AES “Recommendation for Loudness of Audio Streaming & Network File Playback.”

I’d like to share my observations and views on the recently published AES Technical Document AES TD1004.1.15-10 that specifics best practices for Loudness of Audio Streaming and Network File Playback.

The document is a collection of Loudness processing guidelines for diverse platform dependent media streaming and downloading. This would include music, spoken word, and possible high dynamic audio in video streams. The document credits some of the most well respected industry leading professionals, including Bob Katz, Thomas Lund, and Florian Camerer. The term “Podcast” is directly referenced once in the document, where the author(s) state:

Network file playback is on-demand download of complete programs from the network, such as podcasts.”

I support the purpose of this document, and I understand the stated recommendations will most likely evolve. However in my view the guidelines have the potential to create a fair amount of confusion for producers of spoken word content, mainly Podcast producers. I’m specifically referring to the suggested 4 LU range (-16.0 to -20.0 LUFS) of acceptable Integrated Loudness Targets and the solutions for proper targeting.

Indeed compliance within this range will moderately curtail perceptual loudness disparities across a wide range of programs. However the leniency of this range is what concerns me.

I am all for what I refer to as reasonable deviation or “wiggle room” in regard to Integrated Loudness Target flexibility for Podcasts. However IMHO a -20 LUFS spoken word Podcast approaches the broadcast Loudness Targets that I feel are inadequate for this particular platform. A comparable audio segment with wide dynamics will complicate matters further.

I also question the notion (as stated in the document) of purposely precipitating clipping when adding gain “to handle excessive peaks.”

And there is no mention of the perceptual disparities between Mono and Stereo files Loudness Normalized to the same Integrated Loudness Target. For the record I don’t support mono file distribution. However this file format is prevalent in the space.

Perspective

I feel the document’s perspective is somewhat slanted towards platform dependent music streaming and preservation of musical dynamics. In this category, broad guidelines are for the most part acceptable. This is due to the wide range of production techniques and delivery methods used on a per musical genre basis. Conversely spoken word driven audio is not nearly as artistically diverse. Considering how and where most Podcasts are consumed, intelligibility is imperative. In my view they require much more stringent guidelines.

It’s important to note streaming services and radio stations have the capability to implement global Loudness Normalization. This frees content creators from any compliance responsibilities. All submitted media will be adjusted accordingly (turned up or turned down) in order to meet the intended distribution Target(s). This will result in consistency across the noted platform.

Unfortunately this is not the case in the now ubiquitous Podcasting space. At the time of this writing I am not aware of a single Podcast Network that (A) implements global Loudness Normalization … and/or … (B) specifies a requirement for Integrated Loudness and Maximum True Peak Targets for submitted media.

Currently Podcast Loudness compliance Targets are resolved by each individual producer. This is the root cause of wide perceptual loudness disparities across all programs in the space. In my view suggesting a diverse range of acceptable Targets especially for spoken word may further impede any attempts to establish consistency and standardization.

PLR and Retention of Music Dynamics

The document states: “Users may choose a Target Loudness that is lower than the -16.0 LUFS maximum, e.g., -18.0 LUFS, to better suit the dynamic characteristics of the program. The lower Target Loudness helps improve sound quality by permitting the programs to have a higher Peak to Loudness Ratio (PLR) without excessive peak limiting.”

The PLR correlates with headroom and dynamic range. It is the difference between the average Loudness and maximum amplitude. For example a piece of audio Loudness Normalized to -16.0 LUFS with a Maximum True Peak of -1 dBTP reveals a PLR of 15. As the Integrated Loudness Target is lowered, the PLR increases indicating additional headroom and wider dynamics.

In essence low Integrated Loudness Targets will help preserve dynamic range and natural fidelity. This approach is great for music production and streaming, and I support it. However in my view this may not be a viable solution for spoken word distribution, especially considering potential device gain deficiencies and ubiquitous consumption habits carried out in problematic environments. In fact in this particular scenario a moderately reduced dynamic range will improve spoken word intelligibility.

Recommended Processing Options and Limiting

If a piece of audio is measured in it’s entirety and the Integrated Loudness is higher than the intended Target, a subtractive gain offset normalizes the audio. For example if the audio checks in at -18.0 LUFS and you are targeting -20.0 LUFS, we simply subtract 2 dB of gain to meet compliance.

