Now, for the most difficult and mysterious phase of completing an audio project.  Mastering is usually left for master engineers who’ve invested a great deal of money and time into creating a room, buying the gear, and developing the ear which is needed to get the job done.  For this step, the spectrum is essential.  Watching how every move affects the curve helps to maximize frequency response and get the levels that sound good.  Loudness is a word that gets used too much in the audio world, and this is especially so when we’re discussing how to master a song.  One might even assume that an emphasis on loudness is vital to the goal of mastering.  This is contraindicative, however.  Loudness also has to be taken out of the equation.  The mix should stand up to commercial masters at any matched volume.


While working on your song, loudness can be measured through your spectral analysis tool and the numbers that will be important are:


RMS – root mean square, (which is a fancy way of saying, the average overall level). This number can lie around -23db if you’re only getting your track ready for someone else to master.  It’s also a good number for extremely dynamic content like orchestral pieces or symphonies.  For most mastering -14db is adequate.  Pushing it into -12 is only necessary in hard rock or EDM, and great care should be taken to not distort the track.  If you have a soft clipper, more loudness can be achieved at less cost than just limiting, but there is still a ceiling to how hard a song can be pushed.


Broadcasting has recently adopted a fair measuring system for overall perceived volume.  It’s called R128 and can be measured using LUFS.  Using an LUFS meter can be helpful when matching the volume of one track to another, and can reveal the difference between a loud overall mix volume and a full and balanced mix.


Another measurement, the peak number, is always supposed to be below zero, and if you’re doing a pre master, then, -6db would be better.  If these figures don’t get you close to commercial tracks, your numbers may need to be calibrated.


Finally, there is only one way to know how much a track is compressed.  That’s with your ears!  You can, however, use the peak to crest numbers to assess how the electrical signal is reacting to the compression.  The smaller the number, the less headroom there will be.  Typically the number hovers around ten or less.


Once you understand the numbers, you come to see that it’s not just about perceived loudness.  In order for your mix to translate from the perfectly sounding monitoring environment in which it was made to the variety of places where it may be played, smoothing out, or compressing the frequency response curve becomes necessary.  This is done more with equalizers and multiband processing than using compression on the whole mix.  Watching the mid and side of the mix while working on the curve can help to bring the balance across the entire pan field so the mix sounds clear through and through.  Using multiple monitors, or even taking the mix to other places, such as the car or mp3 player, is not just a suggestion; but a necessity.  Using various EQ settings on playback devices should help to reveal weaknesses in the master.  Also, putting it in a playlist with some other songs from a similar genre will expose things in a very obvious way.  It takes some experience to understand what fixes will get you where you want to go, but these are tried and true methods for mastering your song.  An ideal curve will sit right no matter how they choose to listen.  All things have to be considered and fixed before the master can be printed.  Printing is where you establish a line across the top of your loudest parts of the song through limiting, compression in the bands you don’t want distorted, and perhaps some analog or tape saturation to push things into the red without getting horrible results.  If you have a hardware soft clipper, use it on an extended play (EP), to get that extra sliver of fractional db gain off the whole CD.  Just don’t destroy the loudest track while doing so.  The end result is a wave that looks flattened or chopped of at its loudest parts, and if done correctly, should not be noticeable to the listener.