reading an SOS article where Tom Elmherst was talking about what he did as mix engineer for Amy Winehouse's "Rehab" and he refers to shelving her lead vox at 40 Hz.

Dean

More for eliminating room rumble I would bet. Except for the grunts, or possibly residual mic effect there is not much below 150Hz in her vocals. At least no primary notes.

More for eliminating room rumble I would bet. Except for the grunts, or possibly residual mic effect there is not much below 150Hz in her vocals. At least no primary notes.

Like I said, there's plenty of ambient BS well above that filter bracket that worries me a lot more - unless the rumble is there, of course.

@mike

Maybe rephrase the question?

@junski

Long live Voxengo Glisseq - Span *with* the handles.... and you see it, indeed.

.

I don't think it can be done with Sonitus. using a low shelf on band 1
Freq = 80,
Q= 2=
Gain = -18 (max)
and a peak/dip on band 2
Freq = 94
Q = 1.8
Gain = -7.8
gives a curve from around 62 to somewhere around 140 hz so that's less than 18db/octave. Stacking additional bands can tighten it up to roughly 60 - 120 hz but it takes 4 bands. Logic's default eq lets you draw a 48db/octave sheer drop off just using one band which is pretty handy.

.

.

36dB/octave HP filter looks like this (here I use it in my software based RIAA 'stage'):




You can try this steep HP filter or even 48dB/octave HP (and LP filter up to 108dB/octave) by downloading the AiXcoustic Creations .:Electri-Q:. - posihfopit edition plug-in.



Junski

ORIGINAL: mike_mccue
It seems ( to me ) like a way to answer my question is too state something like this:

"Q factor describes the begining of the slope but will never exceed the max slope characterized by the filter design. dB slope relates primarily to filter order design."

The filter order describes the rolloff slope at the asymptote, i.e. far away from the corner and in the filter stopband. For highpasses and lowpasses, that slope will be N x 6 dB/octave, where N is the filter order. For (symmetrical) bandpasses, the slope is the same per side, but the order is doubled (i.e. a 2nd-order bandpass will have a slope of 6 dB per octave on each side).

Q controls the shape of the filter as the response transitions from the passband into the stopband, i.e. at the corner. As the Q increases from a small fraction to about 0.707, the slope at the corner will be typically less than the asymptotic rolloff. As it increases up to around 1 or more, the corner slope can become higher than asymptotic, and as it gets much above 1, the filter becomes resonant (highpasses and lowpasses will have a peak at the corner that's higher than the passband).

The relationship between Q and bandwidth really only makes sense for bandpasses. There, Q = the -3dB bandwidth of the bandpass divided by the center frequency of the bandpass (Q = BW/fc).

ORIGINAL: mike_mccue

Junski,

did you read that article about coding a low pass filter before suggesting it?


I'm still trying to figure out what a 36dB/octave filter sounds like... (or at least what Q and gain setting in Sonitus FX EQ would approximate it) and if you were teasing about it or not :-).

Perhaps you are stacking multiple instances?


best regards,
mike

No, I didn't read the whole story but as losgy replied later (post #46) ... part of what I meant to point out could have been possible to read from there.

Also, some terminology related to HP filter:

http://en.wikipedia.org/wiki/High-pass_filter


Junski

The problem of using a smooth highpass filter during mastering is that you can loose a lot of info in the lows if you set it wrong, so you have to be conservative with it. But with a very steep highpass filter you can cut the Bass at a higher frequency without affecting the lows as much. I'm gonna give the AIX Acoustic Creations EQ a try next time I master for sure. I didn't have an EQ this steep before, so it's gonna come in handy. Thanks for the link Junski!

Hey CJ,

I really like the sound you get on your tunes and it seems you're able to make great use of every ounce of headroom that's available. How big a part of that do you think comes from filtering as you've described? I'd like to experiment more with the approach you're taking by applying filtering on the tracks and busses.