Conversely when the measured Integrated Loudness is lower than the intended Target, Loudness Normalization is much more complex. For example if the audio checks in at -20.0 LUFS, and the Integrated Loudness Target is -16.0 LUFS, a significant amount of gain must be added. In doing so the additional gain may very well cause overshoots, not only above the Maximum True Peak Target, but well above 0dBFS. Inevitably clipping will occur. From my perspective this would clearly indicate the audio needs to be remixed or remastered prior to Loudness Normalization.

Under these circumstances I would be inclined to reestablish headroom by applying dynamic range compression. This approach will certainly curtail the need for aggressive limiting. As stated the reduced dynamic range may also improve spoken word intelligibility. I’m certainly not suggesting aggressive hyper-compression. The amount of dynamic range reduction is of course subjective. Let me also stress this technique may not be suitable for certain types of music.

Additional Document Recommendations and Efficiency

The authors of the document go on to share some very interesting suggestions in regard to effective Loudness Normalization:

1) “If level has to be raised, raise until it reaches Target level or until True Peak reaches 0 dBTP, whichever occurs first. Thus, the sound quality will be preserved, without introducing excessive peak limiting.”

2) “Perform what is noted in example 1, but keep raising the level until the program level reaches Target, and apply either peak limiting or allow some clipping to handle excessive peaks. The advantage is more consistent loudness in the stream, but this is a potential sonic compromise compared to example 1. The best way to retain sound quality and have more consistent loudness is by applying example 1 and implementing a lower Target.”

With these points in mind, please review/demo the following spoken word audio segment. In my opinion the audio in it’s current state is not optimized for Podcast distribution. It’s simply too low in terms of perceptual loudness and too dynamic for effective Loudness Normalization, especially if targeting -16.0 LUFS. Due to these attributes suggestion 1 above is clearly not an option. In fact neither is option 2. There is simply no available headroom to effectively add gain without driving the level well above full scale. Peak limiting is unavoidable.

1

I feel the document suggestions for the segment above are simply not viable, especially in my world where I will continue to recommend -16.0 LUFS as the recommended Target for spoken word Podcasts. Targeting -18.0 LUFS as opposed to -16.0 LUFS is certainly an option. It’s clear peak limiting will still be necessary.

Below is the same audio segment with dynamic range compression applied before Loudness Normalization to -16.0 LUFS. Notice there is no indication of aggressive limiting, even with a Maximum True Peak of -1.7 dBTP.

2

Regarding peak limiting the referenced document includes a few considerations. For example: “Instead of deciding on 2 dB of peak limiting, a combination of a -1 dBTP peak limiter threshold with an overall attenuation of 1 dB from the previously chosen Target may produce a more desirable result.”

This modification is adequate. However the general concept continues to suggest the acceptance of flexible Targets for spoken word. This may impede perceptual consistency across multiple programs within a given network.

Conclusion

The flexible best practices suggested in the AES document are 100% valid for music producers and diverse distribution platforms. However in my opinion this level of flexibility may not be well suited for spoken word audio processing and distribution.

I’m willing to support the curtailment of heavy peak limiting when attempting to normalize spoken word audio (especially to -16.0 LUFS) by slightly reducing the intended Integrated Loudness Target … but not by much. I will only consider doing so if and when my personal optimization methods prior to normalization yield unsatisfactory results.

My recommendation for Podcast producers would be to continue to target -16.0 LUFS for stereo files and -19.0 LUFS for mono files. If heavy limiting occurs, consider remixing or remastering with reduced dynamics. If optimization is unsuccessful, consider lowering the intended Integrated Loudness Target by no more than 2 LU.

A True Peak Maximum of <= -1.0 dBTP is fine. I will continue to suggest -1.5 dBTP for lossless files prior to lossy encoding. This will help ensure compliance in encoded lossy files. What’s crucial here is a full understanding of how lossy, low bit rate coders will overshoot peaks. This is relevant due to the ubiquitous (and not necessarily recommended) use of 64kbps for mono Podcast audio files.

Let me finish by stating the observations and recommendations expressed in this article reflect my own personal subjective opinions based on 11 years of experience working with spoken word audio distributed on the Internet and Mobile platforms. Please fell free to draw your own conclusions and implement the techniques that work best for you.

-paul.

Loudness Meter Descriptors …

In the recent article published on Current.org “Working Group Nears Standard for Audio Levels in PRSS Content”, the author states:

“Working group members believe that one solution may lie in promoting the use of Loudness Meters, which offer more precision by measuring audio levels numerically. Most shows are now mixed using peak meters, which are less exact.”

