The Art of Delivery —
Preparing Audio Files for Mastering

Among the most common questions I receive from musicians, proper preparation of audio files for mastering is always at the top of the list. Following are my opinions relating to the three most common topics: volume, resolution and editing.

Maybe the most common mistake the well-meaning musician makes is an attempt to get his/her mixes as loud as possible. As far as I can tell, this is because most people don't know whether this issue is addressed in mastering (FYI, it is). Many artists — believing it's better to be safe than sorry — do one of two things: normalize their tracks or send them through a compressor/ limiter (or some sort of "finalizing" plug-in). Unfortunately, both degrade sonic quality and neither delivers the proper desired result.

Contrary to common belief, normalizing does not make tracks as loud as possible. The normalization process scans a digital audio file and looks for its peak volume, then moves that point up to digital zero. All other sounds in the file are adjusted proportionally. So, if the file's loudest point is a snare hit that registers one decibel below digital zero (-1 dBFS), normalizing will make that track one decibel louder. (As a point of reference, a decibel is commonly considered the smallest measure of change that the average listener can detect.) This is hardly what most users of this function are trying to achieve. And not only does normalizing generally not make tracks substantially louder, it does nothing to address the average volume of songs: a situation that is crucial for the proper aural perception and flow of a good CD.

The second scenario finds musicians adding a "mastering" plug-in to the mix-down process. I receive a lot of files that have been treated this way and it creates two problems. For starters, it can make the files so hot that any filter applied in mastering (equalizer, compressor, limiter, etc.) will overload and distort. Secondly, the effects of the process cannot be removed, so I have to pull every trick in the book in an attempt to counteract the plug-in's deleterious effects. In this situation, mastering is turned into an audio salvage effort as opposed to a fine-tuning improvement process.

It may be obvious, but the highest resolution is always the best way to deliver digital audio files for mastering. So if you have a hard disk recording setup that offers 20, 24 or 32bit — as well as 48, 88.2, 96 or 192kHz — capability, it's to your advantage to utilize it. Why? Because while it's true that your audio files will eventually find their way to 16bit/44.1kHz for the CDs commercial release, the work that's done on the files before that point will be much more transparent sounding when processed at a higher density and resolution.

Sound simple? Maybe, maybe not. Here's a scenario: Say you have a setup with a built-in CD burner. If you record and mix at 24bit/48kHz and then burn an audio CD for listening in your home or car stereo, you've just lost all your high-resolution information. This is because to burn a commercially compatible audio disc, your recording application (or stand-alone unit) must reduce the bit rate and resolution. And unless you have very specific (and sophisticated) applications – as well as a processor with lots of CPU horsepower running them – this will compromise the quality of your audio. So while it's OK to burn audio CDs to reference your work, always make sure that what you deliver for mastering is the same resolution as what you've recorded and mixed to.

Another common situation comes from musicians saying they've already done intro edits and outgoing fades, and the songs only need EQ and level treatment. Contrary to common lore, this does not make the mastering engineer's job any easier, and in some ways it can make specific tasks nearly impossible. Consider a song needing the removal of electronic hum or microphone hiss. To do this successfully, a bit of that noise must be taken from a part in the song where there is no music, so as to effectively isolate a digital "fingerprint" and create a filter. Ironically, the best place to take this fingerprint is the few seconds before or after a song. If that space has been removed, it is much more difficult to get a good sampling of the problem. If an exact fade or edit point is desired, the best thing to do is carefully document these requirements (mm:ss) on a song-by-song basis for the mastering engineer. Accuracy can be confirmed in the reference disc.

The next time you're working on a project, keep these concepts in mind. They will give your mastering engineer the necessary flexibility to bring out the best in your music and allow you to create a more professional-sounding product.

© Paul Abbott. An edited version of this article was published in EQ Magazine.

Quick Overview

» Don't add any plugin or processor to a mix just to make it louder. You'll want headroom with peaks around -3dBFS.

» Make sure your final two-track mixdown bit/ sample rate is the same as your multi-track files...not truncated. 24/48 is a good standard for tracking and mixing that just about all hardware supports.

» Leave "heads" and "tails" in case noise removal is required.

Questions? Ask us.

Bookmark and Share