Print

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

[KAyo]

Here we go. Wump – 23 Techniques

If you've finished with the listening and commenting page, then, let’s divulge our mastering approaches taken and why so ..

Apart from the settings, do take us slightly into the mind of the man behind the machine, regarding the approach. It just makes for better reading.

Cheers,
KAyo

[MoreSpaceEcho]

ok, my mind was saying "do this as quickly as possible and don't get bogged down in second guessing". i don't know about you guys, but i  really don't enjoy mastering stuff that i mixed...i just try and get it over with....

since i was going fast i didn't write anything down, but from memory it was:
sonoris eq -1.3 at 160? 180? somewhere around there. 1/2 octave q. this might've been on the M only but i think it was L/R.

out via lavry black into drawmer 1961, +1@3.2, +1@10, wide q for both.

drawmer 1969, 20ms attack, release was either 100ms fixed, or 200ms variable, meters just barely twitching.

fatso...input set very conservatively so it wasn't doing much. i think i had the 'warmth' kicking in a little bit on the louder cymbal crashes

back to digi via lavry blue, i let a few of the snares clip a little.

elephant, AIGC3, faster, 50% linked, 4x oversampling, 2db GR.

wavelab internal dither, no noise shaping.

[Patrik_T]

...

[KAyo]

I too wanted to keep it somewhat straight… apart from the plug-in debacle swap etc..
Preferences, was for the midrange tangle clean up and also for the high range to be smoothened out too, with a scoop at some point into the lower mid range etc..
Compress lightly and use very little limiter, if any. Find the gain via Input/ Output modules etc..
Also, watch out for Vocals and that it always stayed poking out.

I had a better time with the TCMD3 into the Masterpiece II, but, since I was judged on the submission, the settings below speaks the process.

Playback via Wavelab >> RME 800FW >>

into the Redline plug >
Low Pass: 37hz
124hz > Gain 1.8 > Q1.8
280hz > Gain -1.4 > Q0.77
1612hz > Gain 0.8 > Q0.78
4800hz > Gain 1.1 > Q0.44
High Pass: 18461hz

into the Masterpiece II >
Input Gain 2.2db

1515 EQ module
600hz > Gain 2 > Q2.0
310hz > Gain 2.5 > Q3.5

Tape Texture
Mid Only
slopes @ 200hz & 2.0khz

1900 Compressor Module
Mid and Low band only
Threshold -15db
Attack 60 mS
Release 175 mS
Ratio 2.1
Soft knee ON

Output Gain 2.1
HPF 60hz
LPF 12hz

>> RME 800FW >> Capture 44.16
No Dither


Thanks to all.
Always, glad to be part of the WUMP’s.

Ciao’
KAyo

[Herbeck]

My big mistake this time is that I wanted to tryout a few things.
I changed the settings on my sub, I wanted a little more low end(on my mastering).
I used a few notches on a harsh demo with some success a while ago, I wanted to try that again.
I've made a new plugin that I just had to try on this song.
Well, of these three things the plugin made the least damage, it may even helped the mix a little.

SRC to 176,4

The EQ:
-2db@200 Q71    -1db@2,7kHz notch    -1db@4kHz notch
low-shelf 3,5db@125Hz    1,3db@1,5kHz Q71    4db@35kHz Q35


The new plugin (I call it color):
This plugin adds color and/or saturation at any frequency and any amount, without changing the frequency.
Using a kind of convolution technique for the coloring part.

Added some color of a Massive Passive at 400Hz just for the hell of it.
Added saturation at 2,5kHz to push the vocal a little.


The limiter:
Non brick wall mode
Channels linked
Fast attack and release (maybe to fast)

SRC to 44,1

I think I used the POW-r 3 in Samplitude for dithering.


Conclusion: developing and tweaking plugins and stuff is fun, but it takes a lot of time and energy to do it.
I need to change focus a little and spend more time on the actual mastering part.

Cheers,

Herbeck

[Patrik_T]

...

[KAyo]

The record option in Wavelab ask's, what sample rate and bit depth. That's the one I use, if recording in Wavelab instead of Sequoia etc.. and if, it is a 24bit I am collecting as a 44/16, then one would hope RME's got that dither covered.

Thanks Patrik_T, for prompting the intrigue. Will investigate further.

KAyo

[MoreSpaceEcho]

why not capture at 24 bit?

[Viitalahde]

OK, here's mine.

My first impression was that the mix sounded good, and I found no major issues with it. Still, since there was a lot happening at the same time, I wanted to focus the sound a little. Like Patrik, I found the mix to be vulnerable to EQ, and I tried very minimal settings (because I still felt these things needed to be done!) but ended up doing quite a bit, compared to my usual gig. In a nutshell, I wanted the sound to groove a little around the  bass guitar and have the midrange be a bit more cracking.

