Print

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

[Phillip Graham]

Ok,

So I am going batty here in the office on a late night of work, so I am going to perhaps jump the gun, take a break, and write what I did processing-wise for WUMP 6.

This was the first WUMP I have participated in.  I really liked the song, and the raw mix, and I thought it would be a fun challenge.

I do a fair amount of archival clean up of live sound recordings, as I have mentioned here before, so those skills come to bear here.

Signal chain-wise, I broke all the "rules."  My WUMP submission was done on a pair of Grado SR80 headphones plugged into my IBM laptop here at the desk in my office.  The entire signal chain consists of adobe audition 2.0, and included or free plugins.  Audition is a 32-bit floating DAW program with VST support.  I double checked my stereo image decisions by burning a copy and making a quick run down to my car in the parking lot.  I did lots of referencing to the raw file, and kept the general adage of "do no harm"

Ok, introduction aside, here is the processing.

I first split the track mid-side using VoxengoMSED:

Mid-processing
1. Audition has a built in LINEAR PHASE GRAPHIC eq, with adjustable procoessing precision.  I choose 4000 points, and 20 bands.  I mention ONLY the adjusted bands below:
31hz +3dB
44hz +1.5dB
63hz -.5dB
125 -2dB
180 -1dB (would do more on a second run at it)
500 -.5dB
710 -1dB
2khz -.5dB
2.8k -1dB
4k +.5dB
8k +.5dB
11.3k +1.5dB
16k +2dB

2.  A digitalfishphones Blockfish.  Air on, no low cut, minimum saturation, slowest attack release, opto mode, complex mode.  
This only kicked in on the creshendos and served as a faster alternative to automating them down.

3.  dfphonse Dominion. +3 attack at 5ms, +5sustain at 229ms, full soft saturation, saturation 25.  HF 5k at level 6.  This  was to give a touch of snap, It is barely noticeable before and after with close listening.

Side processsing
1. 20 band linear phase again.
31 +1
44 +1
500 -1
710 -1
2k -.5
2.8k -1
4k +1
8k +1.5
11.3k +1
16k +2

2. Same as 2 above

3. Similar to 3 above except at attack of 9ms, and HF at 6.8khz

4. There is a large plate reverb on the side buss, a built in audition plate verb.

Main bus processing
1. Mid Side decode with voxengo  (Mid attenuated .4dB using audition send)
2. DFphones endorphin compressor.  saturation 25, soft mode, opto mode, attack 8.3ms, release 155ms, compression low and high both at level 57, output drive 1.9dB.  This was kicking in 3dB of reduction on the HF band at the loudest points.
3. Audition internal multiband compressor (based on izotope ozone).  Used only for the limiter (bypassed all 4 multibands).  Limiter is in brickwall mode, attack 5ms, release 200ms, margin -.1dB.  Worst case GR is about 4dB.
4.  Spliced a high diffusion ambience verb into the cymbal decay tail, then did an automated log fade out.
5.  Downsampled with pre/post filter, and dithered down to 16bits using Audition's triangle dither, no noise shaping.

And there you have it!  I will reveal which track is mine when the time is right.  I wish i could say it was my favorite, but it's not.  I do feel pretty good about a pretty "natural" final result that compares well with many of the other masters.

Night!

EDIT:  Brad says to now open up which mix is ours.  Mine is 1389

[Patrik T]

Shouldn't we maybe hold the tech presentations until the feedback part is done and Brad's given out who's who?

Best Regards
Patrik

[ATOR]

Patrik T wrote on Wed, 26 July 2006 12:50

Shouldn't we maybe hold the tech presentations until the feedback part is done and Brad's given out who's who?

If we wait untill after the feedback round we could also mention our number with the tech specs. Then you don't have to search multiple posts to link the tech to the number.

[bblackwood]

OK, let's open up the discussions...

[Ged Leitch]

Ok, here goes...

