Another of my main roles towards the production of 5150 was to mix the dialogue for the entire film. This has been an incredibly challenging element as EQ and compression had to be automated throughout the process. Tom Fleischman who has mixed films such as: Goodfellas, The Silence of the Lambs, Cape Fear & Miami Blues describes dialogue mixing as: “the most complex, the most time consuming, and the most difficult part of mixing for film” (cited in LoBrutto, 1994, p.77) and goes on to say:

Each scene is shot from several different angles, so the sound tracks that accompany those different pieces of picture have different qualities to them. When they’re spliced together, very often it’s bumpy and you’ll hear the sound change on every cut. That all has to be smoothed out and equalized so that the voices all sound like they’re in the same place. Just getting all of that together is a very painstaking process.

What Fleischman describes is part of the painstaking process I endured for 5150. When we were provided the VT recorded during audio project 1, we discovered that the crew had decided to capture the audio themselves with little training and using poor quality equipment with bad preamps such as the Marantz portable recorder. On top of that, the location was a manor house with high room resonances reflecting off the wooden floorboards and furniture. Therefore as Fleischman describes, I had to arrange an ADR session for the unusable lines and then use EQ to smooth over the different microphones, rooms, angles etc. 

I found the easiest way to make the performances believable was to assess each line by ear and begin removing imposing resonances using subtractive EQ. However, when purveying for tips, I came across an audio for film blog suggesting the following for dealing with distance recordings:

First apply a Parametric EQ at 300Hz with -4db and a Q=2. This can remove some of the “room boominess”. The second Parametric EQ at 4kHz with +6db and a Q=.25 will brighten up the dialog. Even a slight improvement will make the entire mix sound better. Also…I like to EQ dialog while the music, background sound and room tone is playing. If you solo the dialog and EQ in a vacuum…you won’t hear how it’s interacting with the environment and the mix (Nedomansky, 2014).

Of course, this isn’t a copy-and-paste-settings scenario, but they can certainly be applied as a rough guideline. For the character, Ursula in 5150 I applied the method after sweeping the spectrum and pulling down the room with a wide Q; notching out the whistles, removing the honky frequencies around 500Hz and high/low passing any unwanted frequencies such as low rumble and sizzle distortion; the final EQ curve looked like this (minus the high/low pass that was actioned on a separate plugin for proximity adjustment automation):

The full dialogue chain consisted of the following of which set the framework towards all dialogue processing:

• DeEsser:- tuned to each performers *s* range.
• High / Low Pass Filter:- automated according to the distance of the actor/actress.
• Subtractive EQ (before the compressor to avoid it reacting to removed frequencies.
• Compression to smooth transients, sporadic dynamics and to artistically draw out the personality of the performer.
• Expansion: follows the compressor as the compressor raises the noise floor. Expansion pushes it back down again retaining the voice.
• Additive EQ: used to solidify the fundamental of the voice; add midrange for listening universality, boost the presence region (4-7k) to make the consonants more intelligible.

The following is an example of the problem line delivered by the character Ursula from the original VT recorded by the FTV crew:

I considered the line to be too problematic to use and was originally scheduled for ADR, however due to actress availability I had to do the best I could. This is the version following the entire chain: