my mastering chain
First I will do a transfer from analog if needed. At this point I add 1sec silence to the beginning & end (or cut each track to that) and will run any noise reduction if needed. I use Adobe Audition v1.5 for all of my cuts & fades. Version 2.0 (at least the demo) seems to have much fancier graphics at the expense of speed. I prefer function over form. But v3 is out, I should try the demo one of these days.
http://www.adobe.com/products/audition/
Example of an original recording:
http://ictybtihky.com/harbal/examples/a ... iginal.mp3
Next I go into Har-Bal. My method is somewhat like the Emphatic technique from the help files, but instead of adjusting each individual 1/12th octave, what I do is make the center areas of the peaks & valleys (on the peak power plot line) be average across the overall curve that I'm going for. But I do not correct for any roughness between large and small deviations that may be right next to each other, unless it is audibly a problem. This helps to preserve the dynamics of improperly mixed sounds due to them being improperly harmonically balanced. This allows me to correct for that problem, without taking away from the original intent of the mixer in regard to that particular sound's harmonic relationships.
Here's an example of what I mean...
http://ictybtihky.com/harbal/examples/a ... arbal1.gif
I have smoothed out the averages across a few peaks, but I havn't smoothed out the difference between the peaks. I have allowed there to be extra large peaks & valleys, and also extra small peaks & valleys in the middle too. You can also see the frequency response of the filter, here:
http://ictybtihky.com/harbal/examples/a ... arbal2.gif
An example of a har-balized recording:
http://ictybtihky.com/harbal/examples/a ... harbal.mp3
Now I move into the finalization stages. My main goals in Har-Bal have been to smooth the balanced of the harmonics, but also to compensate for particular sounds/instruments/tracks that didn't have proper spectral/harmonic balancing which caused them to be improperly mixed into the recording. This is often done without priority for the overall larger audible spectral balance, because I'm using har-bal mainly as a tool to re-align those problems in the mix, but also the spectral balance from an engineering point of view as opposed to only audible. They both need to have compromises made while doing this. I don't want to create something that sounds so bad that it can't be compensated for in the finalization phase, or something that would create a sound that's too far from what the original artist had intended.
So for the finalization stage, I use (drumroll please) Winamp as my host. Before you think I'm crazy I think I should mention that I've tested the setup and found that it makes a 1:1 perfect copy of a WAV file when it's run through it using Nullsoft Disk Writer output. So quality is not an issue at all.
I use Nullsoft in_wave.dll v2.06 because the new ones only output 16bit or 32bit, not the input WAV file's bit-depth. It's the last version they released without the new "feature", and you'll need it if you plan on playing back 24bit WAV properly. You can get a copy I have here:
http://ictybtihky.com/harbal/apps/in_wa ... .v2.06.zip
I use the Adapt-X plugin chainer v3.61c
http://www.chronotron.com/content.php?p ... cts#adaptx
with "force 24bit floating-point" enabled.
I also use DirectiXer for making VST-only plugins available to DirectX hosts like Adapt-X.
http://tonewise.com/
Now for the plugins, I'll list the main ones I got to often, and
I'll put an asterisk next to the ones that I only use optionally if needed. I'll also be linking to screenshots of my default starting settings, if applicable or not the plugin's own default.
Voxengo PHA-979
http://www.voxengo.com/product/pha979/
this is a pretty simple plugin, it adjusts the phase of the channels, pan, and also has a gain control. it's useful for adjusting the center from analog transfers and/or acoustic recordings, and compensating for the slight phase tilt that usually happens with vinyl & from tapes recorded with badly aligned heads. a good technique is to use MSED to remove L+R (mid) and adjust the phase and pan until the bass/center is the most quiet.
