Reversioning Audio Content For Multiple Loudness Delivery Specs
There is a growing need to be able to reversion mixed content to deliver it on a variety of different platforms, often with different delivery specifications. In this article, loudness guru Mike Thornton takes you through the issues, challenges and a range of solutions.
This all came about after we received an email from community member Robert Head, who asked…
“Hello, I work on a lot of radio ads/promos etc. These are mixed to -24 LKFS. Increasingly I am being asked to deliver multiple files with differing loudness level for the web, streaming, YouTube etc. I would be interested in an article on best practices for this multiple loudness delivery workflow from someone with more experience. How best to keep dynamics intact, noise floor/noise issues in files with a wide loudness range etc. Thank you for reading this. I don't really have the opportunity to work with more experienced engineers and really enjoy reading the articles on Production Expert and find them very informative.”
Why Do We Need This?
Before we get into the practical side of this, we will start with why we need it and look at the challenges this requirement presents to understand what a solution needs to achieve before outlining a number of solutions.
Back in the day, with just one broadcast delivery format, we could produce one mix, and that would be that. The content didn’t tend to move from one platform to another. That is no longer the case. First, we had remixes of films to suit broadcast and DVD platforms, but that only tends to happen on large-budget projects that can afford a rerecording mixer the time to remix a film for playback in the home or for broadcast.
Cost Effective Solutions Required
With growing budgetary constraints, we need to find ways to do this as quickly and cheaply as possible.
In essence, there are two major changes. Taking a film and repurposing it for broadcast. Or taking a broadcast mix and reversioning it for online.
The Problem
As with all things, it isn’t quite as simple as that, as there is no loudness delivery spec for theatrical cinema releases in the same way as we have for broadcast and online, and to complicate matters further, there are multiple delivery specs in both the broadcast and online sectors. In addition, there is the challenge of content produced for streaming services like Netflix, which has chosen to resurrect Dolby’s dialog gating measurement system, so there is plenty to talk about.
However, this table only tells one part of the story: the integrated loudness requirement for each platform. Other parameters need to be considered, including maximum true peak level and loudness range (dynamic range).
RTW has produced an excellent resource, Worldwide Loudness Delivery Standards, which lists the two pass/fail criteria, integrated loudness and maximum true peak, for a wide range of platforms worldwide. As most delivery specs don’t make Loudness Range (LRA) a pass/fail criterion, some do make recommendations for appropriate LRA figures for their platforms. For example, the Digital Production Partnership (DPP) updated their unified UK delivery specs for all UK broadcasters and added this guidance on Loudness Range…
Loudness Range - This describes the perceptual dynamic range measured over the duration of the programme - Programmes should aim for an LRA of no more than 18LU
Loudness Range of Dialogue - Dialogue must be acquired and mixed so that it is clear and easy to understand - Speech content in factual programmes should aim for an LRA of no more than 6LU. A minimum separation of 4LU between dialogue and background is recommended.
In Canada, both CBC and Radio Canada request an LRA that is less than 10. If you want to learn more about this, check out my article Loudness and Dialog Intelligibility in TV Mixes - What Can We Do About TV Mixes That Are Too Cinematic?
In their Netflix Sound Mix Specifications & Best Practices v1.5, for Netflix branded content, they state the following in respect of LRA…
The following loudness range (LRA) values will play best on the service:
5.1 program LRA between 4 and 18 LU
2.0 program LRA between 4 and 18 LU
Dialog LRA of 10 LU or less
However, in the article Netflix Loudness And Dynamic Range Developments Update, Netflix has stated that they have chosen to use metadata, specifically the ‘anchorLoudness’ metadata set. It measures the dynamic range and dialogue levels and then sends this in the metadata to the consumer’s equipment so that the dialog levels in all their content will match when played back on the consumer’s device. We suspect that this is to cover the non-branded content on Netflix.
A More Detailed List
With this additional information, here is a more detailed table, which includes Integrated Loudness, maximum True Peak and Loudness Range.
Reducing Peak To Loudness Ratio
This table highlights the challenge we have to address with content requiring a higher integrated loudness target. This results in the average coming closer to the maximum true peak level. There is less headroom to work with, so the available dynamic range is reduced.
The reason for this is that some platforms have a higher integrated loudness target due to the limitations of mobile devices, both in terms of gain in the playback chain and restricted dynamic range due to battery power.
There is a measurement for this: the peak-to-loudness ratio (PLR). This is the difference between the maximum true peak level and the integrated loudness (which used to be called the Crest Factor). Here are two examples…
ATSC A/85 PLR = -2-(-23) = 21
Spotify Loud PLR = -2-(-11) = 9
Why does this matter? Check out this excellent video from the team at MeterPlugs…
This video shows very clearly the challenge when reversioning content to a higher integrated loudness spec. For example, if we are reversioning a piece of broadcast content with a PLR of 21 to be used on a streaming platform like Apple Music, which requires the content to have a PLR of 15.