Peak Meters are exact – when they are used to display what they are designed to measure:Sample Peak Amplitude. They do not display an accurate representation of average, perceived loudness over time. They should only be used to monitor and ultimately prevent overload (clipping).

It’s great that the people in Public Radio are finally addressing distribution Loudness consistency and compliance. My hope is their initiative will carry over into their podcast distribution models. In my view before any success is achieved, a full understanding of all spec. descriptors and targets would be essential. I’m referring to Program (Integrated) Loudness, Short Term Loudness, Momentary Loudness, Loudness Range, and True Peak.

Loudness Meter

A Loudness Meter will display all delivery specification descriptors numerically and graphically. Meter descriptors will update in real time as audio passes through the meter.

Short Term Loudness values are often displayed from a graphical perspective as designed by the developer. For example TC Electronic’s set of meters (with the exception of the LM1n) display Short Term Loudness on a circular graph referred to as Radar. Nugen Audio’s VisLM meter displays Short Term Loudness on a grid based histogram. Both versions can be customized to suit your needs and work equally well.

meters-480

Loudness Meters also include True Peak Meters that display any occurrences of Intersample Peaks.

Descriptors

All Loudness standardization guidelines specify a Program Loudness or “Integrated Loudness” target. This time scaled descriptor indicates the average, perceived loudness of an entire segment or program from start to finish. It is displayed on an Absolute scale in LUFS (Loudness Units relative to Full Scale), or LKFS (Loudness Units K Weighted relative to Full Scale). Both are basically the same. LUFS is utilized in the EBU R128 spec. and LKFS is utilized in the ATSC A/85 spec. What is important is that a Loudness Meter can display Program Loudness in either LUFS or LKFS.

The Short Term Loudness (S) descriptor is measured within a time window of 3 seconds, and the Momentary Loudness (M) descriptor is measured within a time window of 400 ms.

The Loudness Range (LRA) descriptor can be associated with dynamic range and/or loudness distribution. It is the difference between average soft and average loud parts of an audio segment or program. This useful indicator can help operators decide whether dynamic range compression is necessary.

Gating

The specification Gate (G10) function temporarily pauses loudness measurements when the signal drops below a relative threshold, thus allowing only prominent foreground sound to be measured. The relative threshold is -10 LU below ungated LUFS. Momentary and Short Term measurements are not gated. There is also a -70 LUFS Absolute Gate that will force metering to ignore extreme low level noise.

Absolute vs. Relative

I mentioned that LUFS and LKFS are displayed on an Absolute scale. For example the EBU R128 Program Loudness target is -23.0 LUFS. For Podcast/Internet/Mobile the Program Loudness target is -16.0 LUFS.

There is also a Relative scale that displays LU’s, or Loudness Units. A Relative LU scale corresponds to an Absolute LUFS/LKFS scale, where 0 LU would equal the specified Absolute target. In practice, -23 LUFS in EBU R128 is equal to 0 LU. For Podcast/Mobile -16.0 LUFS would also be equal to 0 LU. Note that the operator would need to set the proper Program Loudness target in the Meter’s Preferences in order to conform.

ab-rel

LU and dB Relationship

1 LU is equal to 1 dB. So for example you may have measured two programs: Program A checks in at -20 LUFS. Program B checks in at -15 LUFS. In this case program B is +5 LU louder than Program A.

Placement

Loudness Meter plugins mainly support online (Real Time) measurement of an audio signal. For an accurate measurement of Program Loudness of a clip or mixed segment the meter must be inserted in the DAW at the very end of a processing chain, preferably on the Master channel. If the inserts on the Master channel are post fader, any change in level using the Master Fader will result in a global gain offset to the entire mix. The meter would then (over time) display the altered Program Loudness.

If your DAW’s Master channel has pre fader inserts, the Loudness Meter should still be inserted on the Master Channel. However the operator would first need to route the mix through a Bus and use the Bus channel fader to apply global gain offset. The mix would then be routed to the Master channel where the Loudness Meter is inserted.

If your DAW totally lacks inserts on the Master channel, Buses would need to be used accordingly. Setup and routing would depend on whether the buses are pre or post fader.

Some Loudness Meter plugins are capable of performing offline measurements in certain DAW’s on selected regions and/or clips. In Pro Tools this would be an Audio Suite process. You can also accomplish this in Logic Pro X by initiating and completing an offline bounce through a Loudness Meter.

-paul.

