CHAPTER 9
Achieving Loudness
This chapter is about achieving loudness—ideal loudness, extreme loudness, and everywhere between. The subject is actually quite sensitive in mastering because of extreme loudness trends called the loudness wars. As a result of the loudness wars, the loudness of recordings has been pushed in a way that has seriously degraded recording quality. When mastering is performed best, recordings are set to a loudness level that is the highest possible without negatively affecting tone or dynamics. However, mastering engineers may be called on to produce loudness anywhere from ideal to extreme, so it is an important part of the mastering knowledge set.
Apparent/Perceived Loudness
The way the human ear perceives loudness is quite complex. When you view a waveform in a digital audio workstation (DAW), you are seeing the peaks and troughs of the pulse-code modulation (PCM) wave—a very poor visual representation of perceived loudness. No meter fully replicates it, although root-mean-square (RMS)/average meters set to slow response can do a somewhat decent job but are still lacking. LUFS meters are the most accurate meters for measuring loudness. The most useful LUFS meters are typically those that include a display of recent levels. In mastering, the short term (S) and loudness range (LRA) modes are the most common. However, most mastering engineers set the final levels by ear—which is always the most accurate tool.
Beyond Ideal Loudness, There Is Quality Loss
Pushing the loudness beyond its ideal level causes a loss in sound quality. Pushing it to the extreme can cause severe distortion. When higher-than-ideal loudness is demanded of a mastering engineer, everything must be done to preserve the original tonality of the recording as much as possible. It often helps to compare the original and the processed versions at the same level to get a sense of the impact of the processing.
When Did the Loudness War Begin?
Many people believe that the loudness war was something that has come along within the past two decades with the proliferation of digital recording. The fact is that the loudness war has been going on since the days of vinyl. Some engineers in pursuit of loudness on vinyl would push the levels so high that listener’s needles would physically jump out of the groove!
Why Does the Loudness War Exist?
The loudness war is mostly due to the fact that people associate higher loudness with a more intense listening experience and higher fidelity. Producers want to leverage this fact to compete, so they want their recordings to be louder than others to gain an edge.
With radio processors such as the Orban and Omnia processors, recordings mastered to be louder will not be louder compared with any other recordings when played on the radio. However, if someone is listening to a recording on an MP3 player, smart phone, or playlist, the variation in loudness will be heard and will likely affect the listener’s perception.
Future of the Loudness War
Judging from the past, we might assume that the loudness war will go on forever, as it almost always has. There is one movement that challenges this future. Apple has begun to integrate a feature called Sound Check into iTunes, iPods, and iPhones that detects the average loudness and adjusts the volume so that no matter how much limiting is applied, all recordings will play back at generally the same loudness. Mastering engineer Bob Katz has worked diligently for many years to get manufacturers to implement these types of processes. They would spell the end of the loudness war and ultimately would lead to better sounding recordings.
Loudness Potential of a Recording
Some recordings have a higher potential for loudness than others. High-quality recordings, when degraded with loudness techniques, sound better than recordings that are not as high quality to begin with. The loudness potential of a recording is something that starts with production and mixing.
Digital Clipping
Digital clipping occurs when a signal attempts to go over the highest possible level—0 dBFS. When clipping occurs, there will be two or more samples that will be at the same level. Severe clipping usually involves four or more samples at the same level. The sound of severe digital clipping is very harsh. While working to achieve loudness, perhaps the most basic goal is to avoid harsh digital clipping artifacts.
Using Clipping
Clippers, such as gClip, as well as clipping features in some limiters, will allow a type of wave shaping that can sound more natural than some types of limiting depending on the material being clipped. Usually this is used when there are only few very fast peaks, where a shaped clip might affect the sound less than the action of a limiter. This type of process has its own character.
Compression Before Limiting
When trying to achieve loudness, it sometimes can help to compress before the limiter. There is a balance that should be achieved. Often, if the master must be very loud, little compression might be used, and the limiter will be used for the majority of the increase in loudness. Using too much compression can take away from the tonality that will already be reduced by the limiter. Sometimes it can be best to use no compression at all.
Digital Limiting
Digital limiters are one of the most important tools for maximizing loudness, and each one has its own character. Limiters such as Fabfilter’s Pro-L have different character options that allow for a diverse selection of sounds. Digital limiters such as the Ozone Maximizer’s IRC III and the Voxengo Elephant have amazing transparent qualities. Digital limiters have a major advantage over analog limiters because they can have look-ahead features that allow them to quickly react to a transient, even before it has occurred.
Operating a Limiter
Limiters are simple to operate. There is almost always a gain control that is raised until the desired level of loudness is reached. Limiters also usually have a ceiling control that allows the user to set the highest-allowable peak. Limiters with a slow attack may exceed the set ceiling because the limiting action will not react quickly enough to grab fast peaks. Also, today’s limiters have dithering and oversampling built in so that separate processors will not have to be used for these features.
Sensitivity of the Ear
When maximizing loudness, it is important to consider the equal loudness contours (or their predecessor, the Fletcher-Munson curves). These indicate that the ear is most sensitive around the 3-kHz region. Therefore, boosts in this area or cuts in areas below it will raise the perceived loudness. This frequency range requires less energy than the bass or low-middle frequency ranges. As such, the loudness can be raised even higher with a limiter, without clipping, when the 3-kHz range is boosted. However, this kind of approach must be taken in a way that minimizes ear fatigue for the listener. Boosting loudness with this approach is used when the most extreme level of loudness is demanded. A very slight adjustment around 3-kHz goes a long, long way.
