If you are mixing audio for delivery using your NLE, you need to not only monitor levels, but also check the loudness of the entire programme. Oliver Peters explains levels, loudness and LUFs.
I started in broadcast and back then the trusty analog VU meter was gospel. Everyone knew that a level between -3VU and 0VU was the sweet spot. Occasional peaks up to +3VU were still fine, but if your meter pegged all the way to the right and never moved off the mark, then that was clearly no good.
Fast forward to the digital age and we’ve all gotten used to graphical full scale meters with a range from infinity to 0dB. These are peak reading meters, so in theory anything exceeding 0 will clip or distort. If you are mixing your audio hot with an average level of around -1dB, then you are clearly doing it wrong. That might be fine for mastering hit records to CD, but it’s unacceptable for broadcast and most streaming services, including YouTube and Spotify.
Three factors are important for a broadcast deliverable - a proper mix, adherence to loudness specs, and proper stereo phase. These same guidelines will also improve your mix for any other form of distribution. I’ll take a deeper dive into how to approach audio within Final Cut Pro’s architecture to achieve the best results.
Under the hood
Like other aspects of Final Cut Pro, working with audio can appear deceptively simple - even simplistic - but if you dig deeper there’s plenty of complexity. Most DAWs and NLEs use mixer interfaces that mimic physical console layouts. It’s easy to understand routing and see levels at interim stages within the path. That’s not the Final Cut approach. I would characterize the audio flow through Final Cut as more akin to a Russian nested doll.
When you edit a clip to the timeline (project), you can adjust its audio parameters in the inspector. Audio enhancements, pan modes, and effects are pre-fader, i.e. before the volume slider. If you add a single-channel, mono clip to a stereo timeline, the audio passes through the enhancement and effect sections as a mono signal. It then hits the fader and is split into stereo before flowing to the timeline. A two-channel, stereo clip flows through the same path in stereo.
Mono clips must adhere to the “pan law.” If your actual clip level is -20dB (pan mode set to “none”), it will pass through the effect filters at -20dB. Effects are processed in mono, but the level will then appear on the full scale audio meters as -26dB. That’s because the signal has been split into stereo. However, if you change the pan mode of the -20dB mono clip to “left/right,” then the signal is split into stereo before the effects. The signal passes through the effect filters in stereo at -26dB and the full scale meters still show the same -26dB level.
In short, both pan modes show the output of the clip as -26dB, but any clip effects are internally processed at different levels. I recommend that you first set the pan mode of any mono clip to “left/right” when working in a stereo project.
The signal flows through stages
The clip on the timeline is effectively a container and is actually the second stage of this architecture. When you expand audio components on the timeline, you are revealing the first stage. Components have the same inspector parameters as the clip. If you have an audio clip with a hot level that you’d like to lower prior to applying any effects, you can either add a gain filter as the first effect in the stack or you can expand the audio components and lower the volume of the component.
When you nest a complete project into a compound clip, you have effectively created “submix” tracks. The clips within the compound have been automatically combined into audio “stems.” There’s further complexity when you use audio audio roles and lanes, but let’s keep this simple.
The compound clip also has a two-stage structure of clip container and audio components. This means that if you’ve edited clips to a timeline and then created a compound, you now have four potential audio stages. Level adjustments and filters can be applied at each stage. If you aren’t confused yet, then let me point out that you can create compound within compound within compound, and so on!
Levels and loudness
A traditional VU meter is calibrated so that 0VU on the meter is equivalent to -20dB on a full scale meter. The VU scale tracks the dB (decibel) scale, so the amount of dB change is comparable to the same change in VU. However, since the full scale meter readings are based on peaks and a faster response time, the real-time mix level during playback may look proportionally higher on the full scale meter than on a VU meter.
While analog VU meters were the norm in US broadcasting, European broadcasters often used PPM meters. In addition, some studios and control rooms also installed Dorrough loudness meters. Ultimately it’s an issue of how to measure a mix based on perceived volume. Does it sound too loud - not - what do the meters show?
Stations often competed based on the loudness of their outgoing signal and commercial blocks were great offenders. This led to the CALM Act (commercial advertising loudness mitigation act) passed in 2010 in the US, as well as similar regulations by the EU and other others. These regulations led us to current delivery specs based on loudness requirements, also referred to as “dialnorm” (dialogue normalization).
Optimal loudness levels are defined by a number of standards that vary by country. Generally the spec will specify an average program loudness level of -23 or -24 LUFS, with some allowed variance from that number, like plus or minus .5 LUFS. LUFS stands for “loudness units full scale” and one LU is equivalent to one dB. Loudness specs typically refer to LU-I, or “integrated” loudness units. This represents the volume averaged over the length of the program. Therefore, you can have quiet and loud sections, but as long as the average meets the LU-I specs, then your program is compliant.
Many programming outlets will use different standards for long-form (shows) versus short-form (promos, commercial) content. Long-form is to be measured based on dialogue (dialogue normalization), not the full mix. However, short-form should be the full mix, including dialogue, music, and sound effects. If you are delivering to a network or content distributor, then you will need to pay close attention to their delivery specs for such nuances. Otherwise, you risk rejection for a non-compliant file.
Mixing, mastering, and metering in Final Cut Pro
How do you check loudness in FCP? There’s no fancy TC Electronic’s RADAR meter like in Premiere Pro and Audition. That’s only partially true, because Final Cut Pro does include Logic Pro’s multimeter audio filter. It’s a comprehensive tool that measures a number of values, including peak volume, LUFS, and phase correlation. Mixing according to the LUFS number on the multimeter enables you to produce a compliant mix. The following are practical steps that work for me.
First, edit your program as you normally would. Make sure that any mono clips have been set to a left/right pan mode for stereo projects. Mix your levels so that the full scale audio meters bounce within a reasonable range - for instance, around -12dB. Then, nest the completed timeline into a compound clip. Expand the audio components of the compound clip and add a compressor to the component. I like the Logic Pro compressor, but use whatever sounds best to you and properly tames wide swings in dynamic range. There are plenty of good third-party plug-ins, too, such as the Maximizer module in iZotope Ozone. Even the Elements version is fine.
Apply the multimeter to the main compound clip. I also like to use old school VU meters and am a fan of the free plug-in from TBProAudio. I set that to its VU standard preset with a -20 calibration. Play the program to see where levels land. Give it enough time for the multimeter to track a cumulative LUFS average (at least :30 or longer).
In order to adjust the volume reaching the meters, you will need to change the volume of the audio component or of the output gain of the compressor effect, not the volume of the clip itself. Play it to check the levels again and tweak until you are within the target loudness range. A mix with a narrow dynamic range, like a needledrop music track, should be consistent across all three meters. Typically a LUFS reading of around -23, should also show the VU needle bouncing around -3 and the full scale meters in the -14 to -12 range.
With a bit of practice, this style of mixing will become second nature. However, if you simply want a single effect to bring your average level to the target spec, then I recommend the LUFS Meter 2 filter from Klangfreund. This simple audio processing filter will adjust the overall level to the desired LUFS value.
Final Cut Pro doesn’t work like a traditional mixer, but that won’t prevent you from completing a polished mix. Just think through the steps and practice some mixes to accomplish successful results.
Loudness vs. Speech Normalization in Film and Drama for Broadcast
Difference Between LUFS, RMS, True Peak Loudness Meters Explained