Audio: Mastering
Adventures in Re-Mastering Music
Introduction
It's 2012 and I decided to release my album digitally on the internet. I selected TuneCore to distribute my album and decided to embark on remastering my CD "Ghosts". I do use audio-editing a lot, mainly SoundForge XP 4.0 by Sonic Foundry (now Sony SoundForge) which came free with my PC SB Live! sound-card from 1999. Since I'm going to remaster, I might as well start using VST plug-ins and learn more about modern mastering.
I found 2 excellent articles on audio mastering from Mastering Elements from TutsPlus and Cubase's Multi-Band Compression by SoundOnSound(both highly recommended).
When I recorded the album in 1999, I took great pains to make sure all the songs were mixed the same so as to define the album as having one sound (like a concept album) . This means that whatever I do for one song, I can also apply to the other songs (yay).
Programs Used
Exact Audio Copy - CD Audio Extractor - Free!
Wavosaur - Audio Editor with VST support - Free!
C3 Multi-Band Compressor by SlimSlowSlider from KVRAudio, VST plug-in - 3 Band Compressor - Free!
UpStereo from QuikQuak, VST plug-in - Stereo Enhancer - Free!
SpAn by Voxengo, VST plug-in - Spectrum Analyser - Free!
Filtrate LE by LiquidSonics, VST plug-in - Parametric Equaliser - Free!
You'll notice that there's a common theme here.
BACKGROUND {non-essential reading}
In 2000, my band called "Fishball" released it music CD entitled "Ghosts". Musically it was fine but there were shortcomings and I spent all my mastering time on corrections using audio-editing software.
• The performance of my vocals had moments of pitch misses.
Since the recording had some effects like flangers and chorus anyway, I use a tiny bit of vibrato to do pitch correction. How it works is, if my vocals were a bit flat, I would zoom into the centre of the flat bit, select about 100ms and apply an "A" shaped pitch change from 100% to 101% and back. This 1% pitch effect sounds a lot like when the flangers and choruses are moving in phase (by coincidence).• The lead-guitar clean sound had clicking.
For recording the lead-guitar sound, I used a Tom Scholz Rockman X100 effects box on setting called "Clean 2". This Clean 2 setting is actually for funky choppy guitar and so had quite treble bias which occasionally would click a bit. The click sound would be fine in funk or grunge but not in my slower songs. To remove the clicks, I would zoom in to find them. When zoomed 1:1, the clicks look like a jagged saw on an otherwise smooth wave. I used "smoothen/enhance" to smoothen the wave and remove the click.• The audio wav-file had tiny segments of silence at random places.
At the time, I mixed-down the songs directly from the Yamaha ProMix-01 mixer via SPDIF out to a PC with Creative SoundBlaster Live! daughter-card SPDIF input. The missing bits of audio could have been either equipment but I suspect it was the SB Live! card. Digital audio was new and PC weren't very fast then. The gaps were perhaps only 20ms in length and I corrected this by copying and pasting from nearby wave data. It's mainly trial and error and you have to do this a few times to get it right. Or you can take the sample just before the gap, paste that, reverse the wave and invert it (a bit like sample looping).• The synth-bass had buzzing in a few spots.
I didn't notice then but one Yamaha ProMix 01 input was faulty and causing this buzzing. I would eventually correct this during the re-mastering process.• For the original mastering, I did some basic gain normalisation, a bit of equalisation and voila! That was my "first cut" mastering process done.
Preparation
Install the programmes! Wavosaur is a surprisingly small programme and runs directly from an exe file. Wavosaur needs to know where your VST plug-ins will be so create a folder (directory) like C:\\Program Files\VstPlugins\ and put all your VSTs there. Then tell Wavosaur to find them by clicking Options \ Editor configuration \ VST path.
A. Transfer from CD
Transferring the audio data from CD to wav file was simple enough using Exact Audio Copy. The set-up lets you choose accuracy or quickness as preferred mode. We want accuracy. The transfer takes maybe 3 ~ 4 times longer than usual... but this is still better than risking bad data and having to transfer all over again.
