Intro from Jeff Towne: We all know the experience - listening to the radio or watching TV, a loud commercial jumps out, sending you grabbing for the volume control. Or conversely, your favorite program is too quiet leading you to crank up the volume, even though you know louder levels will startle you as the next program starts. Variations in program levels have been a problem for public radio stations, and now the public radio world is attempting to even out this difference by implementing a new loudness standard. Adhering to this standard will soon be required when submitting programs to the Public Radio Satellite System’s ContentDepot, PRX, and other organizations, and it's also just a good tool for making sure your mixes keep a steady level.
Rob Byers tells us about the concept of Loudness, and how to use it to create better mixes that are both internally consistent and compatible with other programs in the public radio system.
Today as you mix audio productions you most likely monitor levels with a peak meter — those two little bars that jump up and down in tandem with the waveform — and you know those meters don’t always line up with what you hear! You look at two very different pieces of tape on the meter (say, your studio-recorded voice against an interviewee on the phone), tweak those two voices until they appear the same on the meter, and your ears tell you they play back at quite different volumes. You might decide to forego the peak meter for the RMS meter, which can provide a small advantage over a peak meter, but they too do not take perception into account. This is a problem that a new audio measurement method, loudness, can help you solve. Finally there’s a way to simplify levels.
In this article I will walk you through a number of ways you can incorporate loudness technology in your work. I’ll start with the tool that offers the most immediate improvement to your production: the loudness meter.
If you’re already sold on using loudness and want to skip to the how-to part, click here.
What’s wrong with the meter I use now?
Short answer: The meter you use now has nothing to do with the way you perceive sound.
The most common way to represent audio is to display electrical level. That usually means a waveform, a VU meter, or a peak meter — those bouncing bar graph meters ubiquitous in digital audio. The problem? These tools don’t correspond to how the audio actually sounds to our ears. It’s possible for the audio to look one way on these meters but sound quite different. Phone tape is a frustrating example of this phenomenon: on a peak meter it often looks hotter than it sounds. Modern recorded music often displays the opposite effect; it will sound louder than it looks on a peak meter, especially when matched up against voice.
Loudness meters measure audio similarly to the way humans perceive sound. The meters analyze audio taking into account duration as well as frequency — the human ear is sensitive to some frequencies (the wail of a baby, the rustle of leaves), not so sensitive to others (the rumble of a bus, the low notes of a bass guitar). That measurement is much more consistent with the way we hear. Sounds that we hear as “loud” display higher on the meter and vice versa.
The unit of measurement, the Loudness Unit [LU], actually represents an audible difference. A change of one LU is a difference you can noticeably hear — you certainly can’t say that for one dB, the unit used on most other meters.
Loudness meters also remove an issue that other meters are prone to perpetuate: interpretation.
Let’s try a hypothetical experiment. You and I are both going to deliver a piece to a radio program that requires audio to be mixed to -15 dBFS (average levels hovering around -15 dBFS on a typical peak meter). If we both try to mix to that target, we will end up with different sounding mixes, because we will interpret the target differently. Should the peak meter consistently hit -15dBFS, go through it, or just tickle it? Loudness meters remove this interpretation issue because they give a numerical readout; either you hit the target or you don’t, no interpretation needed.
Loudness meters offer another benefit: they give you a better picture of your audio when you mix in less-than-ideal environments. Perhaps you need to mix a quick spot on headphones from home or deliver some vox pop from the noisy local coffee shop. No matter how fancy the headphones you use to monitor your audio in those environments, you will be at a disadvantage when it comes to balancing elements and judging levels. Loudness meters can make up for some of what you can’t hear because you can now trust that what you see on the meter matches the true loudness of the elements in the mix.
These meters are way more useful for production purposes than the traditional peak and VU meters of yesterday. You could say loudness meters actually show you what you should be hearing!
Is anyone else using Loudness?
Loudness was first implemented in European television broadcasting half a decade ago and is the technology behind the CALM Act in the US (legislation passed a few years ago to ensure that TV commercials aren’t blastingly loud). It’s caught on in other media since — iTunes Radio and Spotify both use their own iterations of loudness technology to balance levels from one song to the next. Pandora doesn’t. Have you noticed?
In early November 2014, the Public Radio Satellite System, public radio’s largest distributor of content, adopted a loudness standard intended to bring consistency to the system and reduce level jumps between programs. The new loudness standard is set to begin some time in 2015. PRX will follow suit, as well as major producers like American Public Media, NPR, and PRI. All of these organizations are jumping on board because loudness technology will bring marked consistency to their programming.