The settings go like this:

BPEQ:

31Hz, -4dB, shelf (6dB/oct)
82Hz, +1.5dB Q2
1k75, +1.5dB, Q1.4
8k90, +1dB, shelf (6dB/oct)

..into Knif Pure Mu compressor:

Input +8dB, threshold 7 (midway), ratio 2 (2nd lowest), SC filter 100Hz, attack 6 (midway), release 2 (2nd fastest), link 0, dual release off. The GR VU was swinging under -0.5dB mark most of the time, a couple of times hitting -1db.

..into Western Electric 111C repeat coils

Capture @ 44k1, cleaned heads and tails off. I think I was the only one in my group that cleaned the stick sound from the middle of the tail.

Some additional limiting with Fabfilter Pro-L: Input +1.4dB, transparent mode, lookahead 0,96, attack 6100, release 168, transient link 0%, release link 37%.

-0.2dB ceiling, dithering down to 16bit/44k1 with triangular spread noise.

[Twerk]

Upon first listen, I was pretty happy with the mix as it was, and didn't hear any glaring issues that I immediately wanted to correct. After a few listens I started to feel the upper frequencies could pop a little more, giving clarity to the percussion. I also wanted a little more weight in the bottom. It was also pretty important to maintain the physical impact/feeling of the drums and not fold them into the vocal with overuse of compression.

In the Daw:

FabFilter Pro-Q:
High Shelf 1.0dB at 5.75kHz with a Q of 0.122
+0.5dB at 40Hz Q 2.0
+1dB at 80Hz Q 2.0

Then out of the HEDD192 and into Foote Control Systems P3S ME (compressor):
RMS and A/R timing modes mixed with an attack of 8, release of 2, feedforward compression, threshold -8dB, ratio 1.05:1, HPF engaged, output gain +4dB, output transformers engaged. Not more than 0.25dB of compression at any given point.

Into Manley Massive Passive:
10 O'clock position at 1kHz - Q fully counter-clockwise
9 O'clock position at 2.7kHz - Q 12 O'clock

Into the Manley Vari-Mu:
Compression mode, input at 2 O'clock, output at 9 O'clock, slowest attack, max threshold, recovery in the middle position, HP SC engaged and linked channels. Barely saw the needles moving.

Back into the HEDD192 with the Pentode setting at 3, captured at 44.1/24

Cleaned up the beginning and end, and applied FabFilter Pro-L: dynamic mode, gain +1.5dB, lookahead 0.25ms, attack 4093ms, release 59ms, transient link 0%, release link 100%, 4x oversampling, 16bit dither (optimized noise shaping), Output 44.1/16  at -0.3dB

[KAyo]

why not capture at 24 bit?

Big believer in capturing the sound at the intended target from analogue, without breaking the encasing again and applying dither and taking back to that delivery point. Over the years, I’ve concluded, unless otherwise… I prefer to capture it at the delivery target rate. Analogue is such a great medium to massage the audio, it acts as the best catalyst for sample rate or bit depth change. It’s not a hard and fast rule, in fact nothing is, it just sounds better to me, especially on this run.

Ciao'
KAyo

[MoreSpaceEcho]

..into Western Electric 111C repeat coils

i don't even know what this means. what does this mean?

i love reading what you guys did. a question for you fabfilter folks...what's up with the limiter attack times? thousands of milliseconds?

[Patrik_T]

...

[ggidluck]

Western Electric 111C

Old telephone repeater coils. Transformer saturation

[Viitalahde]

Old telephone repeater coils. Transformer saturation

Yes, they're transformers. But they will not saturate, I think I'll blow up my A/D converter before the transformer gives any sort of significant distortion.

Don't oversweat it, guys. The 111C's have only a very slight sonic stamp here, and the master could have well been done without it, too. I just like the way these things sound.

This is interesting. I initially found that either I had to let the EQ be out of the way or use all 5 available bands in order for nothing to "sink". I very much preferred the situation and sound of no EQ since the drift-from-mix became too much for me to accept.

Indeed. I remember the mix felt like it crumbled down easily, but when the EQ was right, it was right.

With mine, I'm still happy about the EQ. I shouldn't have crushed it this much though, and perhaps just leaving the limiter out would've helped. Doesn't sound too crushed, though.

Oh, I think I forgot to mention the D/A - A/D loop. I'm using the Crane Song HEDD192.

[Viitalahde]

EDIT: I quoted instead of modifying.

