Podcasts and -23.0 LUFS ...
I stumbled upon a few interesting comments in response (though not directly) to an article that I wrote back in July regarding the lack of delivery loudness consistency in the Podcasting space. I’m specifically referring to Integrated (Program) Loudness of Podcasts (or videos), distributed by independent producers and/or Podcast aggregation networks.
Real quick - the Integrated Loudness descriptor for any audio segment is basically the average perceived loudness of a piece of audio, measured in it’s entirety. I reiterate - an audio segment must be measured in it’s entirety in order to achieve accuracy. This can be done in real-time (Online) or “Offline”, as an independent process. Both options have their pluses and minuses.
Please note that I’m not referring to Peak Amplitude, that in fact has no bearing on audible perception. It’s simply a representation of amplitude , including the possible occurrences of overloud, also known as clipping.
So here is the problem and the basis of my article:
Podcast subscriber Joe queues up 20 or so Podcasts on his mobile device in preparation for a long commute. It could be a car, plane subway, bus - you name it. I think it’s pretty safe to use Joe’s commute scenario as a gauge, considering the amount of people who now consume their media on mobile devices in a wide range of ambient environments.
Joe’s Podcast selections are quite diverse. A few are distributed by certain independent producers. Others are also distributed by independent producers, however they are part of a collective network, like 5by5, TechPodcasts, PodcastOne, etc.
I would argue that the first problem Joe will encounter is a lack of consistency in Podcast Program Loudness from show to show, forcing him to make constant playback volume adjustments as he listens. This is due to producers sort of winging it, publishing programs at random, with no guidelines to abide by.
Please don’t confuse level inconsistencies of inter-program audio components (multiple participants, music beds, spots, intros, etc.) … with Integrated Loudness inconsistencies FROM program TO program. Of course level management between individual audio components is an important aspect of any competent post production workflow. However it is in fact a totally different issue.
The point is without even the slightest reference to some sort of Integrated Loudness target or best practice, producers have nothing to shoot for. This results in a less than ideal experience for the listeners and/or subscribers.
The EBU (European Broadcasting Union) and ATSC (Advanced Television Systems Committee) have implemented Loudness Standardization guidelines for Broadcast Media. In fact in case you are not aware - we have actual laws in the U.S. (The CALM Act) that require distribution networks to comply. The Integrated Loudness target for EBU R128 is -23.0 LUFS. The ATSC A/85 target is -24.0 LUFS.
I’m not going to get into the specifics in terms of how and why these targets were implemented. I will say that some of the most savvy audio people on earth were involved in the studies and implementation.
So, why not just dive in and support/advocate the Broadcast Standards for internet/mobile audio distribution?
Back to the reaction to the article that I wrote …
A few people sort of discounted my call to action. They noted that the broadcasters set the standards, and that’s what we *should* abide by. This of course is if the producer ever decided to establish a level of consistency with what they distribute.
And so they argue, if the broadcasters target -23.0 LUFS, so should the Podcasters. Note that there is a very well designed audio DAW out there that advertises support for non-customizable Loudness Compliance processing (to -23.0 LUFS). All well and good - but herein lies the problem …
I respectfully challenge anyone out there to queue up a Podcast on your iPhone/Android, - whatever… that is Loudness Normalized to -23.0 LUFS, with a wide LRA (Loudness Range), and listen to it on a NYC Subway, bus, train, or airplane. In fact - try listening to it in a noisy cafe, where ambience is elevated due to a chattering crowd. And while you’re at it - try listening to the same program out through an iPad speaker under the same circumstances. Factor in an extended dynamic range, and the problem intensifies. I’m convinced you will quickly realize that for this type of media consumption, -23.0 LUFS simply does not work. It’s just not loud enough.
It is for this reason that audio distributed as Podcasts (or any form of internet/mobile audio distribution) requires revised “guidelines.” I say guidelines because I totally get the variables as well the sheer scope of such an endeavor.
Let’s face it - in the internet/mobile space there is wide range of tools, skills, and experience out there directly related to audio post best practices. I’ve also noticed that many people in the Podcast Production space expect everything for free in terms of software applications and even audio plugins. This mindset combined with the lack of easy to use tools further complicates matters. Any level of standardization would be near impossible. However, the Podcasting space is so much more of a collective “community.” All it would take would be for a few heavy hitters with a voice to advocate something, and the rest will follow.
Bottom line - I understand that the mere suggestion of implementing yet another “standard” further complicates an issue that is already quite complex. However - there are valid reasons why revisions are a necessity.
That’s my take.
If you’re looking for a more technical approach and/or discussion - please feel free to express your thoughts. Also, feel free to contact me privately.
By the way - a few prominent Public Radio Engineers who are responsible for the distribution of Public Radio shows as Podcasts now support the “revised” Loudness Standardization targets for internet/mobile distribution. I’m hearing they expect widespread acceptance of these standards in the near future.
For the record: -16.0 LUFS Integrated (Program Loudness) with a reasonable allowance for deviation. And, True Peak Ceiling of -1.0 dBTP in the lossy distribution file. To play it safe we recommend that you target -1.5 dBTP in order to safely avoid any possible occurrences of Intersample Peaks.
For additional insight, please refer to "Managing Audio Loudness Across Multiple Platforms" by Thomas Lund
-paul.
@produceNewMedia.