It’s clear the audio world accepts loudness as a superior metering technology, and its great that it’s on its way to public radio. But you can start using loudness meters now to help improve your mixes. There are numerous options for integrating loudness metering into your work, from meters built directly into audio editors, to free third-party plug-ins, to super-polished paid plug-ins.
How do I get a Loudness Meter?
Adobe Audition CC and Hindenburg Journalist PRO, already popular in the public radio world, include loudness meters. Both of these meters are extremely useful, but they are each formatted differently. We’ll talk about that later, for now simply keep an eye out for the “Short Term” reading, circled in each image below.
There are also free meters available as downloadable plug-ins. I call them “free” though they are really trials or shareware. (The folks who made these put some hard work into them, show them you appreciate their efforts and pay the upgrade fee or donate.)
And while you're showing appreciation...
Help Transom get new work and voices to public radio by donating now.
If you use an audio editor that takes VST and/or AU plugs (like Audition, Hindenburg, Dalet, DAVID, Reaper, Audacity, GarageBand, or Logic) look at the ToneBoosters EBU Loudness meter or the Klangfreund LUFS Meter. For ProTools, check out the HOFA 4U meter.
Keep in mind that you tend to get what you pay for. These plug-ins offer few customizable options and all minimize the size of the numerical displays, a feature I find important.
If you are willing to spend a little cash you can pick up the TC Electronic LM2n for about $149, which happens to be the meter Adobe chose to include in Audition. The LM2n is compatible with VST, AU, and ProTools plug-in formats (PT10 and up); it’s a great choice if you find yourself working in different audio platforms. TC Electronic is well established in the audio world, so support for this product should continue for some time. They offer a 14-day trial that is worth checking out before you commit to purchasing.
Many other loudness meter plug-ins are available at higher price points. They may have prettier graphics and more bells and whistles, but these features won’t offer much added value to the typical public radio workflow. They will, however, likely remain supported for many years to come. I suggest two in this category: the Waves WLM (ProTools, VST) and the Nugen LMB (compatible with newer versions of ProTools, AU, and VST). Izotope’s Insight also comes highly recommended and is compatible with just about every platform available.
Hardware meters are also available if you prefer a physical box with analog and/or digital inputs. These are mostly appropriate for studio installs, but if you are in the market to fill this need you’ll find plenty of options out there.
Check out the PRSS Loudness Tools list for a comprehensive list of meters and other tools.
An aside for you ProTools users: it wouldn’t surprise me to see a loudness meter included in a future update. Loudness is becoming ubiquitous, and the major platforms are all getting in on the act. Nuendo, Audition, and Hindenburg all have, so it makes sense that ProTools will follow suit… eventually.
What do I do with it?
First and foremost, I’ll set you up so you can actively mix your project — changing levels and making balance adjustments as the audio plays back live. Install the plugin (instructions can usually be found on the plugin’s website), open your project, and insert the plugin on the master track. If your session doesn’t already have a master track, make one. All of the tracks in your project will route through this master track, so it’s crucial for monitoring levels as you mix. Check out the videos below for specifics:
Loudness meters display measurements in a unit called Loudness Units Full Scale, or LUFS (As in “I lufs loudness!” You’ll occasionally see this stated as LKFS, but that’s the same as LUFS). Loudness Units is used when describing relationships. For example, the difference between -24 LUFS and -20 LUFS is 4 LU. If you see a number with “LUFS” or “LU” following, it means the measurement incorporates perception (the frequency and duration characteristics of the human ear).
It’s important to recognize that there’s a difference between the unit used on peak meters (dBFS) and LUFS. The unit dBFS (Decibels relative to Full Scale) is solely a measurement of electrical level, without any of the benefits that loudness brings. Our new world is a LUFS world!
To get started, focus your attention on the Short Term measurement. There’s a bunch of information displayed on most loudness meters, but for improved consistency in the typical radio piece you can simply use the Short Term measurement for mixing. It is an average of a rolling 3-second time window, usually displayed as a number, and it works really nicely with the pace of speech. If you ignore most everything else on the meter and concentrate on this measurement, along with the clipping notification on the peak meter (clipping is bad), you will be just fine.
What number should you aim for? Well, the PRSS already made this decision for us: -24 LUFS. That number happens to correspond pretty nicely to the old PRSS standard of average levels at -15 dBFS, so if you’ve previously mixed this way, you shouldn’t notice too much of a change in peak levels. You will, however, notice your mixes sounding more consistent, and your old meters may read above or below -15 dBFS.
If you remember one thing after reading this article, it should be this: mix voices to -24 LUFS on the Short Term meter. I specify voice because that is the common denominator in the majority of public radio-esque programming. It will help with consistency from one program to another if you target voices to the same place. Remember this, use good production practices, and rely on your ears. Keep the voices at -24 LUFS and the rest of the audio in your piece like music, ambi, and sound effects will fall in line around them.
A bonus: the target of -24 LUFS gives quite a bit of space between the typical speaking level and the maximum level where audio will clip (also called “head room”), so peaks aren’t too much of an issue. That said, you still need to manage your peaks, you can’t just ignore them, because clipping (red lights on peak meters) can cause audible distortion. Volume automation or use of a limiter are often great ways to solve this problem. The easiest way is to follow the PRSS standard and aim to keep peaks at or below -3 dBFS.
To make things a little easier, most meters allow you to set the target level with presets. These presets will sometimes adjust the point at which the meter will change colors which gives you a visual reference when mixing – perhaps green when the target is being hit, and red when it is too high. On most meters the presets are listed relative to the international standard they are based on. The PRSS standard of -24 LUFS happens to be compatible with the ATSC A/85 standard (the standard set by the CALM Act), so pick that preset when available. You don’t have to choose a preset, but it is an easy way to quickly customize the meter and make it easier to use.
Remember: mix voices to -24 LUFS on the Short Term meter!
Integrated Measurement – The Big Average
If you produce programming for national distribution, listen up. The Integrated measurement, described here, is the primary metric that PRSS and PRX will use to determine compliance with their new loudness standards.
Now that you know to mix voices to -24 LUFS on the Short Term meter your pieces will show improved consistency. You will be able to get phoners, narration, field tape, music — all of your elements balanced easier and quicker than you could with a peak meter alone. You might stop now, but there’s a little more to it.
The Short Term measurement focuses on a three-second window — it constantly updates the average loudness over the past three seconds. Short Term is super helpful as you mix a piece in the moment, but it can also be informative to know the average or Integrated loudness of the entire piece from start to finish. Integrated loudness is a decent predictor of consistency between pieces of audio. It’s certainly not foolproof, but if two pieces of audio have similar Integrated measurements and are produced with good production practices, odds are they will sound consistent with each other.
Geek note: Integrated measurement doesn’t allow silences or super low level (like quiet ambi) to influence its measurement. It actually excludes audio below a certain loudness so the measurement isn’t artificially skewed.
There are three methods for obtaining Integrated loudness. The easiest (yet most time-consuming) way is to play the piece in real time, start-to-finish, while measuring with a loudness meter on the master track. Most loudness meters have the ability to actively measure the Integrated loudness while audio plays through it in real time. Some plug-ins require you to start and stop the Integrated measurement, while others simply synchronize the meter with the audio editor: when audio plays in the editor the meter begins measuring.
Another way to get the Integrated loudness is with an offline, faster-than-real-time measurement. Not all audio editors and plugins allow for this, but Pro Tools does, so I’ll use it as an example. I’ll also assume that you have installed a third-party loudness plug-in. To do an offline measurement you need to mix the audio down to a single file, either through a bounce or some other method (there are many ways to do this depending on your audio editor). Import that mixed audio into a ProTools session. Under the AudioSuite menu find the loudness meter plugin and choose it. Highlight the audio file in your session, and click “Analyze” in the meter. The meter will quickly process the file and you will be presented with the Integrated loudness of the piece.
The last way is similar to the second method and requires you to have the audio mixed down to a file. You will use an external application to measure the file. You simply drag the audio file on to the program window, it analyzes the file and reports the measurements. On the Mac I like to use r128x (it’s a confusing webpage – click the “Download” link on the right-hand side to get the program).
There is yet another way to achieve the magic Integrated target loudness of -24 LUFS… but it is not without drawbacks. You’ve probably heard of peak normalization, the process that moves the highest peak in a piece of audio to a chosen level and adjusts the rest of the audio relatively. This process is based on electrical level, considering only the peaks of the waveform, not how the audio sounds. Peak levels have little to do with how we perceive the audio. Enter loudness normalization.
Loudness normalization is a similar concept, but uses perception instead of electrical peaks to normalize the audio. It measures the Integrated loudness then moves the level of the piece of audio up or down until the desired Integrated loudness is achieved. That means the entire file is gained up or down in one fell swoop.
Spotify and iTunes Radio both use loudness normalization for their streamed content — it’s why those services sound so consistent. iTunes also has a function called SoundCheck which uses loudness normalization on the music stored on your computer. If you use the free version of Spotify, you’ve probably noticed that the ads play SUPER loud. That’s because the ads aren’t loudness normalized! iTunes Radio does normalize their ads and it’s a nice experience (as far as listening to ads goes).
