Loud, but not TOO Loud!

Clients always want a mix with impact. In their minds, impact equals loud, but that’s not really true. What you really want is a dynamic mix that fits within a comfortable range. The listener should hear some variance without the broadcasters and/or streaming platforms affecting the mix too badly through loudness normalization. Hence, the battle over loudness and in recent times, the various ways to measure it.

In the early 2000s the so-called loudness wars came to the attention of legislators who, through the FCC, eventually codified US standards with the CALM Act of 2010. This resulted in loudness targets measured in LUFS (loudness units full scale). It has become the standard for broadcast mixers ever since. However, years before the CALM Act, noted mastering engineer Bob Katz offered a solution in a white paper to the AES. His proposal was the K-scale, which is still a valid way to mix and is an available feature in a number of metering plug-ins. This method is known to recording engineers, but probably something that most video editors never knew existed.

The K-scale is based on the concept that mixing should always be done at a reference speaker volume. Based on research done by Tomlinson Holman (the TH in Lucasfilm THX) and Dolby for theater sound calibration, that level is 85 dB SPL (sound pressure level) in movie theaters. The K-scale weights a VU meter’s scale according to three levels of headroom. These are labelled as K-12, K-14, and K-20 for 12, 14, and 20 dB of headroom, respectively. K-20 is intended for mixes with a wide dynamic range like classical music, K-14 is good for pop music, and K-12 for broadcast. Of course, those genres are only suggestions.

The point is to preserve dynamic range in the mix based on the type of material. The key to the weighting of the scale is that 0 VU always equates to a speaker volume of 85 dB SPL. It’s intended for professional music mixing and mastering studios. However, don’t discount its value for modern video post. K-scale metering has two advantages. First, you are mixing with a VU style meter, which has an averaged response. That is more akin to how humans perceive loudness than fast-responding full scale meters common to most NLEs and DAWs. Second, it encourages consistent monitoring levels when you mix audio.

How do you implement this in a small edit suite?

First, set up the room properly if possible – depending on available space. Place your speakers at the front. Your seating location should be about 1/3 of the way from the front with 2/3 of the room space behind you. Sound absorption panels are highly recommended. The speakers should be about five feet from you. Calibrate your speaker volume using a sound pressure level meter. The simplest and cheapest option is a phone app. I use the free NIOSH SLM app on my iPhone set to C weighing. It’s fine for casual monitoring like this, but understand that a phone app is not accurate in critical environments due to how the built-in mic is calibrated.

There are several meter plug-ins that feature K-scale presets. I prefer the free mvMeter2 from TBProAudio, which emulates an analog VU meter. It includes numerous VU and PPM presets along with the three K-scale settings. Place this on your final master or mix bus without any other effects or level changes after the meter. Play a pink noise file in your timeline and set the output level so that it reads 0 VU on the meter scale that you’ve picked – for instance, 0 VU on the K-14 scale.

There are plenty of home theater guides about how far away you should be when measuring SPL. The official distance would be one meter from the speaker, but it’s also valid to measure from where you would normally sit. A comfortable listening volume is going to be somewhere between 70 and 85 dB SPL. Start at 85dB. Test your mix playback. If this sounds too loud, you can lower the level and retest with the pink noise file. You might not operate at that level the entire time that you are editing. However, keeping this consistent level each time you mix will give you more predictable and translatable results.

Using this system to mix

Now, let’s put this into practice. Most of the mixes I do myself are for broadcast TV or some social media outlet, like YouTube. Each uses different loudness targets. I find that the K-12 or K-14 setting is best for my internet work and K-20 is best for broadcast. Note that Katz recommended the K-12 scale for broadcast; however, that was before the industry standardized at -23 or -24 LUFS. I find that if you use the K-20 scale (with 20 dB of headroom) that the result is closer to this spec. That difference is probably because Katz was talking about music broadcast, as in radio airplay of songs, rather than TV programs.

The reading on the meter will react like a typical, analog-style VU meter. If you’ve mixed with VU meters in the past, then you’re probably used to the needle hovering around 0 VU and safely (and often) bouncing up into the red overload zone up to +3 dB or more. When using the K-scale method, you’ll find that a lower meter reading will actually match what sounds right to you, assuming you followed the speaker calibration described above. Most of the time my levels are somewhere in the middle of the VU meter, well under 0 VU. It only bounces above 0 on a few peaks. If you compare this to Premiere Pro’s loudness meter, you’ll find that this equates to around -20 to -14 LUFS, which is a good target for platforms like YouTube.