B. Trimming - 250ms at top & tail
My standard for trimming songs is to keep only 250ms at the start and the end. You can use any audio editor (I used Wavosaur). It's good practice to zoom-in vertically 5 times so you are sure of finding "silence"; and, zoom-in horizontally such that the width of the screen is about 1 second of audio.
The process is very simple: Mark the start [or end] of song; extend your selection into the silence by 250ms; do a "Fade-In" for the start ["Fade-Out" for the end]; and, delete the extremities.
C. Low Freq Cut-Off - at 30Hz
Here we are using Filtrate LE to cut out frequencies lower than 30Hz. These are sub-sonic frequencies which you can't hear. For Filtrate LE, the settings used are:-
• Gain : -20dB• Freq : 30Hz• Q : 0.8• Type : L. ShelfFiltrate LE has 5 bands. We only need to use band #1.
If your music has some rumble effects you want to keep, then perhaps you can cut off frequencies from 25Hz. Honestly, I'd just do 30Hz.
Once, you've finished prepping the audio, save the files (and maybe do a back-up copy).
Notes on using VST PlugIns in Wavosaur
• In the VST toolbar, click "Rack" - This opens up the VST rack for you to load your VST PlugIn.• Select the first empty position "000" and click "Load VST" - Find the VST PlugIn you want, C:\Program Files\VstPlugins\PluginName. The selected VST PlugIn will appear in the rack at position "000".• In the rack, select the VST PlugIn (at position "000") & click "View" - This opens up the interface of that particular VST PlugIn. Adjust parameters to your liking.• When you are ready, in the VST toolbar, click "Apply".MASTERING - using a Multi-Band Compressor
A 3-band compressor is basically a "crossover" which separates audio into low, mid and high frequencies, and then these 3 frequency bands are each fed into their own compressors (ie 3 compressors). So before you start compressing the audio, you have to imagine what you want to do to the music.
Recording - When you're recording individual instruments, you're primarily concerned with capturing the instrument with clarity.
Mixing - When you're mixing the song, you're mainly trying to balance the instruments and giving each instrument some space to shine.
Mastering - When you're mastering a song, you want the whole sound have space to breathe and sparkle.
The 2 crossover frequency points determine what is low, mid and high bands. We want to be sure that the 2 crossover points we pick will capture the essence or soul of the music. Very glossy words, I know.
Crossover Frequencies
D. Spectrum Analyser - finding CrossOver frequencies
The spectrum analyser is the only VST plug-in which doesn't alter the audio in any way. But it's a priceless tool which is going to save many hours of fiddling about.
Since my album has 12 songs, I have a few songs with the drums alone in the intro and a couple of songs with the synth-bass alone in the intro. Let's spectrum-analyse them...
• The synth-bass sound has most of its bass energy around 60Hz ~ 160 Hz.
• The kick drum sound has most of its bass energy around 50Hz ~ 160Hz.
• The snare drum has most of its rattle noise about 4kHz ~ 12kHz.
Basically, the 2 crossover points will define the 3 bands of low, mid and high.
Setting one crossover at 225Hz would safely separate the bass sounds from the mids.
The mids are usually for defining the vocals. Most vocals won't go beyond 5kHz. From the snare-drum spectrum, it looks like 5kHz should be okay for separating out the highs.
So that's the crossover points of 225Hz (for f1 & f2) and 5kHz (for f3 & f4) chosen. C3 Multi-Band Compressor allows you to set f1, f2, f3,& f4 independently but we want our 3 bands to be exactly fitting.
SPAN Spectrum Analyser is very fast, accurate and capable of so much more. The creators, Voxengo, have done a marvelous job.
Compressor Settings
E. Soft Compression Settings - in a Multi-Band Compressor
Before you even start, switch on the limiter. You don't want digital clipping!
Okay, our intent is to remaster, not remix. As such, we want soft or gentle settings which affect the sound in a subtle way; nothing harsh or impactful!
We can focus on the effect of the compression much better if we listen to it one band at a time. Click "bypass" on the bands you are not doing. And remember to be gentle.
• Start the Threshold from maximum and slowly bring it down, until you see the "reduction meter" start to bubble. That's the compressor kicking in. The "reduction meter" bubbling at 10% is subtle. You don't want the "reduction meter" to go more than 25%.
• Set the Ratio at something soft like 1½ : 1 or at most 2 : 1.