MIX >
well i loved this mix, the playing was superb and the cello's were nice and big sounding.However, there was too much low mid clutter and resonances muddying up the articulation i think.

so...>

PLParEQ  

+3db @ 68Hz bell Q 2
-1db @150Hz bell Q 0.5  
-1db @2K    bell Q 0.5

didnt want to cut all the warmth away here, so had to be gentle.

Impulse response EQ - Massive Passive shelf's

ok these sound great for boosting, but i DID overdo the hi shelf.

+ 2db @100Hz shelf

+ 2db @16K   shelf < : i have since found that this particular shelf is a really narrow Q, it cuts the upper mids but boosts the air too much, just like the special shelves in the actual massive passive.

Tube limit
Freeware thing, sounds nice when used very sparingly.

Compression @ 5%
Saturation - OFF
Bright -     OFF
Output -      no makeup.

Marquis compressor

Ratio - 3
attack- 30ms
release- 250ms
Knee -20db

3db makeup gain.

about 1db of GR at maximum, just to tame the crescendo's a bit as requested.

Ozone Limiter

Took exactly 1db off the top with this, set to intelligent mode.
Prevent intersample peaks ON.

SRC with R8Brain free and edit and dither in Samp.

In retrospect, i'd be happier with just a gentler 1db + @16K bell, as that shelf was too much.

But hey this was a GREAT track, and im learning from this all the time.

[Viitalahde]

OK, here's mine then.

>Playback at 48kHz, output at original level to Lynx DAC

>Virtal

[cerberus]

Viitalahde wrote on Mon, 31 July 2006 16:21

stereo linking off

i've been prejudiced against dual mono dynamics processing for years; but of course my amplification is configured this way. i've never heard of an m/s amp in fact; and i wouldn't use a "linked stereo" one. thanks for showing us that; i think it will be the next limiter experiment i will try.

jeff dinces

[Viitalahde]

cerberus wrote on Tue, 01 August 2006 02:29

i think it will be the next limiter experiment i will try.

Yeah, the artifacts of unlinking start to show only when you're hitting really hard - and you're not really supposed to do that in the first place. I think I run the limiter unlinked 95% of the time, and the rest I run maybe 20% linked or so.

I still got no dual mono compressor in my rack, but soon will..

[Luke Fellingham]

Here's what I did:

R8Brain SRC - Unfortunately there's something in my system which isn't working well with high sample rates at the moment which I haven't yet had a chance to look in to yet. Because of this I needed to SRC before doing anything else, I suspect in an ideal world this might have been better done at the end but on this occasion that wasn't an option for me.

Convert to M/S -
TC eq +0.5db@ 5KHz med q on S channel
Raise gain on S channel +0.4dB
Convert back to L R
Although I placed this M/S processing at the beginning of the chain, it was actually the last thing I executed. When I had finished my initial attempt I just felt that the image was a bit closed in and I wanted to draw a bit more attention to the side elements. What I did was fairly subtle but I think it was just enough to achieve this goal. My choice of the TC eq plugin on this occasion was a matter of convenience in that it is the only one I have which can have the two sides unlinked.

Sony Oxford eq - HPF 30Hz, +2.7dB @ 97Hz q 3.4, -0.9dB @ 157Hz q6.3, -1.4dB @ 205Hz q3.6, +0.4dBHigh Shelf @ 2KHz
I used the Oxford eq to help reduce a couple of resonant areas. I also boosted some low but with a fairly high q, I wanted to keep the bass solid without muddying up other stuff too much.

Out Via Lavry Black

Avalon747
Compressor - Low ratio, slow attack and release, tube switched in. Max 1dB reduction. My usual settings, not really doing much to the dynamics but warming things up just a little.
Eq - (no scale in dB) Low shelf boost @15Hz, small boost @ 500Hz, small shelf boost @5KHz, small shelf boost @ 32KHz

Gyratec G14 - Small narrow cut at 230Hz, small wide boost at 2KHz, small wide boost at 18KHz

Apogee AD

UAD Precision Limiter - about 1.5dB reduction
Sony Oxford inflator - Input 0, Effect 100%, Curve -14.2, Output -0.1
Set like this the inflator raises the overall level but with a slight bias towards transients. With a positive curve setting it would have been louder but not so dynamic.