Remember that a VU-style meter is an average. Be sure to limit your signal for peaks. A limiter plug-in should be placed before the meter. In my work that’s usually set to a brick wall level of -3 or -6 dB true peak depending on the program. This is lower than the -1 dB limit often recommended for music mixes. This approach to mixing should give you a decent amount of dynamic range without the need for severe normalization by the platforms.

It’s important to understand that many of these recommendations for streaming service targets are based on music mixing and not TV shows and films, where dialogue is usually dominant. In 2021 the AES posted an updated technical document (linked here) that spells out recommendations for dialogue-based content. Their suggested target is -18 LUFS based on measuring the dialogue tracks rather than the full mix.

Alternatives

The K-scale system is old, though still valid. Of course, NLEs like Premiere Pro and Resolve include their own loudness meters and there are numerous third-party plug-ins using traditional and modernized ways to measure loudness. One modern metering technique is to measure the difference between peaks and short term loudness, aka PSR, and/or peaks versus long-term loudness, aka PLR. A plug-in like Meterplugs’ Dynameter provides you with this feedback. The reading will tell you how the mix level will be handled by the loudness management of the top streamers. Loudness Penalty is an online resource that will tell you this, as well.

Lastly, what about headphone mixing? There are pros and cons, but in order to trust your ears and know whether your mix will translate to various speaker systems, Waves has a solution. Their Nx technology emulates the three-dimensional sound and space of several famed studio control rooms, including those of Chris Lord-Alge, Ocean Way, and Abbey Road.

Place the plug-in last in your mix bus, select the correct headphone or ear bud model EQ curve (270 presets are available), and you’ll experience your mix based on an emulation of those speaker systems and rooms. Each plug-in includes head tracking and Waves sells a separate Nx tracking sensor. However, it can also use your front-facing computer camera.

With the plug-in enabled, you’ll hear a proper sound image as if listening to those speakers (Abbey Road, Ocean Way, etc.) instead of the usual left/right split of headphones. With head tracking, the spatial image shifts as you move your head. If you disable head tracking, you can still manually pan the image around the room and hear the result in your headphones. Since the plug-in is only designed for headphone monitoring, disable it when listening on external speakers and when you output the mix. Since the plug-in is manipulating phase to create these emulations, material that is truly just mono might sound more odd than full stereo mixes.

Granted, you’ll probably never own the kind of speaker systems these studios use, but the real point is that these plug-ins give you a transportable reference based on some of the best control room environments. As you move between different systems with different monitor set-ups, you can maintain a common monitoring reference for all of your mixes.

Happy mixing!

©2023 Oliver Peters

NLE Tips – Audio Track Mixing in Final Cut Pro

In the past I’ve explained how audio in routed through the Final Cut Pro architecture. I’ve also discussed track-based audio mixing, predominantly based on the workflow in Premiere Pro. Today I’d like to extend that workflow into the realm of Final Cut Pro.

Everyone knows that FCP is not track-based. The timeline consists of a string of audio/video clips called the primary storyline, which empowers its magnetic feature. Additional audio and video clips can be attached to the clips on the primary storyline as connected clips – video above, audio below. At this level the software is indeed trackless. (Click on any image to see an enlarged view.)

Understanding audio roles and lanes

Several years ago, Apple added the “roles” feature. Audio and video clips can be assigned default and/or custom role designations, which can be used for visual organization and other functions. For example, do you want to export a “textless” ProRes file from your timeline? Then simply disable the Titles video role in the export dialogue.

Apple engineers have done more with audio roles, which can be further grouped into audio “lanes” through the timeline index window. If you’ve assigned the correct audio roles to each clip, then all dialogue clips are grouped into the dialogue lane, all music clips in the music lane, and so on. If you exported an FCPXML file for an outside mixer, then audio roles help to organize the track layout in other audio software.

