Keeping things CALM with new loudness meters in Sound Forge Pro 11
Commercials are loud. They're often louder than the program that they play in. And over the years, they've been made intentionally louder and louder so that they'd get our attention. It's always been an annoyance, but with the advent of digital TV it became even more of a nuisance for a variety of reasons.
Shortly after the implementation of digital TV, governments began passing legislation to solve the problem. In the US, President Obama signed the Commercial Advertisement Loudness Mitigation Act—known as the CALM Act, which is the term I'll use in this article to generically refer to all similar legislation regardless of country—into law in late 2010. Of course, like any law, it's complicated, but the gist of it is that a commercial can no longer legally be louder than the programming that surrounds it.
So, what does that mean for you? Well, if you are a commercial producer, or if you have clients who come to you to create the audio for their commercial, it means you'd better be in compliance with the CALM Act, or risk the rejection of your work by the broadcaster who doesn't want to risk the penalty of noncompliance. Even if the broadcaster accepts your audio file, they'll probably run it through processing of their own—over which you'll have no control—in order to ensure that it is compliant before it airs. In fact, it looks more and more like this will be the standard for all material, not just commercials, so this touches virtually every audio editor, or at least it likely will in the not-too-distant future.
So, if you're going to deliver CALM Act compliant media, you need a tool to ensure that your work falls within the requirements. Sound Forge™ Pro 11 gives you that tool with the new Loudness Meters. In this article, we'll learn to use those meters so that you can keep your work CALM Act compliant.
To start, open a file in Sound Forge Pro 11. My file is a recording I did for the voiceover of a recent training video. When I recorded this file, I used the Sound Forge Pro Record meters to make sure my levels were in an acceptable range. After editing the file, I normalized it to a peak of 0.0 dB, monitoring my levels with the standard Sound Forge Pro Peak Meters. I chose not to do any sort of compression or limiting on this file. Figure 1 shows my file playing in Sound Forge Pro.
Notice that I'm viewing both the standard peak meter and the VU meter. In those meters, you can see the loudest peak and the maximum perceived loudness (click the picture to see a full-sized version of it). The peak meter shows that this file peaks at exactly 0.0 dB, which is the level to which I normalized. So, there is no indication of clipping and I've maximized my peak volume while avoiding clipping.
The VU meter shows a fairly high maximum perceived loudness, but that's not really as important as the average VU reading of this file which falls nicely around or below 0.
From the looks of these two meters, I've done a nice job of mastering this file to acceptable levels. You can see from the waveform that this file is not heavily compressed and thus there appears to be some variance in wave height, which translates to dynamic range. So, all good, right? Well, if I intend to deliver this file for broadcast, it might be all good. But then, it might not be. Does this file conform to the CALM Act? So far, there's no way for us to tell, and that's why we need to run this file through the new Loudness Meters.
To access the meters, choose View | Loudness Meters. Figure 2 shows the Loudness Meters in action as I'm playing the same file as I used in Figure 1. There are a lot of meters and buttons here, so let's break it down so we can understand what it all means. We'll start with the meter on the far right.
This meter, called the True Peaks meter, does essentially the same thing as the main Sound Forge Pro Peak Meter, but with one very important difference. Look back at Figure 1 and recall that the meter's highest peak value was 0.0 dB. Now look again at the True Peaks meter in Figure 2. Its highest peak value reads 0.1 dBFS and the red box around that value indicates that the file has clipped at some point. The True Peaks indicator on the left side of the window also warns you when you've clipped. Why the difference between the two meters if they're both measuring peak levels?
The main Sound Forge Peak Meter is accurate to the audio sample. But digital processing can make it entirely possible for a waveform peak to occur between audio samples—called an inter-sample peak. If this happens, the main Peak meter will not catch that peak. And if that peak happens to be above 0.0 dB, the Peak meter will miss it and you'll think all is well, when actually you have a clipping problem.
The True Peaks meter over samples the audio. That means it operates with finer granularity and analyzes the audio closer to continuous time instead of sampled time. That's why its value is indicated in dBFS—Decibels Digital Full Scale. This is a much more accurate measure of your file's peak levels. Thus the term, true peak. By comparing Figures 1 and 2, you can see that my True Peaks meter accurately identified a problematic audio peak that my main Peak meter missed.