Voxengo MSED
http://www.voxengo.com/product/msed/
this is a pretty simple plugin as well, FREE and it can adjust the gain on mid & side.
my own custom 2-pass-band de-esser (*)
(sorry, no public information is available at this time)
Waves Renaissance Bass (*)
http://www.waves.com/Content.aspx?id=194
i rarely use this, but i use it on some tracks that just can't seem to get up enough gumption in the sub-bass department through EQing, without bringing up unwanted or otherwise "problematic" sub-bass issues. it's definitely resonant.
I can't post a screenshot of my defaults because I created an XML preset and then edited it by hand, because there's about 4x more controls in most of the Waves preset files than what you see on the GUI. I basically took the default sound, and made it have more resonant bass yet to have much less mixed into the signal without having to reduce the "Intensity" knob at all.
Waves MaxxBass
http://www.waves.com/Content.aspx?id=327
i pretty much use at least a little of this on everything. It's really great, and accurate & honest to the bass input, when used so there's just enough to augment the bass, not replace it. I also don't use this very often in the rare cases that I use RBass. My default
http://ictybtihky.com/harbal/pluginshot ... xxbass.gif
(note: "HighPass" is set to 12db)
BBE Plugin v2 (*)
http://www.bbesound.com/products/maxim/newplugin.asp
I use this sometimes near the end of tweaking all these plugins, if I can't seem to get the right snap or crispness out of a track with EQ. it's basically a distorted compander, with phase-sync on the harmonics. I very rarely use more than a setting of 1 on the "process" knob. I never use "lo contour" because it ruins the midrange.
Nomad Factory EQP-4
http://www.nomadfactory.com/products/an ... /eqp4.html
This EQ is amazing. It's a hybrid layout Pultec EQ emulation. I think if I had to pick any one plugin for an EQ, this would be the one. It sounds amazing even just running audio through it, much less actually EQing which is even more amazing. The lows are very warm and phat, yet it actually improves the solidity of the bass instead of smearing it. Both mids are anywhere from milky warm woodiness to silky warm musical articulation, depending on high high the frequency knob is set. And the treble is super duper smooth and graceful, very musical without over-accentuating individual tones. It can also be used for slight de-essing or to augment a de-esser. My default:
http://ictybtihky.com/harbal/pluginshot ... t_eqp4.gif
URS A-series EQ
http://www.ursplugins.com/ursA.html
Another amazing EQ. This is an emulation of an API (Neve) 550B console EQ. It really excels at bringing out, or pushing back, frequency ranges. But it's especially it great for bringing out high-frequencies. The Q of the bands increases as the gain goes up, so they start out very wide and musical. I almost always use band 3 & 4. And I sometimes use band 1 or 2 if the bass or mid-bass needs to be brought forward or backward. My default:
http://ictybtihky.com/harbal/pluginshot ... lt_api.gif
URS N-series EQ
http://www.ursplugins.com/ursN.html
Yes another amazing EQ. This is an emulation of a Neve 1084 console EQ. If i had to use one word for this EQ, it would be "musical!!!" The low band is the most amazing low shelving EQ I've ever heard. Even at 220Hz, it NEVER sounds muddy, no matter how much gain you give it. The midrange is equally as exciting, and it's soooo musical. If i had to define the coloring of the midrange with one word, I would call it "goodness". It really doesn't add anything to the midrange that isn't already there, it just brings out the good stuff. Generally the "center" frequency of this EQ is more towards the bottom of the actual filter, and it tapers upwards with plenty of beautiful harmonics. And the treble I usually don't use too often, but it is very good at bringing out tones in the upper treble, so I will sometimes use just a twinge of it, often at 12 or 16kHz, whatever is above the main "noteyness" of the treble, to add additional clarity if it's needed. The high-shelf can make some insanely crispy pitchfull treble, even with less than 1db gain, and it can reduce the same just as fast with -1db. So be very careful using the high-shelf, and make sure you have a DA & monitors that will do proper justice to it, or just forget about using the HF band on this plugin. My default:
http://ictybtihky.com/harbal/pluginshot ... t_neve.gif
Voxengo Elephant
http://www.voxengo.com/product/elephant/
This is a broadband brickwall peak & rms limiter. I have done extensive testing of a LOT of brickwall limiters, commercial and free, and Elephant surpasses them all by leaps and bounds, ESPECIALLY with regard to it's lack of any harmonic or inter-modulation distortions at all. Nothing else comes close to how transparent this is. Even Waves L2 looks disgustingly filthy by comparison. It is one of few digital brickwall limiters that can limit L & R separately, and perhaps the only with varying percentages of channel-linking. It also has Bob Katz's K-System metering, and a phase-compensated high-pass filter to eliminate sub-bass. It's best to remove un-needed sub-bass as early as possible in your chain though. My default:
http://ictybtihky.com/harbal/pluginshot ... ephant.gif
Apogee UV22 HR
http://www.apogeedigital.com/products/uv22hr.php
This is considered to be the #1 bit-depth reduction algorithm. I would say that tied with it is POWR 2 which is available on the SADiE, and Daniel Weiss gear, among others... but i've always liked the sound of Apogee UV22 HR more, in every case I've compared the two. So I've always stuck with it, plus I can run it in real-time in my chain without having to route audio back over to the SADiE, keeping the convenience of using Winamp for my final stage. Settings I use are to dither to whatever depth is needed, usually 16int or 24int bit (remember Adapt-X is using 24bit 4 byte float). And I always use the Auto-Black function, which makes it only use as much high-frequency noise bias it needs to.
And then I render the WAV when I'm done tweaking. I also save each chain preset before rendering so I can re-render any version if i ever need to. Then I go into Adobe Audition v1.5 again. If it's 16bit, I'll reduce the bit-depth right away, using NO dithering because it's already been applied. I like to do my fades for 16bit while it's 16bit, because it doesn't add un-needed dithering noise that can sometimes be added to silence that's below the 16bit noise floor before dithering. I usually put 0.14 seconds of a special near-silence on the beginning and end, again using a plugin i made. The reason I don't use pure digital silence is because some crappy cd-players don't enable the DA until there's non-silence and then end up not enabling fast enough to catch the leading edge of the actual audio, which negates the whole point of putting a split-second of silence there in the first place.
And that's about it. For client proofing and final delivery I'll usually provide a 16bit WAV, two high-quality mp3s (320 & vbr), and sometimes 24bit WAV. I use Lame 3.97 with LameDrop front-end, using these settings:
-b 320 -q 0 --lowpass 22.05
and
-V 0 --vbr-new -q 0 --lowpass 22.05
But I'll provide whatever formats they request, of course.
and finally the fully mastered results on that recording...
http://ictybtihky.com/harbal/examples/a ... master.mp3
(slammed? well this actually only has about 2-3db of peak rms limiting, max. There is also a radio promo version with about 5db more headroom/punch and about 5db less "side")
I hope you enjoyed reading this, and I hope you can end up checking out some of the plugins I have mentioned. They all deserve attention, especially Aleksey Vaneev's Voxengo plugins. It's a one-man operation, and his plugins are top quality and very affordable.
later on,
Jesse Graffam
p.s. to zumbido, you should ideally be recording at 44.1 for CD, but if you need to support HD for archival & later re-purposing, I recommend recording at 192kHz if possible. but really, the quality of the output is mostly related to the quality of someone's DA filter, than the sample-rate. At least for PCM anyways, 1-bit streaming formats like SACD are not directly comparable in function.
but if you are forced to record in 48kHz, or need to record in 192kHz, you should re-sample to 44.1 BEFORE doing any of your mastering process. It may also be an improvement to use even a free audio editor on your windows machine for the final "cut & fade" step you are doing in pro-tools. Here's a decent one:
http://www.wavosaur.com/
[edit]
got rid of the SADiE, edited it out of this post too
I also started using Voxengo LF-Max now & then right after the LF-Punch, and I'm demoing the Waves API EQs which DO sound great.
[/edit]