[Twerk]

i love reading what you guys did. a question for you fabfilter folks...what's up with the limiter attack times? thousands of milliseconds?

FabFilter Pro-L has a "transient" stage called lookahead, and a "release" stage that uses attack and release controls. The long attack times apply to the latter. In this plug-in, the attack and release are working on overall dynamics.

[MoreSpaceEcho]

thanks for the explanations guys. my simple mind can understand running a mix through transformers, but that limiter thing sounds complicated! i suppose in practice it's easy to set up once you've got it sorted though.

[Twerk]

thanks for the explanations guys. my simple mind can understand running a mix through transformers, but that limiter thing sounds complicated! i suppose in practice it's easy to set up once you've got it sorted though.

It's not as complicated as it seems, and they've done a good job of limiting (pun?) the amount of controls.

[Viitalahde]

It's not as complicated as it seems, and they've done a good job of limiting (pun?) the amount of controls.

Indeed, I find the Pro-L to be very intuitive, and I hate extra controls. This limiter has quickly become my to-go limiter, I use it for everything. Elephant can sometimes be technically more accurate, but the Pro-L just sounds better.

[KAyo]

I think this is exactly the kind of mix that either needs pretty heavy EQ or no EQ. It would be interesting to hear if anybody else found EQ to be hard to apply here.

Totally agree..
It felt at times as if EQ was a hindrance, but, once administered, then, one had to find a novel solution to resolve. I had good moments with the MD3, until I faulted and moved onto to the Redline and went cross eyed. My biggest impediment to date is the lacking of a sub set-up. All conversations on my end are null and void, until that’s fixed.

Ciao’
KAyo

[aivoryuk]

Hi all

Here is my technique.

I thought the mix had great energy and I wanted to capture that and just clear up of some of the small tonal imbalances.

My equaliser chain was
Sonoris parallel 11 Khz -0.5 mid bell just to tame the the drum cymbals

Abbey Road eq
256Hz -2 mid bell just to clean some mud out
1.05Khz + 0.5ish mid bell to bring the vocals out
4Khz mid bell +0.5ish mid bell to bring a little clarity

I didn't use comperssion initally and to get the level I used Gclip but did not use the input gain I turned the clip down to 33%
To bring the level up I used the barricade limiter but its just used as a gain stage no limiting takes place (every limiter has it's own sound even when they are doing no limiting)

At this point I was pretty happy but the bass end and cymbals were still poking a bit so I used the Waves Linmb at the start of the chain. I never use multiband as when I think it is sounding good I take it and always find it sounds better without but on this occasion I think it helped.
freq rage between 110Hz and 150Hz just diping every slightly.

For the cymbals which was inserted after the Sonoris I used Spitfish at around 11Khz

To finish I used Waves IDR for dither. I don't normally use this one but it sounded more like the original source then some of the other ones I tried.

Great track and I was pleased with my efforts as I haven't mastered anything for 6 months due to other commitments.

[fuse]

In short:

UAD - Massive Passive EQ emphasize mids +3dB @ 680Hz +3dB @ 3k9
UAD - Camebridge EQ Cut down sub low 40Hz
UAD - Multiband 1.05 expansion on mid en hi-mid bands
PSP - Vintagewarmer knee at 15% and drive -4dB and saturation on mid and high band at -3dB (took away too much high)
UAD2192
Ibis +1dB @ 1k48 -1dB @4k98
STC8 some gentle compression (prolly overdone it because it took away too much dynamics)
P38 in full limiting mode with TH just enough to move the needle every so gently
UAD2192
Ozone limiter with TH @ -0.3dB and Mbit light dithering

[ggidluck]

The track sounded a bit distant overall. When the bass started playing I got the impression that it was clouding up the track. I also felt like the vocals needed a bit of something to make the words more intelligible.

Up until this point I have worked totally ITB, and this WUMP was a bit of an experiment for me into a bit of analog compression. Eq was done in the box but I wanted to use the tubes to level out the track and increase the level.

I am using an RME Multiface (for analog in/out) and an ART PROVLA II for the tube compression. The VLA was upgraded to Mullard tubes shortly beforehand. (Hopefully soon an FCS compressor will be in the works).

Separate eq treatment was done for mid and side. Using Samplitude Pro 11 here. As far as method, in the software I bounce a file for mid/side and put each on a separate stereo track. Then eq is done in the object editor. Gain was set for mid and side at 2.2 db each.