• Set the Attack and Release at longish times.
• When you're done, move on the next band (and bypass the others)
• When you've done all three bands, then you can switch "comp on" for all three compressors. That is the magic moment!
The settings I eventually used are as follows:-
Band Threshold Ratio Attack Release KneeHigh -8.4 dB 1.3 : 1 75 ms 752 ms 0.81Mid -10.2 dB 1.3 : 1 50 ms 495 ms 0.74Low -12.0 dB 1.4 : 1 25 ms 247 ms 0.67Listening to the sound one band at a time, some of my compression was hardly audible. Yet switching on all 3 made the sound come alive and with really good definition. The creator of C3 Multi-Band Compressor , Slim Slow Slider, have done amazing work on this software!
Reference EQ
F. Reference Music and equalising using cut/gain
Now that the 3 bands are set-up, we can use the gain/cut of each compressor to equalise the music (as a 3 band equaliser). How should the music sound? Well, listen to some music which you like, unbias your ears a bit, and then adjust the low, mid, high bands to emulate.
You'll notice that different music have different low, mid, high bias. Generally, this is genre dependent. I wanted an 80's new wave sound so I used 3 songs to reference to:-
In my case, the lows and mids fell into place quite obviously using the reference music. The only niggle was how much highs to boost, as all 3 reference songs had differing amounts. Bananarama had the most highs so I referenced to that, and then pulled it down a notch.
The Gain settings I used are:-
• Low Gain : -0.1 dB
• Mid Gain : +1.9 dB
• High Gain : +2.3 dB
Once you're satisfied that this is the mastered sound you want, apply the VST process, and save the files (and maybe do a back-up copy). Once in a while, "applying VST" created a small glitch clip at the start of the file (during the silence), so please check for this. You can fix this by pressing "undo" and then "apply VST" again (and check again).
Added Effects
There are lots more you can do to add to your remastered sound but, quite frankly, the multi-band compressor is 80% of the work done. Although you can create a "chain" of VST plug-in effects, I would recommend that you do the multi-band compression as one single process in isolation. And then treat other further effects as add-ons.
G. Stereo Enhancement
I felt that my album sound could do with a slightly wider stereo field. I'm not sure how UpStereo works but I liked the stereo effect. But, again, I only wanted to widen the stereo by a small amount.
The control "Air" was quite pleasant so I added a tiny bit. I didn't want any "Loudness" or "Bass". I also noticed that "apply VST" would increase the overall amplitude a little, so pressed "undo" and pulled down the "gain" a tiny bit.
The range of settings I use are as follows:-
• Loud : off / Bass : 0• Gain : -1.1 dB• Air : 5% ~ 8%• Width : 63% ~ 67%
H. Tone Correction - using a Multi-Band Compressor
One of my objectives was also to reduce the volume of the snare drum so that the audio mix wasn't so 80's sounding. Applying UpStereo VST to the audio also made the snare drum sound have more noise and less tone.
And so the C3 Multi-Band Compressor makes another appearance...
This time to compress the snare drum sound which will cut the "noise" and bring out the "tone". To do this, I set the crossover points to 4.9kHz and 11.6kHz (see spectrum of snare drum again).
We are only using the mid band compressor so we can set the low band and high band to "bypass".
The range of settings I used are:-
• Threshold : -7.8 dB ~ -8.4 dB• Ratio : 3.3 : 1• Attack : 15.1 ms ~ 20.1 ms• Release : 217.9 ms ~ 297.1 ms• Gain : 0 dB / Limiter : on / Knee : 0.8Basically, the Threshold and Attack will catch the snare hit to trigger the compressor.
The critical factor in this process is the "Release Time" which controls when the noise portion of the snare would be heard. In essence, I wanted to hear the snare drum tone first and then hear the snare noise later.
In the end, I made the decision that the the slower tempo songs and the guitar songs would have more snare drum tone (longer release time). As such, the faster tempo songs and the synth songs would have more snare drum noise (shorter release time).
Conclusion
Well, that's it! Save the files and make a back-up.
I hope you've enjoyed reading about mastering and using VST plug-ins.
The remaster of Fishball's album Ghosts is available online.
Yala