Next, let's jump to the M and S meters. These meters both measure your file's loudness. Note that loudness is different than peak. For instance, you could have a very quiet file (with a low loudness level) that has one very loud, sudden noise in it—like a door slam—that has a much higher peak than the rest of the audio in the file. Such a file would seem quiet even though that single peak may be extremely high.
The M meter measures momentary loudness across all audio channels based on 400 millisecond chunks. Think of it as an instant snapshot of your file's loudness at an exact moment. The loudness is measured in loudness units (LU), and I'll explain more about that a little later. If you want to see where you're at right now with loudness, glance at this meter.
The Momentary field gives you a numerical readout of the same information as the meter provides graphically. So, in Figure 2, you can see that at the instant that I took the screen shot, my file's Momentary loudness was 4.2 LU.
The S meter shows you your file's short-term loudness. Short-term loudness is similar to momentary loudness except that it uses three-second chunks of audio to calculate loudness. Again, while the meter shows you this information graphically, the Short field gives you the exact information numerically. In Figure 2, my Short loudness comes in at 1.2 LU when I took the screen shot.
Both of these meters are helpful because they give you an idea of where your loudness is right now—one with an instant value, and the other with a bit more of an average value, but still quite immediate. Both meters have maximum values at the top which show you the highest value registered in what you've played so far. In Figure2, you can see that my file's highest Momentary loudness was quite high, while my highest short-term loudness seems a little more reasonable (closer to 0 LU). As you play your file, these two values change constantly, much as the True Peaks meter does. It's fine for these readings to be very high on occasion, like my highest Momentary value.
The LRA meter measures the loudness range of the momentary and short-term loudness levels. In practical terms, this gives you a graphical representation of the dynamic range in your file. This can be very helpful information. If your file has an extremely high dynamic range it may be difficult, or even annoying, for your listeners. For example, if the audio you're working on is audio for a full-length action movie and it has an extremely high dynamic range, it might mean that viewers are forced to turn their volume up high so they can hear quiet dialog, only to have their ears blown out by very loud explosions in the action scenes.
On the other hand, if you're working with a recording of a symphony orchestra, the composer has likely used dynamics almost as another instrument—the sweet harp solo shouldn't be as loud as the bombastic timpani and horn section passages. In a case like that, a very narrow dynamic range might indicate a mixing or mastering problem that you'll want to address.
Of course, in both cases, your ears are the final judge, but the visual information you get from the LRA meter can give you an important cue that you might need to pay more attention to your file's dynamics. The Loudness Range field gives an exact numeric value of your loudness. You can see in Figure 2 that my file's loudness range comes in at exactly 3.2 LU. That's a fairly compact dynamic range, but not surprising given that the file is a voiceover narration with not much volume change from one moment to the next over the course of the file.
As you play your file, the LRA value likely changes since this measurement takes in new information all the time and factors it into the overall value. If you want to measure the dynamic range of the first two minutes of your audio only, make a selection in the Sound Forge Pro data window and play just that selection. You could then note the LRA value, select the last two minutes of your file and check the values there, and then compare the two. In most cases however, you'll likely want to play your entire file and get an overall sense of its dynamic range.
This brings us around to perhaps the most important meter of all, the I meter. This meter gives you the information you need to see if you're in compliance with the CALM Act or not. It basically measures the same information as the M and S meters, but it runs its calculation based on the duration of the audio you've played. So, if you play your entire file, this meter gives you the program loudness on average of your entire file. And that's exactly the information you need in order to know whether you're in compliance.
Since this value is so critical, the I meter includes an over-target indicator next to the Integrated field (which shows you the exact numeric value of your integrated loudness). In Figure 2, you see that my file's integrated loudness comes in at 2.2 LU and that the over-target indicator is red. This shows me what my ears could not hear: my file is not CALM compliant!
The red over-target indicator shows me that I'm not CALM compliant, but how far out of compliance am I? What's my target? Here's where the numbers might first seem confusing.
