I've made my first "real" master of a song. It's not optimal, since I've mixed the song myself (wich makes it alot more difficult to "listen with new ears"), but I gave it some time to rest after the mix, before entering the mastering stage.
I'm a newbie at this, and I'm learning. If you have any views regarding the technicalaty here, or have any pointers or whatever, feel free to start a discussion. I made a thread out of this since it is indeed an experiment for me, and I figured some might be interested to learn more about this (just like I am). Best selfwhoring thread ever? :)
- Mastering is the final stage when a product i is prepared for it's final medium. If you intend on playing the song in a nightclub, you will master it one way, if you want it to sound good on low-end radios, you'll have a different approach. If you intend on the audio to be in a flash, you'll optimize for it. There are many aspects to take under concideration when mastering. If the product is going to be an mp3, you must be aware of the fact that mp3 don't have the same dynamic range a 16bit CD quality .wav have. If you make extremley compressed electronic music, for example, it might even be difficult to actually hear any difference between .wav and .mp3, since the processing surley will glue the dynamics to the rough, making it as loud as possible. it's a difficult art to master, but we're all n00bs in the beginning.
- When you've finnished the mix and is satisfied with it, it's mastering time. Yes, you can very well process the stereo bus out in the mix, just make sure you have some headroom. Just because you slapped on a multiband compressor and a limiter on the stereo bus doesn't make it mastered!
When mixing, it is very good if you know a few things about mastering (not a must, but it might help the mastering engineer in the end). For example, if you suspect that the mastering engineer will use a multibandcompressor, then it could be a good idea to isolate certain sounds from the future bands where the multiband compressor will be at.
The most important thing in the mix is hower to make it sound good and leave some headroom!
- With this song, I have exported the mix as a 16 bit 44.1kHz .wav and started anew in Cubase. At school, where I've mastered it, we have quite a few UAD plug-ins, wich I mainly have been using.
First step was to make four copies of the track I then individually filtered them as follows;
Low End - 0-80Hz (in reality, everything below 30Hz is cut, but not completley)
Low Mid - 80-900Hz
High Mid - 900 - 5000Hz
High End - 5000 - 18500Hz
Before doing these cuts I did let the mixx go through a multibandcompressor with dividing at about the same frequencies, but it was a very delicate compression, and in all honesty didn't do much difference.
- I used a Cambridge EQ with a butterworth filter to make the nessecary cuts. The thought was to treat each band individually, and compress them thereafter. Instead of using a multibandcompressor to do the job, I could now use whatever compressor I wanted to do the job!
When cutting like this, it's important to know that the point actually set in many cases (not all cases) actually is the point where the filter is filtering 3 dB. This mean that a 6dB/octave highpass filter set to 200Hz will actually cut 3dB at this point. One octave being the double range of one frequency (100-200Hz is counted as one octave, 500-1000 is one octave etc.), this mean you have to pay attention to what kind of filters you're using to get the same amount of power as the original mix had. There will be a slight difference in the sound due to other reasons, but I won't get into that here.
- I experimented around with adding various effects (I wish I had an exciter...) on the different bands, and ended up distorting the low end (0-80Hz) with a built in Cubase distortion unit (rather crappy, but it got the job done....). After the dist, I set another lowpass filter with the same values as the previous one to keep the bands "intact". What the distortion unit did, was to introduce harmonics of even the lowest frequencies, and it made it heavier and better in my opinion.
- At the end of each band and chain I added a Fairchild compressor. It's a very cpu hungry unit, so I had to use the freeze function when I had set it right. Before freezing the tracks, I used some automation to set the volume of the bands, lowered the volume at some spots to make it seem louder when the song kicks in, for example.
- I took the freeze files (same thing as if I would have exported the different bands) and started yet a new project, with these four files.
I tried a few various things, but nothing really worked out the way I wanted it, except for one thing; A phaser added on the high mid starting at 2 min and 8 sec. It's on until 2.40. After this I once again filpped around with the volumes, automating them to make it right. I tried out a phase-mojo thing to give the highend some sparkle, but it didn't cut it. I also tried to stereo-image the high+mid high, but nothing made it sound better in my opinion. On the stereobus out, I slapped on a neve compressor (wich stole 65% cpu of the UAD-plug in card! But it does the job very good). After this came a limiter, set to a maximum limit of -0.1dB to really squeeze the loudness up. I did automate the limiter, so the threshold differ with 3.5 dB at most, wich makes quite a difference...
- Definitley not. There are many flaws with this experiment. I had problems with the subbass being to loud, and in the final stage (before the neve compressor and the limiter) I actually lowered it with more than 3.5 dB, to be able to squeeze a little more out of the limiter. The lowend STILL is a little overpowering, but I probably should have checked the song in more systems (i checked at home where I have no subwoofer, at the studio and at another computer with decent listening at school). I do think that the final result it too compressed. But electronic music like this is very forgiving. Especially the high-end is too compressed, just listen to how hard the hi-hat hits. This is me setting the attack a little badly somewhere along the way, probably both in the mix and the master. I can live with it though. The mastering process took me around 8 hours during three days. Obviously alot time was spent trying things out.
- The fun thing about making music is to make the songs from skratch (allthough it's extremley timeconsuming). The bassdrum that hits before the song really kicks in is a sound-puff that have been processed mad. When the song hits in, a loop is playing. The "wind puff bassdrum" is highpassed high up (700-800Hz) to only let the midrange of it stay. The "weird" samples are all foley things I recorded around school with a H4-zoom handy recorder (i must buy one...). The synths are all made from skratch using Reason 3.0's Malström Graintable synthesizer. The beat is an old loop of mine. And in that loop there are some toms that really muddies up the mix alot, but I didn't feel like going back to the actual loop and alter it, since I had already started to slice and process it in Cubase. The skratching is made by me, and it was quite sloppy to begin with, and have been edited to fit more slick.
- To make it easier to compare the two I have slapped on a limiter just to raise the volume of the pre-master. The transients may suffer the effects of limiting, but the dynamics and the balance should be rather clear to hear the difference of anyway.