Fresh Air Podcast: Audio Analysis …

In my No Free Pass for Podcasts post I talked about why the Broadcast Loudness specs. are not necessarily suitable for Podcasts. I noted that the Program Loudness targets for EBU R128 and ATSC A/85 are simply too low for internet and mobile audio distribution. Add excessively dynamic audio to the mix and it will complicate matters further, especially when listeners use mobile devices to consume their media in less than ideal ambient spaces.

fa-processed

Earlier today I was discussing this issue with someone who is well versed in all aspects audio production and loudness processing. He noted that ” … the consensus of it all is, that it is a bad idea to take a really nice standard that leaves plenty of headroom and then start creating new standards with different reference values.” The fix would be to “keep production and storage at -23.0 LUFS and then adjust levels in distribution.” Valid points indeed. However in the real world this mindset is unrealistic, especially in the internet/mobile/Podcasting space.

The fact of the matter is there is no way to avoid the necessity to revise the standards that simply do not work on a platform that consists of unique variables.

And so considering these variables, the implementation of thoughtful, revised, best practices that include platform specific targets for Program Loudness, Loudness Range, and True Peak are unavoidable. Independent Podcasters and network driven Podcasts using arbitrary production techniques and delivery methods simply need direction and guidance in order to comply. In the end it’s all about presenting well produced media to the listener.

Recently I came across a tweet where someone stated “I love the show but it is consistently too quiet to listen to on my phone.” They were referring to the NPR program Fresh Air. I’m not exactly sure if this person was referring to the radio broadcast stream or the distributed Podcast. Either way it’s an interesting assertion that I can directly relate to.

I subscribe to the Fresh Air Podcast. This will probably not surprise you – I refuse to listen to the Podcast right out of the box. When a new show pops up in Instacast, I download the file, decode to WAV, convert to stereo, and then reprocess the audio. I tweak the dynamic range and address show participant audio level variations using various plugins. I then bump things up to -16.0 LUFS (using what I like to refer to as “The Lund Method”) while supplying enough headroom to comply with -1.0 dBTP as my ultimate ceiling. I’ll get into the specifics in a future post.

According to the leading expert Mr. Thomas Lund:

“Mobile and computer devices have a different gain structure and make use of different codecs than domestic AV devices such as television. Tests have been performed to determine the standard operating level on Apple devices. Based on 1250 music tracks and 210 broadcast programs, the Apple normalization number comes out as -16.2LKFS (Loudness, K-weighted, relative to Full Scale) on a BS.1770-3 scale.

It is, therefore, suggested that when distributing podcast or Mobile TV, to use a target level no lower than -16LKFS. The easiest and best-sounding way to accomplish this is to: 1) Normalize to target level (-24LKFS); 2) Limit peaks to -9dBTP (Units for measurement of true peak audio level, relative to full scale); and 3) Apply a gain change of +8dB. Following this procedure, the distinction between foreground and background isn’t blurred, even on low-headroom platforms.”

In this snapshot I demonstrate the described workflow. I’m using two independent instances of the bx_control plugin to apply the gain offsets at various stages of the signal flow. After the initial calculated offset is applied, the audio is routed through the Elixr True Peak Limiter and then out through the second instance of bx_control applying +8dB of static gain. You can also replicate this workflow on an off-line basis. Note that I’ve slightly altered the limiting recommendation.

Lund-small

So why do I feel the need to do this?

Podcast Source

These are the specs. and the waveform overview of a recently published Fresh Air Podcast in it’s entirety:

raw-specs
fa-source-complete

Next is a 3 min. audio segment lifted from the published Podcast. The stats. display measurements of the attached 3 min. segment:

source_revised
source-1

Podcast Optimized for Internet/Mobile

Below is the same 3 min. segment. I reprocessed the audio to make it suitable for Podcast distribution. The stats. display measurements of the attached audio segment:

web-specs-2
source-2

The difference between the published source audio and the reprocessed version is quite obvious. The Loudness Normalized audio is so much more intelligible and easier to listen to. In my view the published audio is simply out of spec. and unsuitable for a Podcast.

Bear in mind the condition of the source audio is not uncommon. The problems that persist are not exclusive to podcasts distributed by NPR or by any of their affiliates. Networks with global reach need to recognize their Podcast distribution platforms as important mechanisms to expand their mass appeal.

It has been noted that the Public Radio community in general is exploring ways to enhance the way in which they produce their programs with focus on loudness standardization. My hope hope is this carries over to their Podcast platforms as well.

-paul.

For more information please refer to “Managing Audio Loudness Across Multiple Platforms” by Thomas Lund at TVTechnology.com.