The CALM Act (and remember, I'm using that term to refer to all other similar laws in other countries) places the integrated loudness target at -24.0 LUFS (European standards set this level at -23.0 LUFS). LUFS stands for Loudness Units Full Scale and it is the absolute value that's being measured. It's extremely helpful to know that 1 LUFS = 1 dB, so that should make us all feel more comfortable working with these new labels which the CALM Act directs us to use!
As we've seen, the new meters in Sound Forge Pro show most values in LU, not LUFS, so how do they help us conform to CALM? Basically, where LUFS is the absolute value, LU is a value relative to LUFS that makes the meters easier to read. In other words, the LU value shows you exactly how far off the target LUFS value you are. An LU value of 0 equals -24 LUFS (or -23 LUFS as we'll see shortly). Further, as with LUFS, 1 LU = 1 dB. So, if your file's Integrated value reads 1.0 LU, then you can easily see that you are 1 LU over the target relative value of 0 LU. And, you are also 1 LUFS over the target absolute value of -23 or -24 LUFS. So, in Figure 2, we see that the Integrated value of my file at 2.2 LU is 2.2 LUFS louder than the target value.
If you'd rather read your meters in the absolute LUFS values, right-click the meters and choose Loudness Scale | Absolute from the menu. Figure 3 shows my meters in the absolute values of LUFS (note that this screen shot may not correlate exactly to that in Figure 2 since they were taken at different locations in the file).
Compare Figures 2 and 3 and note a couple of things. First, -23 on the meters in Figure 3 correspond to the location of 0 on the meters in Figure 2. Remember, 0 LU = -23 LUFS (or -24 LUFS). Also, notice that if you do the math to figure out the difference between the Integrated value and the target value of -23 in Figure 3, you get the Integrated value of 2.2 in Figure 2. You can see then, that working in the default LU values makes it very easy to see what's going on because you're not forced to do the math! Right-click the meters and choose Loudness Scale | Absolute from the menu again to switch back to LU.
I've mentioned a couple of times that the target value is -23 or -24 LUFS. Why the ambiguity? Here's one place where different laws display their differences. The European Broadcasting Union (EBU) sets its target at -23 LUFS, while the North American Advanced Television Systems Committee (ATSC) sets a more conservative target of -24 LUFS. You can work in either mode with these meters. To change the metering mode, right-click the meters and choose either EBU R128 Mode or ATSC A 85 Mode from the menu. In both cases, 0 LU corresponds to the target value measured in LUFS. These modes also affect the maximum True peak value, which is -1.0 LUFS for EBU R 128 mode and -2.0 for ATSC A 85 mode.
Finally, you can change the scales that your meters use if you need more or less meter detail. To do so, right-click the meters and choose Loudness Scale from the menu. Then choose either EBU +9 or EBU +18 from the submenu. Choosing a wider range allows you to see low-level signals at the expense of precision display at high levels. To change the scale of your True Peaks meter, right-click the meter and choose True Peaks Range from the menu and then the scale you want from the submenu. Note that you can customize your meters further with the other options in the right-click menu.
OK; now that we understand how to read the new meters, how do we actually use them in practice? What do we do with the information we glean from these meters? Let's start with the most important information, the Integrated value which tells us whether or not we're in compliance. In my case, my Integrated value is 2.2 LU over the target, so I have a problem. To solve this problem, recall that LU and LUFS both equal dB. So, if I'm 2.2 LU over my target, I can simply lower the volume of my file by 2.2 dB and this should put me right on target. To lower the volume of your entire file by 2.2 dB, choose Process | Volume. Set the Gain slider to -2.2 (or whatever value is appropriate for your file). If you're having trouble hitting that value exactly, hold the Ctrl key while you adjust the slider for finer detailed movements. You can see my meters after I've made this adjustment in Figure 4, and my Integrated value is right on the money.
You might notice that as you play through your file, the Integrated value bounces in and out of compliance and the Over-target indicator goes on and off at various times. This is a key point to understand. In order to be in compliance, the file must register at or below the target as a whole. Out-of-compliance moments don't matter as long as the file as a whole complies. That's why it's critical for you to run the entire file through the meters and let them perform their calculations on the whole.