[Luke Fellingham]

Oops I forgot to mention the dither!
Pow-r 1

[Pingu]

Viitalahde wrote on Tue, 01 August 2006 14:07

cerberus wrote on Tue, 01 August 2006 02:29 i think it will be the next limiter experiment i will try. I still got no dual mono compressor in my rack, but soon will..

Mastercomp is not bad for this.

[ericjenson]

ok

I liked the piece very much and enjoyed working with it,
so first off, thanks for supplying a very interesting mix.

split to m/s

side channel:

eq with electri-q beta 1.54(no longer beta at present)
thanks to christian budde for letting me try out his plugin on this.
anyhoot:
low cut bug pass linear phase at 30Hz, cleanup
-1 at 60Hz with linear phase triangular at0.33Q,trying to make the bass more controllable

-1.2 at 120Hz with the same, again, bass control

-0.8 at 240Hz with the same, just sounded good while sweeping through, so i adjusted gain to taste.
+0.4 at 3500Hz "    "   ", i think i settled here to make up for the cuts in the bass region of the side channel, again found through sweeping with the filter.


mid channel:

same eq plug as above:
+3.6 at 7400Hz  linear phase triangular at 1Q, brought out some shine, maybe too much; guilty...

+1.9 at 500Hz   with the same, it just needed the body and some mud, i was probably compensating for the aforementioned boost at the top end, to try and keep the balance,  failed i think...

combined to l/r
waves linMB
created a band at 100-400Hz
att: 40
rel: 400 opto manual, was looking for about 2 dBs of reduction max to tame some resonance, in retrospection i think the band should have been 60-400

created another band from 600-6000Hz
att: 60
rel: 600 opto ARC, for a little over 2 dBs of reduction at crescendos, in order to let the toms punch through and perhaps gain a tad of headroom at these louder parts.
no makeup

PSP MasterComp

-16.7 thresh
att: 50 auto
rel: 600 auto
100% linked RMS mode, FAT turned off
sidechain lowcut at 350Hz
this was an attempt at autoleveling the crescendos, seemed to work ok.
got about 1.4 something GR from this at the crescendos, anymore and i think the compression would have been audible.

scanned for peaks
and nipped 2dBs with an L2 plug, release at 60 ARC,
did the left and right seperately with same gain so essentially unlinked, cieling at -0.1
this was to even out some peaks for..

Magneto:

input +0.1
i.p.s. 15
drive settled at 2.1, was trying to be a little more conservative with this plugin this time around.  wanted some tape compression, not distortion.

SpectraQ:
enabled only the limiter section, took off another 3.6 dBs with this
cieling at 0.6

downsample to 44.1 via Audacity high quality sinc interpol

waves L3

thresh at -0.9, cieling -0.1, did nothing to the bands or releases, default settings here.

fades done in sequoia
dithered to 16 with pow-r 2

and that's it



this was a good practice piece and helped me to see some areas where i'm lacking.  this is not the type of stuff i usually work on and i like it for that.  good sharpening stone.

i think my first mistake was using a plugin i'm not familiar enough with for the EQ, with the PEQ orange i know i would have heard better results. lesson learned..

next?