At this point the clips are still individual. However, once you combine all clips in the sequence into a single compound clip, then the audio for all clips within an audio lane are summed together. This is similar to a group or submix bus in a DAW. The combination of lanes are in turn summed together and sent to the mix output. In essence, each audio lane within the compound clip is similar to a summing track stack in Logic Pro. You can adjust volume and apply effects to the entire lane, on top of anything done to individual clips contained inside of that lane.

Mixing in FCP on real-world projects

I’m working on an Alaska travelogue series – on-camera host on location, voice-overs, voice-over pick-ups, and music. The host stand-ups were recorded in two environments – close to the shoreline and in a quiet wooded area.

The location sound mixer recorded both a Lavaliere mic and a boom mic on separate channels. My personal preference is the boom, but sometimes the waves on the beach created too much background noise. In those cases, it’s the lav mic, but then I have to contend with the duller sound of the mic under the clothing, along with some rustle.

The next challenge is getting the voice-overs to sound close to the on-camera audio. These were recorded on location, but in a quiet room. The final challenge is to match the sonic quality of the voice-over pick-ups (done by the host at his home) to the original voice-overs.

Step One

The first step in this process is to assign the proper audio roles before clips are edited into the FCP sequence. Roles are quite versatile. If you had multiple speakers, each one could be assigned a separate role. In this project, my audio roles are Dialogue, VO, VO2, and Music. Once clips are imported and roles assigned, I can edit as I normally would in Final Cut. I personally add very few audio effects at this point to the individual clips, because I will do that later. In addition, certain effects, like noise reduction simply don’t work very well with short clips (more on that in a minute). So I only add what I need to “sell” the cut.

Step Two

Once the cut is approved and locked, I can move on to a final mix. To start, I’ll remove any audio effects that I’ve added to individual clips. Then, I meticulously go through and even out any level imbalances. Final Cut Pro features multiple gain stages. You have the clip volume control, but if you expand the audio, you see the individual channels, which each have volume controls, as well. Each of these can be raised by up to 12dB. So if you’ve applied 12dB to the clip and it’s still too quiet, expand the audio and bump up the channel volume. Or work this process in reverse. My objective is to end up with a clip volume that’s a bit hot in the peaks and then use the range tool to highlight the larger peaks and duck them down a bit.

Expand audio and make sure you have overlaps with fade handles between all clips. This is somewhat time-consuming. It’s far simpler in Premiere Pro to add audio dissolves (crossfades) across all audio edits in the timeline in a single step. But it’s a necessary step, including the addition or room tone/ambience to fill any gaps in the speech.

Finally, check the music. Make sure the edits work musically. Overall, the music volume can be a bit loud at this stage, but you want to make sure the balance is right for the entire sequence. So pay attention to the proper and graceful ducking of music around spoken audio.

Step Three

After you’ve made everything as uniform as possible, compound the sequence. Open the timeline index, enable “show audio lanes” which expands the audio of the compound. You’ll now see a “track” or summing bus for each audio role – Dialogue, VO, VO2, and Music. When you select an audio lane, you can adjust its volume and apply audio effects to only that lane. That lane’s audio parameters are shown in the inspector pane.

Selecting the topmost level of the clip, displays the output (i..e mix) bus parameters. Additional effects can be added here. It’s fine to apply and adjust such “master” effects, but I recommend that you do not make any changes to the volume. That’s because the volume control comes after any effects, which would include a meter plug-in, such as the built-in multimeter plug-in. Leave the volume slider alone if you want to see accurate volume levels.

Aside from mixing in tracks/busses, audio roles add another value at the time of export. My deliverables include a ProRes file without titles, as well as audio that’s split into separate tracks. In Final Cut Pro’s export setting, I can select the Multitrack Quicktime and then arrange the combination and order of roles. For this project, it’s a ProRes file with four stereo tracks corresponding to the four roles that I’m working with.

Note that when you export a multitrack file, each lane output also has any master output effects added to them. For example, if your mix uses a compressor and a limiter on the main output of the compound clip, then each lane/bus/track of the multitrack will now also have the added effect of that compression and limiting. If you don’t want this, then make sure to disable these effects prior to exporting a Multitrack Quicktime file.

Which effects should you use?

