Mix-master flow
It’s possible for performing artists to tackle mixing and mastering their recordings nowadays. There’s a knack to it, for sure, but with some patience the process isn’t too grueling, and the benefit is a fine final product, as nobody has more incentive to patch up a recording better than the artist themself.
I use Audacity as my DAW, because it’s free and I like open-source software, generally. It’s used by a good deal of the audio-book readers, and podcasters alike. Who knows if the tool is powerful enough to put oodles of tracks together, but for one instrument, with only a pair of input mics to deal with, Audacity is up to it, for sure.
Source
I try to ensure that each movement I record has at minimum two takes. It’s a hard rule for me as patching up a flub when there’s no backup is not fun to do.
I tackle each movement in turn, listening to each take to get a sense what version is the cleanest, most creative version I can use as a master track. When I find a good candidate, I name the track ‘master’ and put that in my working directory.
Dirs
Using all my source files, I splice together a version and put that ugly hairy monster into a directory named ‘attach.’ I rename it to its apt track number — e.g. ‘seven’ — and remove the master copy.
From there, I work on the next phase of editing the track. I copy ‘seven,’ rename it ‘master,’ and begin smoothing out the roughness in my main working directory (more on what tools I use to smooth it out, below).
When this editing phase, is done, I copy ‘seven’ over to a directory named ‘blend,’ and blow away the ‘master’ file as before. Now I have incremental backups of the track over the course of its maturity.
In the final phase of editing, the gain and bookends of the track need to be dealt with. This is a pretty easy step, so I don’t create another Audacity project just for that. Rather, I just output the final product (a 32-bit ‘.wav’ file) to a directory named ‘crown.’
In depth
When I’m in the first ‘attaching’ phase, I practically press the ‘z’ key every second. I’m constantly making sure the cuts and pastes don’t create any pops, and ‘z’ mitigates that for sure.
When I’m ‘blending,’ I live and die by the following ‘Effects’:
- crossfade clips
- sliding time/pitch shift
- spectral edit multi tool
- spectral edit parametric eq
- equalization
In addition, the ‘Analyze → plot spectrum’ is helpful, and outside Audacity, the following website has the most useful tool to pair up with most of the above:
When it comes time for the final phase of putting it all together, some rules of thumbs and hacky tricks are all that I use. More on that right now!
Mastering classical
Although rules are meant to be broken, I don’t use limiting, compression, nor do I perform any wholesale noise reduction. The former messes with the great dynamic range of classical music, while the later causes audio artefacts to show up.
However, for cutting an audio CD, approaching 0dB at the peaks is desirable, though I have a system for doing it more in keeping with what one might expect from a classical audio product.
A somewhat arbitrary peak-level I’ve decided to gain to is −6dB. I like the number six, and it’s a decent middle way considering rock albums peak at zero, and quieter solo instruments of which violin is not, can be found in the wild peaking somewhere around −8dB.
My goal is to have the loudest moment(s) in my whole album to peak at −6dB, so I fish through the louder tracks to seek out which one is loudest. I do this in a slight hacky way.
For the candidate tracks, I toggle on Audacity’s ‘View → show clipping.’ I select the stereo track, and apply ‘Effect → amplify.’ Within the amplify dialog, I set ‘new peak amplitude’ to ‘0.1,’ and toggle ‘allow clipping’ to ‘on.’
The track now shows a single point in the red. I zoom in, take a close look at the note/chord in question, then highlight the area for an eventual ‘copy.’ (One can also ‘Edit → save region’ for good measure.)
I reverse the amplification edit on the track, copy the note, and paste it into a new Audacity project file. From there, we can discover the amount of gain the whole album needs in order to just reach −6dB.
In the new single-note file I make, I select the stereo track and engage with ‘Effect → amplify’ once more. This time, I toggle off ‘allow clipping’ and in the field labeled ‘new peak amplitude,’ I set the value to ‘−6.0’ and jot down the yielded value from the ‘amplification’ field.
With these values, one can suss out which track is the loudest (by what track yields the smallest number) and it also gives the exact dB number I must apply to every single track such that the loudest moment in the album is only −6dB.
Bookends
I’m not sold on how much ‘quiet’ before and at the end of each track there should be, generally. I’ve been finding that at the start of the track, two to four seconds of quiet is a decent window. At the end, due to some animated feedback from my first self-titled album, I now leave a huge eight to ten seconds window of quiet. This much quiet time is rarely found in the track itself, so I have to copy/paste/munge together (always using ‘z’) and replicate this quietness from the naturally occuring ‘quiet’ there to work with.
I’ve chosen to fade in/out from the ‘quiet’ with a one-second ‘Effect → fade in’ for the track’s start, and ‘Effect → fade out’ for the end. Again, there’s no reasoning behind one-second, it’s just a simple number that works.