What does it take to create the sound edit and design elements of a feature documentary? Answer: a lot of different samples, plugins, speakers, and most importantly, patience. Here I’ll share a crash-course outline of the process I took to go from Adobe Premiere project to finished sound mix for our 2018 film, Saving Atlantis.
Here’s a quick overview of some of the gear we relied on over the course of the process:
- DAW: Logic Pro X. This pick made sense for our unit when I came on in 2016, when our audio budget was minimal and we liked the flexibility offered by the array of native plugins.
- Audio Interface: UAD Apollo 8 Duo. Using hardware emulations from Universal Audio gave us a substantial boost in production quality, whether I was trying to increase the clarity of an element that was supposed to sound natural, or trying to effect a sound to fit in with a specific subset of uniquely colored sounds. The fact that we were able to offset the DSP onto the hardware of the Apollo allowed this project to stay nimble and responsive, something I would have otherwise expected to see get bogged down with a project of this size. With the addition of a UAD Satellite Quad this project still only uses about 71% of the available DSP with the plugins I used.
- Sound Editor: Izotope RX6 Advanced. We used this to repair audio that had any range of common issues you run into in documentary filmmaking, such as windy plosives, clicky vocals, high noise floors, and noisy recording environments.
- Monitors: Tascam VL-S5’s. This was another early selection to keep costs down; however to their credit they’ve led me to produce reasonably translatable results.
The chronology of the process was a tricky thing to determine early on because there were so lingering unresolved elements of the film edit. I won’t go into the details of how many revisions the film edit saw after “picture lock” (note the ‘TIME SHIFT 1/2/3’ markers in the project window), but generally the film still followed the steps of:
- Sound Design
- Sound Edit
- Sound Mix
The sound design elements were added at any various stage of the process. Some of it was added by the editors for effect early on, some of it when I first stepped in, and some even after the mix was done. This was determined by the revisions that were applied after the first round of screenings was completed, such as footage or music we couldn’t get clearance on. Sound design (or “editing,” depending on who you talk to) is the most challenging part of the process for me since that hasn’t historically been my emphasis and I don’t have a very extensive library of sounds to draw from. You can only get away with using the same WAVES_CRASHING.wav samples so many times. Some SFX were recorded on locations with a Zoom H4n. Many SFX were sourced from open source libraries, where there can be quite a few high-quality samples when you filter search results by sample rate.
When I jumped into the task of approaching the overall edit, I was bouncing back and forth between using Logic and Izotope RX6. Unfortunately, my job required quite a bit of organizational work up front since we were unable to successfully export a functioning .AAF/.OMF file from the project in Adobe Premiere (I still don’t know why, and the time I spent forum-diving was fruitless).
My suggestion to the editors at that point was to organize all the character dialogue onto the same audio tracks, the music onto its own respective tracks, and the SFX onto their own separate tracks as well. This also meant that all of the volume automation and gain changes across separate regions in Premiere had to be reset or manually removed and normalized. These tracks were all individually exported from Premiere as ‘stems’ that I imported into Logic along with an export of the full film in stereo, immediately SMPTE locked and got to work splitting regions and categorizing onto tracks and into track stacks. Color coding regions was essential for this part of the process: visual tastes aside, the utilitarian advantages offered by chromatically indicating which regions contain what kinds of information are indispensable.
I allocated a track for every unique audio source. For the English mix, that meant that every character had a track; if their interviews were recorded in two or three different locations, they had multiple. For the Spanish dubs mix, I created a new track only for every speaker who recorded parts for characters. Our Spanish voice talent would read for multiple characters since a unique reader for each character would have been unreasonable.
The music tracks had a little less reason behind their organization. I started off with two or three to keep overlap separate between tracks that had long fades, and then created new tracks for each song that needed unique processing. For example, some songs needed specific attenuation of sub-bass frequencies while many others sat in the mix better with a wide ~5dB reduction around 1-2kHz. A couple lower-quality music tracks required a more intensive approach, where I used parallel processing techniques and the UAD Little Labs VOG plugin to give them a fuller feel.
The SFX tracks were the most sprawling of all since each sample typically needed customized attention to get it to sit in the mix properly. Most of the time, basic compression and EQ was necessary to get them to live in the same space as everything else, and a subsequent reduction of levels so that they remained present without distracting from the principal elements of the mix.
Overall, this process followed a similar form that most of my audio post work involves when working on pieces for Oregon State Productions, but at a much larger scale. It’s been laborious but highly rewarding to have the opportunity to work on my first feature documentary as the audio lead. I’m sure I’m still making my share of rookie mistakes, but I look forward to learning more with every new project that comes our way.