For radio production, you would typically use loudness normalization after mixing to make sure the final product hits the target loudness exactly. You would want to mix towards the target anyway, but the loudness normalization step afterward would put you spot on. You could also use it to normalize individual elements in your piece prior to mixing.
To be clear — loudness normalizing a final mix will not change relative levels between elements in the mix. If the mix includes a poor level shift between two interviewees or two sections in the mix, loudness normalization will not improve that, it only adjusts the level of the overall mix. The poor level shift will still be present after loudness normalization as all elements are shifted up or down equally.
Another drawback to loudness normalization involves peak management. Because this form of normalization focuses solely on the Integrated loudness of the audio, it doesn’t necessarily take into account what happens to the peaks. With lower-end tools it’s certainly possible for a file to be loudness normalized so that peaks end up clipping. To prevent this, most normalization tools include a limiter to prevent clipping. Preventing clipping is good — but sometimes the limiter can have a negative impact on the sound of the mix.
Loudness normalization is integrated into a couple of DAWs, but it’s not widely available in that form. Adobe Audition includes a rather powerful tool for this purpose, which uses a limiter to prevent clipping. It’s the most powerful tool I’ve seen integrated into a DAW in terms of customization, but that also means there are quite a few options to maneuver around.
Hindenburg takes a unique and user-friendly approach. It calls loudness normalization “Auto Level” and allows the user to apply normalization either manually or automatically to clips and recordings.
First you have to specify the target loudness, which Hindenburg only allows you to do by choosing among presets. Make sure you have upgraded to the most recent version of Hindenburg PRO. Older versions of Hindenburg PRO aren’t compatible with the PRSS standard of -24 LUFS. To set the target loudness, go to Preferences and choose the Advanced menu, then select “US” in the dropdown box. This sets the target loudness to -24 LUFS for voice and -25 LUFS (just noticeably lower) for music.
To use Auto Level manually, select a clip and then choose “Auto Level” in the Tools menu and watch the waveform level change. To enable it automatically, turn on “Perform automatic leveling after import and record” in the Preferences/Advanced menu. Now, after you import a file or make a recording, you’ll see the waveform adjust to match the target loudness. It’s pretty slick!
If you just have Hindenburg Journalist, never fear. That version automatically presets to -23 LUFS, which is the European standard. Though you will hear a difference between -23 LUFS and -24 LUFS they are so close that your mixes will still be compliant with the new PRSS spec.
There are just a few standalone loudness normalization tools available and many are pricey. The TC Electronic LCn is a recent standout, but it’s not cheap. Auphonic is a cloud-based service that offers loudness normalization and a bunch of very interesting audio processing features for a monthly subscription. Most other standalone normalization tools worth their salt are much more expensive.
If you decide to implement loudness normalization, don’t get lazy… you must continue to use your ears. Relying on loudness normalization can work, but you give up all of the benefits that loudness meters bring, such as balancing different elements within a mix. The best approach is to:
- actively mix with a loudness meter
- keep loudness normalization on hand to prep audio before a mix
- implement it only if you come into the process late and can’t make adjustments to the mix itself.
What about recording in the field?
No portable recorders currently implement loudness meters, so there is not yet a loudness solution for recording in the field. But that’s okay — you’ll want to make loudness adjustments when it comes time to mix anyway. Record at a comfortable level, and after you import the files you can actively mix with a loudness meter. If you have access to loudness normalization, you could quickly process your audio to -24 LUFS before mixing.
I LUFS Loudness! (say it out loud)
There you have it. You now have access to new loudness tools to bring consistency to your work. When presented with different audio elements in a mix — a voice recorded in the studio against a phoner, for example — you’ll be able to balance them so they sound consistent. You’ve also got a tool to help when mixing in less-than-ideal environments. If you produce national programming, you now know how to meet the new distribution guidelines using a combination of metering and loudness normalization.
As with any new tool, it may take some time for you to adjust. The meter will be the biggest change, since it looks and reacts differently than what you are probably used to. You’ll quickly find that loudness meters do indeed match up with what your ears tell you, or, if you are in a place where you aren’t able to listen carefully, what your ears can’t tell you!
If you have questions about implementing loudness in your work, please post a comment below and don’t hesitate to contact me via email or Twitter. I strongly believe loudness technology will improve the public radio system and I enjoy helping folks figure out these new methods.
Resources you might find useful:
Why You’re Doing Audio Levels Wrong, and Why It Really Does Matter – from Adam Ragusea at Current.org