I’ve now discussed how the process works, but what combination of effects should you be using? Obviously that’s a question of style and personal taste. The type of effects for me will be similar to my description in the Premiere Pro article. I tend to stick with native Final Cut Pro effects, so that I don’t have to worry about what’s installed if I move to another Mac or a different editor has to step in. Also, Final Cut Pro is often a poor host for some third party audio plug-ins. I don’t know the reason, but have been told it’s up to those developers to optimize their tools for FCP. In most cases these same plug-ins work well in Logic Pro, not to mention other non-Apple applications. Go figure!

I’m happy with most of the built-in Apple audio plug-ins, with the exception of noise reduction and other audio repair tasks. The Accusonus tools are my go-to, but they are sadly no longer available. After that it’s the RX package from iZotope. If you have a really challenging piece of audio, then use the standalone RX package on that clip and re-import. If you don’t own either of these, then the newly added voice isolation feature in Resolve is pretty sweet (and better than what’s in FCP). Another impressive contender is Adobe’s Podcast beta. The AI-powered voice enhancement feature is available for free use through their web portal. I’ve used it for some really poor Zoom interview audio and it did an outstanding job of cleaning up all manner of audio defects.

Where this explanation is most pertinent is on location-based dialogue recordings. These are the ones that often benefit from noise removal/repair. These tools require consistency and some lead-in to the first audio, so they are best applied to full tracks and not individual clips. That’s why I make sure I have overlaps and fill in gaps and do all of this processing on the lanes of the compound and not on individual clips. If you have different dialogue sections – some noisy and some clean – then it’s best to organize these into separate audio roles, so that they are sorted out correctly once you compound the clip.

My typical processing chain

My FCP effects layout is similar to the description in the Premiere Pro post. Dialogue and VO tracks get some noise reduction, EQ, and compression. Voice-overs are particularly susceptible to plosives (popping “p” consonants) and sibilance, so plosive and de-essing filters are useful. For music, I usually spread the stereo image more and dip the EQ in the midrange. Plus some compression. All of this is designed to allow the dialogue to sit better in the mix. 

The last level of processing is what you do to the top level of the compound clip itself. That’s a bit like mastering in audio production. Applying effects to the compound clips is analogous to applying effects to a mix or output bus in the DAW world. On this particular chain, it’s EQ, exciter, compressor, adaptive limiter, and the multimeter. The effects stack is processed before the volume slider. Since I’m judging peak and loudness levels with the multimeter plug-in, I don’t want to make any volume slider changes on the compound clip, because those would be applied after the reading on the multimeter.

You’ll notice from my screen grabs that different compressor models have been used. These are all from the same Logic Pro compressor in FCP. This single plug-in features various presets designed to emulate tried-and-true analog compressors favored by top recording engineers/mixers.

Final thoughts 

As with my other Final Cut Pro audio articles and posts, I can already hear some screaming that this is just a workaround for the fact that Final Cut Pro has no “true” audio mixing panel. While that may be true, it’s also irrelevant. Until such time as Apple’s ProApps engineers redesign the audio section or add a “roles-based mixer” to the tool set, this is the software you have. If you want to mix in Final Cut Pro and deliver a properly mixed master file without using specialized audio software, then it’s best to understand how to achieve the required results.

If you step into the compound clip to make any editorial changes to the sequence or to individual clips, then you will not hear the results of the top-level mixing and effects. The proper mix is only heard when you step back out. This is a short-coming compared with this same process in Premiere Pro. Therefore, when you are editing in Final Cut Pro, it’s best to leave all of the final mixing until the end. In Premiere Pro, I tend to mix as I go.

Hopefully this post gives you some insight into the “guts” of the software. If you can’t send the audio to a mix engineer and don’t want to bounce over to Logic Pro, Pro Tools, or Resolve (Fairlight) yourself, then there’s no reason Final Cut Pro can’t be made to work for you.

©2023 Oliver Peters

Final Cut Pro + DaVinci Resolve

The concept of offline and online editing goes back to the origins of film editing. Work print was cut by the film editor during the creative stage of the process and then original negative was conformed by the lab and married to the final mix for the release prints (with a few steps in between). The terms offline and online were lifted from early computer lingo and applied to edit systems when the post process shifted from film to video. Thus offline equates to the creative editorial stage, while conforming and finishing services are defined as online.

