How We Edit Podcasts (Patreon)

Hey Friends,

The most recent episode of The White Vault (4.07) was particularly complex, and we wanted to offer some insights into how we do things in post production. This will contain mild episode spoilers.

[Production - A directed recording session with Daniel Muñoz, who voiced one of the monsters]

A real quick outline of what production is:Pre Production is brainstorming, writing a script, editing the script, language translations, casting, signing contracts (so we can use the actors' voices), setting up everything before you record, etc.Production is the recording process, which for us typically has actors around the world reading their lines at home, at their own speed, self-directed by the script.Post Production is combining the raw audio files (Mp3's and WAV's) from the actors into the Mp3 you listen to, editing them in a DAW (Digital Audio Workspace or 'editing software'), and sending the finished Mp3 to your phone or computer. All of the editing takes place in POST Production, and that's the focus of this writeup.

[We communicate via Word document, using comments and colors]

Step 1. Dialogue Cut

[The finished dialogue cut by Marisa in Protools]First I submit the list of who is in each episode, the episode script, and raw actor audio to Marisa Ewing (Hemlock Creek Productions), who marks up the script in Microsoft Word and makes the initial dialogue cut. This averages around 3-9 hours per episode, depending on the complexity. She also cleans up the audio using some fairly advanced tools in her software of choice: Protools. For instance, the volume of each track is made comparable to the other tracks (normalization), and the best takes are selected. There is a true art to dialogue editing and Marisa has given lectures on the topic, so I'll keep this part fairly brief. Marisa also makes an initial list of "retakes" we'll need and sends us the edited files as "Stems", which we'll explain at the end. For now, just understand that the Stem files are audio files that are in sync with one another (so they line up when imported in any editing software) and that they're organized by character, and renamed for easier organization.

From there, I import the edited voice actor's files (Stems) into my editing software (Mixcraft), where they line up really easily, and divide them into scenes. The gaps I made between the files you see below are the scenes within episode 4.07. Each horizontal line/lane is a "track" which represents a character voice, sound effect, music track, etc.

I organize tracks with prefixes and colors, which greatly assists in visually identifying what I'm editing at a glance. For example CHA = Character (green), AMB = Ambience (purple), SFX = Sound Effect (orange or yellow if it's clothing Foley). This is very helpful due to the complexity of our editing session, which has the entirety of The White Vault (S1-4) loaded at all times. Organizing the session, and having the entire show up at once saves a lot of repetitive work and makes editing a lot easier!  I go through and erase all of the silence manually (most of the tracks are 70% silence), further tweaking track EQ settings, reverbs, and swapping out performance takes as needed. I typically work with some of the location sounds (ambiences) present while I edit, so I get a feel for how the characters will sound "in the room." Remember, none of our actors recorded with any of the other actors, they just have our scripts for direction. So I spend about 2-5+ hours adjusting the timing of the performances, swapping out words for understandability for the less English-savvy actors, and making my own list of retakes, after confirming that there isn't a better take in the original tracks. When it's needed, I also fix things like manually removing their room noises using manual fades at the beginning and end of every breath... For instance, episodes four through eight a certain actor recorded near a heater, which Marisa reduced the sound of, and I removed the final traces of. I email actors for retakes if they're absolutely needed, and edit them into place as I receive them. I also start to make notes directly in my editing software, which you see as blue lines with text above them. These remind me about sound effect placement, panning, volume adjustments, or anything else I might forget.

[A clear visual representation of three scenes once the dialogue cut is almost complete]

Step 2. Sound Design

[Above are notes for an 8 minute scene from episode 4.06]

Sound design varies from scene to scene, but generally I start by adding footsteps, swapping in any retakes from actors I've received, and any other major sound effects that influence timing. By this point I've noted every sound effect we'll need via the blue lines, and I spend a fair deal of time making sure that nothing gets missed, including sound effects that characters talk about but that aren't expressly written in the script (like someone saying "give me your hand"). Establishing the timing of each scene is important before we record Foley (manually recorded sound effects). Foley is typically the last step in what I do because it requires accurate timing, and you can see the finished 'clothing' Foley visible as the yellow tracks in the below scene/screenshot where Eva and Rai have a conversation in the same area as Simon. I'll typically record at least 2 clothing/motion Foley tracks in a scene with 2 more or characters, in addition to 1-3+ tracks of Foley for non-clothing sounds as necessary.

[I play with jackets, vests, and bags for maybe 2 hours a week]

After the sounds are all recorded, I do another pass of each scene to double check that I've established the proper PoV for the scene. This is where I ask "does it feel like we the listener are in the right part of the room?" I'll adjust the volume levels of every character and sound effect to match the desired PoV. 

If needed, I'll add additional sound effects and Foley, and often times adjust vocal effects for monsters to make them sound even more scary. Prior to this point, the most terrifying of creature vocalizations will probably sound weak in the context of the scene. This is also where (project pending) I'll manually pan the tracks (left/right) in certain areas. Panning is a really important and underutilized tool in podcasts. Small changes (5%) yield significant results and that's the difference that made this episode work so well. In episode 4.07 we have one scene play identically from 2 different points of view - the protagonists and the monsters stalking them. The panning, volume, and reverb adjustments really make the second listen-through a lot more unsettling, allowing the listener to walk with the monsters, stalk the protagonists, creep up from behind obstacles, and attack them. The same tricks are applied feel the PoV fall during the shotgun blast, to hear the doctor slowly stumble closer, pick up the camera, and walk away with it, or, in VAST Horizon, to help us "feel" the spatial location of the detached AI -  Ally as she projects her voice from various points around the Bifrost.

Step 3. Mixing and MasteringThe tracks are exported as Stems (see below), Zipped (WinZip or WinRAR), and dropboxed to our mixing engineer, Brandon Strader, who imports them into his DAW (mixing software - Digital Audio Workstation) and works his magic. I send him copious notes on reverb, panning, and what tracks go to which character using timestamps, scene by scene. About 8 hours of work later, Strader sends us a single Mp3 of the first cut. In this instance, it's all 56 tracks that make up episode 4.07 combined. I'll send him a .txt list of further adjustments and typically 1-3 mixes later, we're at the final mix! Keep in mind that I listen to each mix 2-5 times from end to end to ensure that there were no exporting errors, and that by changing one thing we did not create new issues.

[Strader and I use .txt files to share scene notes]Stems explained:Every track (horizontal lane) in the editing software is a voice or sound effect type. For Episode 4.07 (which has a total running time of 28 minutes), each track is exported as a single 28 minute WAV file with the same start/stop as the other 56 tracks. There is a LOT of silence on every track, but it's needed so they can all be combined to recreate the timing we have on one computer/DAW to another easily. It's useful because we don't use the same editing software, but all audio editing software can read WAV files, and all of the tracks have the same start/stop point. If we were all using Protools we wouldn't have to do this, but Strader and I are attached to our respective editing setups.

Finally, we upload the episode for you to listen to on Patreon, and schedule the post for our release date. We hope that you enjoy this insight into some of the processes we go through to bring each episode to life!