A Simple Gain Adjustment Won’t Fix It
A simple gain lift of 6dB will bring the integrated loudness up to -16LKFS, but the 6dB gain lift will take the maximum true peak level through digital headroom to 6dB above the spec (and 5dB above digital headroom), creating significant peak distortion.
To avoid this, as part of the reversioning process, we need to reduce the dynamic range of the content to be able to fit it into the smaller PLR window.
How Do We Do This?
Now we understand the challenges where the integrated loudness gets closer to the digital headroom, let’s look at some ways we can achieve these changes as part of the deliverables reversioning.
Remix The Content From Scratch
In many respects, this is the ideal option. However, it is not always practical or, in some cases, possible and is outside the remit of this tutorial.
Use Conventional Dynamics Tools To Contain The Dynamic Range
I did this for a long time, especially on the Production Expert podcast. Each voice would have its own track in the session with conventional EQ and dynamics plugins set to deliver -23LUFS content. Then I set up a ‘Limiter’ channel on an Aux track, which had plugins on it to take the conventional broadcast mix and change it into an online podcast mix. As you can see, there are three plugins on this channel.
First, I have an Avid Pro Compressor plugin, effectively providing a soft limit, with the knee control set to provide a gentle transition from no compression to full 20:1 limiting.
Next, we have an instance of the BBE Sonic Maximiser plugin, which I have used for many years, starting with a hardware unit before graduating to a plugin to help with top-end clarity without using EQ.
Finally, I have an instance of Avid’s Pro Limiter, which is an excellent true peak limiter set to bring the gain up from a -23LUFS mix to a -16LUFS mix and set to limit at -3dBTP, which is my preferred setting to protect any lossy encoders downstream.
The only plugin to come after this is an instance of Nugen Audio’s VisLM set-up so that -16LUFS displays as 0LU and my preferred colour banding in green from +2 to -3LU, which is my comfort zone for this kind of content.
Use Specialised Dynamics Tools To Contain The Dynamic Range
This is a variant of the previous option. Instead of using the Avid Pro Compressor plugin, I am using POWAIR from Sound Radix and use its Loudness Leveller option to contain the dynamic range. I still need the Avid Pro Limiter as a True Peak Limiter in the final stage, but as POWAIR is handling the gain lift, I can set the Pro Limiter just to be a true peak limiter set to -3dBTP.
You can learn more about POWAIR in these articles…
Use A Tool Designed For The Job
There are a number of tools out there that are designed to undertake this task in one step. My preferred tool for this is LM-Correct from Nugen Audio.
Above, we have a basic AudioSuite version of LM-Correct in Pro Tools. The left-hand side of the plugin displays the integrated loudness and maximum true peak of the source file. The right-hand side displays the target integrated loudness and the maximum true peak level. In the middle, there's a Presets button. When we click on it, it displays an extensive range of presets covering most of the delivery specs worldwide, as shown on the right-hand side of the image above.
But if you need something different, you can type in the appropriate values into the target boxes. I prefer to deliver -3dBTP.
Click the Analyze button to display the source file measurements. If some of the numbers are displayed as amber, it denotes that those measurements are outside the target's spec.
But if we're all happy with that, we can simply press the Render button. LM-Correct will make a new loudness-compliant version of that file.
Now this is great for fixing a mix that wasn't quite compliant. However, with the proliferation of channels and delivery systems, more and more often we need to be able to deliver different versions of the same mix to be compliant to different delivery specs.
LM-Correct includes an intelligent true peak limiter and can constrain the short-term and momentary loudness, which you can enable and disable from the Options window. This feature is really helpful for producing mixes compliant with the EBU R128 S1 spec for short-form content like ads and trailers.
Back in the main window, you can see that we now have a new row of parameters and can choose to display either the maximum short-term or momentary measurements. For short-term content, the EBU spec restricts the maximum short-term loudness, which is averaged over 3 seconds, to a maximum of minus 18 LUFS, or 5 LU above the target loudness.
If you've bought LM-Correct with the optional DynApt extension, then Dynamic Range Changes, or LRA in loudness speak, can be made, which is really useful for repurposing content (the aim of this article). For example, taking a film mix and repurposing it for TV or producing a podcast from a radio mix.
You can see that we now have the LRA Source and Target boxes in the main window. We need to make sure that all the appropriate features are enabled by checking these blue icons in the middle.
In this example, I'm going to be taking a BBC Radio 3 documentary and reversion it as a podcast. I want a true peak max of -3 dBTP, so I don't overload the lossy codecs. Set the Integrated Loudness to -16 LUFS, which is my preferred loudness for online content, and set the LRA to 6 because it will now be a podcast. But as this is a long-form documentary, I'm not going to enable the short-term max option.
Click on the Render button, and you will get a file that is not only loudness compliant but also dynamically optimised for online content suitable to be listened to on mobile devices in noisy environments.
You can learn more about LM-Correct from Nugen Audio in this article…