4391

[ATOR]

This is what I did:

SRC with R8brain to 44.1 (my sound-card doesn't do anything higher than 48kHz)

I separately edited the cello part at the beginning with
narrow PLPeq cuts at 100, 200 and 400Hz  of 1 to 2 dB

PLPar EQ
63Hz / 1.7dB / Q0.8
196Hz -1dB Q1.6
339Hz -0.7dB Q0.9
727Hz +1.3 Q0.5
7533Hz +3dB Q0.5

SIR
0.8sec room ambience reverb

UAD Fairchild
timeconstant 1
L&R unlinked
GR of max 2dB

Here in the chain I automated the gain to enhance the dynamics, I pulled down the beginnings of the sirens and the quiet middle part. Steve wanted the first kick/cello attack to hit zero so the only way to enhance the siren dynamics was down.

UAD PrecLimiter
rel 0.05
GR 1.5dB

Apogee UV22HR dither


That's it


edit: to give this some meaning, my master is 2323

[cerberus]

 as some regulars know,  i have been using wump for a science lab to experiment with very complex chains. (e.g. including entire gangs of faders, plug-ins and routing paths such as an "ersatz k-stereo" section). so i took a tip from chrisj and tried to see if i could get by with removing sections of the chain; to simplify; for  efficiency.

but it didn't work.   and i also was stressed and rushed from deadlines, when i finished three hours earlier than i expected, it was a relief because i  was exhausted; but it didn't work.  one of my "techniques" even failed completely at one point...it needs to be revised to accomodate all situations.  

now i think that cutting these corners is not a path to success.   maybe next time i will try and cut some others, or perhaps not. i hoped it wouldn't show, but the reviews are scathing in terms of attributes that  usually i am said to control better.

jeff dinces

[mbruce333]

Allrighty then, here was my take on the whole things...

First in Sound Forge 8, I wanted to do a little work on the gliss sections.  In Steves request, he said they needed to be brought to a similar level of the rest of the tune without losing the impact of the crescendos, and they were a bit brittle.

So, in SoundForge 8, just on the gliss sections:

Waves Ren EQ, -2.4db, 3750hz, Q=3.86 to take off some edge of the violins

Waves Ren Comp,  1.44:1 with about 3db reduction on the peaks of the crescendos.  

Saved as a new version, than opened up in a WaveLab 6 montage:

Waves LinEq Lowband:
HP@32hz with a slope of .26
1.5db boost at 96, Q=.9
-1.3db at 419 with a wide Q of .29

I wanted to help clean up some lower midrange mud and try and "under" the cello's to give the kick a little help to drive the track a little more.

External gear plugin wiith a 3db gain which feeds my usual setup...

Crane Song HEDD D/A

Manley Vari Mu to try and get a little more glue going.  Med-slow attack, slow rel, 2-3db of reduction, maybe 4 on the peaks.


Millennia NSEQ-2 Solid State Mode:

Low Shelf, +4db at 100
-2db at 200hz, Q=1.5
+2db at 7500hz, Q=.6
High Shelf, +4db at 10K

The goal was to continue to clean up the mud around that low "G" and try to get some air in the top end.  I think I could have been a little more aggresive in my cuts, and the HF boosts, though.

Crane Song HEDD A/D
No Triode or Pentode, Tape at 3.  

Trying to get a touch more "glue."


Back in WL6:
Voxengo Elephant:
In +4.5
Out -.3
Oversample = off
Mode AIGC-4
Lim Speed = Med

When listening to the gliss sections, I decided that I wasn't quite happy with impact at the end of each section. So, back in SoundForge, I used an L2 to just bump up (+2db) the kick drum hit that provides the impact at the end of the gliss sections.  I just wanted a little more "umph" to help release the wonderful tension built in the glissandos.

While I felt this worked pretty well for the first section, that same hit and the end of the second section just isn't as solid, so it didn't work as well there.  I thought about trying to copy the kick from the first one and mix it into the second, but felt I would be over stepping my place, so I went with the same L2 treatment as the first one.

I felt the head of the tune was a little tight with only 29 or so ms, so I added another 100 ms to keep the front edge from being clipped off by slow CD players.

R8Brain to conver the sample rate, fade in/out and dither in SoundForge.

This was a really good WUMP, tons o' fun!

Mike Bruce