===== MID =====

EQ116:
-6 @106 Hz Q=10 (notch out that bass a bit)
+4 @230 Hz low shelf
+3 @10K hi shelf

Parametric:
low cut @ 80 hz
+6 4500 Q=4.0 (add a little something for the vocal)

===== SIDE =====

Parametric:
low cut @ 120 hz
+6 @ 6K Q=4 (something to help the vocal here too)
+3 @ 9200 hi shelf

===============

out of the DAW into the PRO VLA
threshold -2.0 (about 2 o'clock position)
ratio about 2.5:1
output level +1 (about 12 o'clock position)
attack 0.25 (fastest it will go)
release 1 sec (3 o'clock position)

No compression or limiting was done in the box.


 

[DOMC]

im learning less is more from this

[Twerk]

im learning less is more from this

That's the case for just about every job I do. I think it's easy to get carried away with processing because we love the tools, the plug-ins, the theories, the process, the twiddling.

[Patrik_T]

...

[KAyo]

Yup yup..
Less is more, especially on this one. Good learning. Restrain is key many a times.

By the way, bought the sub- B&W ASW 650. Now, comes the task of getting it to sit well with the speakers etc..

Ciao’
KAyo

[nisilen]

i love reading what you guys did. a question for you fabfilter folks...what's up with the limiter attack times? thousands of milliseconds?

I can answer this as I developed two of the algorithms (transparent and dynamic).

I actually tried getting the fabfilter folks to "hide" the values of the knobs and instead just write fast -> slow at the edges of the knob, to force people to actually listen and not think "gee.. these values are ridiculously large and make no sense!".

The thing is, to make FabFilter Pro-L a hard clipper, well, ALMOST a hard clipper (the transparent algorithm will always be program dependent no matter what values you set), you'll need to open up the attack fully. Yeah, set it to that ridiculously large value of 10 seconds. 

The attack values really do not make sense because the knob itself controls quite a lot of internal parameters and basically it is a balancing knob.. or you could call it a mix knob. It balances the amount of audio that is sent to the various internal detectors. The fast detector/envelope path and the slow detector/envelope path. If you set attack to minimum, Pro-L's slow capacitor/envelope will be dominating and thus give you much longer program dependent release times. If you set attack to maximum then the short, almost instantaneous detector path will dominate and Pro-L becomes a clipper.

It's a bit tricky to explain but think of it as a balancing and timing knob that tells the limiter how large "chunks" of transients it will let through before it engages the slow envelope to prevent distortion from happening.

All this is probably explained better in the manual but I'm too lazy to open it up and paste it from there.

Cheers and sorry for the offtopic stuff! Interesting reading the WUMP techniques even though I haven't heard the song in question!

Cheers!
Niklas

[MoreSpaceEcho]

hey thanks for that explanation! that makes it much easier to understand.

[zeropoint]

dx:
needed some "room". bass needed to be thinned where there was overlap. cymbals tamed subtly. vox brought a little more out front.

prescribed treatment:

upsamp to 96 via izotope.

k-stereo max depth, mid width. -1 mid, +2 side, LC 145.
 
brought to the or for surgery:
plparEQ
made a deep and thin notch at 107, about -9 on the mid-channel, to thin out the resonant bass freqs, gain a little bandwidth. pretty close to 107 is where there was already a dip in the spectrum, so it seemed the most non-invasive and worked rather nicely. i couldn't find the right place without running an fft at 256k window length (mri with contrast). also a -2db, q2 bell at  7200 on mid channel to smooth it out a little. added a 4-pole HPF at 15 Hz to make the subs more stable and clear.

patient was cold and a little pale after this. added some color with colortone at the vmoo IR with 15 on warmth, 1.2 on color, 3a window mode, no trim.

to aid in convalescence, i used some glue via waves puigchild 670 at time const 5, thresh 0.9. input to meters barely twitching, max gr just a hair past 1.

fortified the diet with waves puigtec EQP1a, boost 4.5, atten 1 at 30 cps to give it some balls and some beef for the the kick; boost 3, atten 0 at 3 kcs to bring out the overtones in vox and axe. just overall vitality.

noise redux and de-click at the end via izotope.

downsamp via izotope.

limit via izotope, to max 1.5 gr at intelligent 2 release mode, character 3.5, inter-sample clip checked.

fade in/out with wavelab. cut off the extra fat at the end on a 75 fps marker, assuming final destination is cd.

tpdf dither to 16.

final save.


to do again, i would change the k-stereo to low-mid width, +1 mid, -2 or -3  side.
i would have ran the 670 to almost 2 gr max, and limiter to a 2 gr max, with same settings on the other parameters. i believe these changes in my treatment of it would bring it closer, without becoming a carbon copy, to the standards that seem to have been established in this group.

thanks again.