Digital nonlinear edit systems evolved to become capable of handling all of these stages of creative editorial and finishing at the highest quality level. However, both phases require different mindsets and skills, as well as more advanced hardware for finishing. And so, the offline/online split continues to this day.

If you are an editor cutting local market spots, YouTube videos, corporate marketing pieces, etc, then you are probably used to performing all of these tasks on your own. However, most major commercials, TV shows, and films definitely split them up. In feature films and high-end TV shows, the film editors are separate from the sound editing/mixing team and everything goes through the funnel of a post facility that handles the finishing services. The latter is often referred to as the DI (digital intermediate) process in feature film productions.

You may be cutting on Media Composer, Premiere Pro, or Final Cut Pro, but the final assembly, insertion of effects, and color correction will likely be done with a totally different system and/or application. The world of finishing offers many options, like SGO Mistika, Quantel Rio, and Filmlight Baselight. But the tools that pop up most often are Autodesk Flame, DaVinci Resolve, and Avid Symphony (the latter for unscripted shows). And of course, Pro Tools seemingly “owns” the audio post market.

Since offline/online still exists, how can you use modern tools to your advantage?

If Apple’s Final Cut Pro is your main axe, then you might be reading this and think that you can easily do this all within FCP. Likewise, if you’ve shifted to Resolve, you’re probably wondering, why not just do it all in Resolve? Both concepts are true in theory; however, I contend that most good editors aren’t the best finishers and vice versa. In addition, it’s my opinion that Final Cut is optimized for editing, whereas Resolve is optimized for finishing. That doesn’t make them mutually exclusive. In fact, the opposite is true. They work great in tandem and I would suggest that it’s good to know and use both.

Scenario 1: If you edit with FCP, but use outside services for color and sound, then you’ll need to exchange lists and media. Typically this means AAF for sound and FCPXML for Resolve color (or possibly XML or AAF if it’s a different system). If those systems don’t accept FCPXML lists, then normally you’d need to invest in tools from Intelligent Assistance and/or Marquis Broadcast. However, you can also use Resolve to convert the FCPXML list into other formats.

If they are using Resolve for color and you have your own copy of Resolve or Resolve Studio, then simply import the FCPXML from Final Cut. You can now perform a “preflight check” on your sequence to make sure everything translated correctly from Final Cut. Take this opportunity to correct any issues before it goes to the colorist. Resolve includes media management to copy and collect all original media used in your timeline. You have the option to trim files if these are long clips. Ideally, the DP recorded short takes without a lot of resets, which makes it easy to copy the full-length clip. Since you are not rendering/exporting color-corrected media, you aren’t affected by the UHD export limit of the free Resolve version.

After media management, export the Resolve timeline file. Both media and timeline file can go directly to the colorist without any interpretation required at the other end. Finally, Resolve also enables AAF exports for audio, if you need to send the audio files to a mixer using Pro Tools.

Scenario 2: What if you are doing everything on your own and not sending the project to a colorist or mixer for finishing? Well, if you have the skillset and understand the delivery criteria, then Resolve is absolutely your friend for finishing the project. For one thing, owning Resolve means you could skip purchasing Apple Motion, Compressor, and/or Logic Pro, if you want to. These are all good tools to have and a real deal from a cost standpoint; however, Resolve or Resolve Studio definitely covers most of what you would do with these applications.

Start the same way by sending your FCPXML into Resolve. Correct any editorial issues, flatten/collapse compound and multicam clips, etc. Insert effects and titles or build them in the Fusion page. Color correct. When it comes to sound, the Fairlight page is a full-fledged DAW. Assuming you have the mixing chops, then Fairlight is a solid stand-in for Logic Pro, Pro Tools, or other DAWs. Finally, export the various formats via the Deliver page.

Aside from the obvious color and mixing superiority of Resolve over Final Cut Pro, remember that you can media-manage, as well as render out trimmed clips – something that FCP won’t do without third-party applications. It’s also possible to develop proxy workflows that work between these two applications.

While both Final Cut Pro and DaVinci Resolve are capable of standing alone to cover the creative and finishing stages of editing, the combination of the two offers the best of all worlds – a fast editing tool and a world-class finishing application.

©2023 Oliver Peters