In my case, lowering the volume of my file not only brought me into compliance with the Integrated value, but it also solved my clipping problem as indicated by the True Peaks meter. But what if it hadn't? What if I was still peaking? I may not want to lower the volume of my file any further because there's no reason for me not to be as close to the Integrated target as possible. But I may still need to tame a clipping peak. You could handle that in a few different ways in Sound Forge Pro. For one, you could isolate the offending peak, select it in the data window, and lower the volume of just that peak until it stopped clipping. If you have several clipping peaks and don't want to handle each of them individually, you could run the file through a limiter—like the Graphic Dynamics filter set to act as a limiter—catch those high peaks, and limit them to a level that keeps them from clipping.
Finally, if you decide that your dynamic range (as indicated by the Loudness Range value) is too wide, you could use something like the Wave Hammer filter to apply compression to the file and "squeeze" the dynamics so that the portions with the highest peaks are closer in volume to the quiet portions.
Just make sure as you make any of these adjustments, that you run your file through the meters again to not only verify that you've solved the problem, but to ensure that you haven't created new ones in the process.
Beyond the new Loudness meters, there are a couple other tools in Sound Forge Pro that you can use to help conform to the CALM Act. For instance, the Statistics tool now reports the final Integrated Loudness value, the maximum Momentary and Short-term Loudness values, and the Loudness Range value. It also shows the True Peak maximum per channel and position in time. These can all be handy—especially on long files—if you don't want to play the file repeatedly to see these values. To see the Statistics tool, choose Tools | Statistics. This generates the statistics for your file and displays them in the Statistics window. Figure 5 shows the statistics for my file. Notice the True Peaks information in the list in the top half and the loudness information in the section below. If you need to send these statistics to a client or a broadcaster along with the file, click the Copy to Clipboard button and then paste them into a text document.
Finally, the Clipped Peak Detection tool can operate on True Peaks and has ASTC and EBU presets that will place markers where the file exceeds the maximum True Peak specifications. To use the Clipped Peak Detection tool, choose Tools | Detect Clipping. From the Preset drop-down list, choose the preset you want based on whether you're using the ASTC or EBU standard. Click OK.
These tools combine with the new Loudness meters to make it easy for you to comply to the CALM act, and in my opinion, the sooner we all do that, the better. Once everybody is complying to these regulations the "volume wars" will finally be over. And not only for commercials, but more importantly to many of us, also in music mastering. Up until now the main consideration when doing a mix/master project was peak level. So producers could make their records sound louder by compressing all of the dynamic range out of them and then raising the overall gain as high as possible. When you look at a waveform for a rock song and it looks like a solid bar, you're seeing a piece of music that's been drastically compressed (look at any cut off of The Red Hot Chili Peppers Californication sometime to see what I mean). Such music has little dynamics and can be fatiguing to listen to. The subtleties of the music are sacrificed at the altar of the volume king.
Conforming everything to the CALM Act would solve this problem once and for all. Since all audio would have to come in at the same Integrated loudness level, there would be no volume benefit gained by compressing the dynamics out of a recording. Then, in fact, wantonly compressing a music file will be something mastering engineers will avoid since most music is simply better with its natural dynamics intact. Compression will become an artistic tool as opposed to a blunt-force volume raiser.
One day we may get to that magical place. But for now, regardless of whether we ever get there for all audio, the CALM Act has at least brought us there for commercials headed to broadcast, and that's a good start. If you need to make sure that your audio files comply with the CALM Act, the new tools and meters in Sound Forge Pro are going to be lifesavers.
For more training resources, including free article archives, videos, webinar archives, and more, visit the newly organized training section at www.sonycreativesoftware.com/training. And finally, for more information on the CALM Act and loudness metering, check out the document, Practical guidelines for Production and Implementation in accordance with EBU R 128 which you'll find at https://tech.ebu.ch/docs/tech/tech3343.pdf. I pulled heavily from that document for this article and it does a great job of explaining these concepts in as much detail as you care to dive into!
Gary Rebholz, is the training manager for Sony Creative Software. Gary produces the popular Seminar Series training packages for ACID Pro, Sound Forge Pro, SpectraLayers Pro, and Vegas Pro software. He is also co-author of the book Digital Video and Audio Production. Gary has conducted many live training webinars as well as countless hands-on classes in the Sony Creative Software training center and at tradeshows such as the National Association of Broadcasters show.