ORIGINAL: mose

ORIGINAL: Junski
You can try this steep HP filter or even 48dB/octave HP (and LP filter up to 108dB/octave) by downloading the AiXcoustic Creations .:Electri-Q:. - posihfopit edition plug-in.

I downloaded the plugin from the link, but it does not seem to have the options you show in your picture. Are you using the full version?

It's the free posihfopit as the screen title says.

- clear a preset by removing all 'balloons' from the line (click the balloon and press del key from keyboard)
- set one point anywhere by clicking the line w/ left mouse key
- by right mouse key, click the 'balloon' so you get the menu available
- select the filter type
- double click the 'balloon' and yo can enter the parameters (Freq, Gain, BW)

As there is something odd in saving presets on Electri-Q, don't use those User 'preset slots' but use the "01 Default" for your own preset (it stays that way).

Specs for Posihfopit version:

Features:

* intuitive GUI
* up to 64 bands
* two basic algorithms ('Digital' and 'Analog')
* 64-bit Assembler optimized
* Economy Mode for less CPU Usage
* all basic filter types (Peak, HighShelf, LowShelf, Highpass, Lowpass, Bandpass, Notch)
* 30+ special filters
* Mono/Stereo (linked)
* exact edit of frequency, gain, bandwidth
* neatless Add/Delete of a band
* customisable Default
* A|B compare
* import/export posihfopit presets
* 50 Presets


The full version of Electri-Q has much much more.

Features:

* intuitive GUI
* up to 64 bands
* two basic algorithms ('Digital' and 'Analog')
* additional algorithms ('Scream', 'Linear Phase' and 'Maximum Phase')
* 64-bit Assembler optimized
* Economy Mode for less CPU Usage
* all basic filter types (Peak, HighShelf, LowShelf, Highpass, Lowpass, Bandpass, Notch)
* 80+ special filters
* Mono/Stereo (linked)
* Draw Filter Curve Estimation
* exact edit of frequency, gain, bandwidth
* neatless Add/Delete of a band
* Psychoacoustical Spectrum Analyser
* customisable Default
* A|B compare
* import/export posihfopit presets
* 72 Presets

NEW Features:

* New linear phase IIR mode (linear phase filter with low CPU usage)
* Chebyshev LP filter up to -72dB/Oct
* several bugfixes
* color theme support added
* performance of several filters improved
* undo/redo refined
* bypass added ([ALT] + click on mode)
* volume slider added for Winamp version
* installer includes optional WinAmp version as well
*
maximum phase mode *WORLD INNOVATION*
-The "Maximum Phase" mode is like the "Linear Phase" mode, but instead of having a linear phase, it has maximum phase. With this, the whole filter is preringing only. So a techno bass might turn from "boom" into a "whoop". If you now take into acount, that loudspeakers (especially sub-woofers) are minimum-phase, these effects will compensate each other. At least in theory...-


Junski

I don't know why this happens

I think Mike already spotted the reason for that (Chebyshev vs Butterworth, which is much smoother in any order).

ORIGINAL: doncolga

Hey CJ,

I really like the sound you get on your tunes and it seems you're able to make great use of every ounce of headroom that's available. How big a part of that do you think comes from filtering as you've described? I'd like to experiment more with the approach you're taking by applying filtering on the tracks and busses.

Thanx Don, A big part of it is from cutting and bossting, ala complimentary Eq. Another part is that i plan each sound before it goes into sonar, meaning the sound is what you hear, for the most part when im tracking in sonar and i make sure that each level of the gain stage isnt clipping or isnt too low. I usually have 3 to 4 gain stages before it hits sonar form using outboard gear. I EQ every track and every bus in every song ive done. Sometimes it just a highpass filter and sometimes im cutting and/or boosting.

The secret is that cutting is always better than boosting, because when you boost one frequency range, youll need to boos all the other frequency ranges to even out the spectrum...
Cj

I often roll off the low end. I came to this by simply thinking about what was down there. Not much useful stuff in most cases. There is a lot of energy that is wasted in the 40 and below range. It is important to adjust what is cut out depending on the material but most of the time there is little usable musical content at those frequencies.