No Free Pass for Podcasts …

I think it was in the mid to late 1980’s. I was still living home, totally fixated on what was happening with Television devices, programming and transmission. Mainly the advent of MTS Stereo compatible TV’s and VCR’s. I remember waiting patiently for weekly episodes of programs like Miami Vice and Crime Story to air. I would pipe the program audio through my media system in glorious MTS stereo. For me this was a game changer.

vice

I also remember it was around the same time that Cable TV became available in the area. I convinced my Mom and Dad to allow me to order it. Initially it was installed on the living room TV, and eventually made it’s way on to additional TV’s throughout our home. For the most part it was a huge improvement in terms of reception and of course program diversity.

However there was one issue that struck me from the very beginning: the wide variations in loudness between network TV Shows, Movies, and Adverts. In fact it was common for targeted, poorly produced, and exceedingly loud local commercials to air repeatedly throughout broadcast transmissions. Reaching for the remote to apply volume attenuation was a common occurrence and a major annoyance.

Obviously this was not isolated. The issue was widespread and resulted in a public outcry to correct these inconsistencies. In 2010 The CALM Act was implemented. The United States and Europe (and many other regions) adopted and now regulate loudness standardization guidelines for the benefit of the public at large.

If there is anyone out there who cannot relate to this “former” problem, I for one would be very surprised.

Well guess what? We now have the same exact problem existing on the most ubiquitous media distribution platform in existence – the internet.

I realize any expectation of widespread audio loudness standardization on the internet would be unreasonable. There’s just too much stuff out there. And those who create and distribute the media possess a wide scope of skills. However there is one sort of passionate and now ubiquitous subculture that may be ripe for some level of standardization. Of course I’m referring to the thousands upon thousands of independenlty produced Podcasts available to the masses.

In the past I’ve made similar public references to the following exercise. Just in case you missed it, please try this – at you own risk!

Put on your headphones and queue up this episode of The Audacity to Podcast. Set your playback volume at a comfortable level, sit back, and enjoy. After a few minutes, and without changing your playback volume setting – queue up this episode of the Entrepreneur on Fire podcast.

waves-1

Need I say more?

From what I gather both programs are quite popular and highly regarded. I have no intension of suggesting that either producer is doing anything wrong. The way in which they process their audio is their artistic right. On the other hand in my view there is one responsibility they both share. That would be the obligation to deliver well produced content to their subscribers, especially if the Podcast generates a community driven revenue stream. It’s the one thing they will always have in common. And so I ask … wouldn’t it make sense to distribute media following audio processing best practices resulting in some level of consistency within this passionate subculture?

I suspect that some Podcast producers purposely implement extreme Program Loudness levels in an attempt to establish “supremacy on the dial.” This issue also exists in radio broadcast and music production, although things have improved ever since Loudness War participants were called to task with the inception of mandatory compliance guidelines.

I’ve also noticed that many prolific Podcast Producers (including major networks) are publishing content with a total lack of Program Loudness consistency within their own catalogs form show to show. Even more troubling, Podcast aggregation networks rarely specify standardization guidelines for content creators.

It’s important to note that many people who consume audio delivered on the internet do so in less than ideal ambient spaces (automobiles, subways, airplanes etc.) using low-fi gear (ear buds, headphones, mobile devices, and compromised desktop near fields). Simply adopting the broadcast standards wouldn’t work. The existing Program Loudness targets are simply unsuitable, especially if the media is highly dynamic. The space needs revised specs. in order to optimize the listening experience.

Loudness consistency from a Podcast listener’s perspective is solely in the hands of the producers who create the content. In fact it is possible producers may even share common subscribers. Like I said – the space is ripe for standardization.

Currently loudness compliance recommendations are sparse within this massive community driven network. In my view it’s time to raise awareness. A target specification would universally improve the listening experience and ultimately legitimize the viability of the platform.

For the record, I advocate:

File Format: Stereo, 128kbps minimum.
Program Loudness: -16.0 LUFS with acceptance of a reasonable deviation.
Loudness Range: 8 LU, or less.
True Peak Ceiling: -1.0 dBTP in the distribution file. Of course this may be lower.

Quick note: when I refer to Podcasts, from a general perspective I am referring to audio programs and videos/screencasts/tutorials that primarily consist of spoken word soundtracks. Music based Podcasts or cinema styled videos with high impact driven soundtracks may not necessarily translate well when the Loudness Range (and Dynamic Range) is constricted.

For further technical insight, please refer to “Audio for Mobile TV, iPad, and iPod” – Thomas Lund, TC Electronic.

-paul.