Clipping a High-Quality A/D Converter
A common technique to achieve loudness is to raise the gain of a signal in the analog domain before digital-to-analog (D/A) conversion. Essentially, high-quality A/D converters act in the role of a limiter when used this way. Of course, this requires a high-quality gain control to be somewhere within the signal path. Some engineers may use this only for a very light clipping effect while relying on a digital limiter afterward for the heavy lifting. In this way, if the client requests a change in loudness, it can be accomplished without the time required for analog processing. Also, some engineers never use converter clipping, especially given the quality of modern digital limiters.
Serial Limiting
This is a technique for achieving loudness in which several limiters are used at different stages, each taking a little bit off the top to achieve loudness. The idea is that several processors, with their various compression/limiting actions, can achieve loudness better than a single processor. Of course, there can be diminishing returns with the number of processors used for this approach. One prime example is to use a small amount of transparent digital limiting or compression before converting to analog, slightly compress in the analog chain and gently clip a high-quality A/D converter; and finally, use a high-quality digital limiter once back in the digital domain. At each stage of this approach, just a little more limiting/compression is applied for the purpose of achieving the most transparent loudness.
Digital Limiter Ceiling of –0.3 dBFS
Using a ceiling of –0.3 dBFS for limiting is virtually a mastering standard. This ceiling helps to prevent consumer D/A converter clipping and intersample peaks. This ceiling can be set with a limiter or maximizer. Instead, it is possible to set a digital audio workstation (DAW) master fader, track fader, object fader, or a transparent gain control after the limiter to –0.3 dB while leaving the limiter set to 0 dBFS. While this is very widely accepted, a few mastering engineers may use a ceiling of –0.1 dBFS, although there is virtually no audible difference.
NOTE In this section I am discussing –0.3 dBFS, not –3 dBFS. There is a big difference!
Maximizers/Multiband Limiters/Inflators
A maximizer is usually a limiter that involves multiband or other “smart” processing. Some processors designated as limiters could be called maximizers, so the terms aren’t definitive. For example, the Slate Digital FG-X responds to transients differently depending on frequency, but it is still called a limiter. The Waves L3 is a well-known multiband maximizer. The iZotope Ozone maximizer provides great results and is extremely transparent. Normally, a maximizer is used for the same function as a digital limiter.
Loudness can be raised in ways other than with dynamics processing with processors such as the Sonnox Inflator, Voxengo VariSaturator, and Universal Audio Precision Maximizer. The Sonnox Inflator has gained the highest popularity for this role and some see it as a “secret weapon” for loudness and thickening. When it is used, often the “curve” is adjusted with the effect at 100 percent, then the percentage is lowered to mix in the effect. However, there are many mastering engineers who do not use this type of processing. When it is used, it is almost always just before the limiter.
Ideal Loudness
Ideal loudness is set by ear and according to taste. Some engineers use a target RMS level, such as –10, –12, –14, or –20 dbFS RMS. One of the most noted proponents of ideal loudness is Bob Katz, who declared a system for guiding toward ideal levels. Essentially, it is a system where recordings are set to a standard RMS level of –12 dBFS RMS for broadcast/radio, –14 dbFS RMS for pop/rock/country, and –20 dbFS RMS for film/classical/hi-fidelity, with the peaks of each at –0.3 dbFS. His K-Meter shows RMS levels and peaks to make this easy. The K-System is discussed in more detail later in Chapter 11.
Processing While Focusing on the Loudest Passages
Working with a focus on the loudest passages in mastering can help to quickly find the optimal loudness potential of a recording.
True Peak Ceilings
When working with recordings intended for broadcast, especially television broadcast, and the broadcaster requires adherence to the Advanced Television Systems Committee (ATSC) A/85 RP, European Broadcast Union (EBU) R128, and International Telecommunication Union ITU-R.BS.1770 standards, then the true peak level must be visualized and set. There are many true peak meters available, including the Waves WLM. Limiting in a way that conforms to the true peak specifications requires limiter oversampling of 4x, and it is usually a good idea to use a limiter with intersample peak protection and a ceiling of at least –0.1 dBFS, although –0.3 dBFS is widely thought to be best.
Broadcast Loudness Standards
In some areas of broadcast, there are recommendations and legal standards on loudness. These include
• Advanced Television Systems Committee (ATSC) A/85 RP
• European Broadcast Union (EBU) R128
• International Telecommunication Union ITU-R.BS.1770
• Commercial Advertisement Loudness Mitigation (CALM) Act
In the United States, broadcast standards are mainly for television, whereas in Europe, the EBU R128 covers radio, television, and other electronic media. However, EBU R128 is not adopted into law, although a vast majority of European broadcasters adhere to it and require the recordings they broadcast to adhere to it.
ITU-R BS.1770
International Technical Union Recommendation Number BS.1770 outlines many standards for broadcast audio. It defines loudness unit (LU) and loudness unit referenced to full scale (LUFS), true peak, foreground loudness, and dialog loudness as incorporated into the EBU R128. The Dolby Media Emulator is software designed to help with this metering and other film/TV related monitoring tasks.
EBU R128
European Broadcasting Union Recommendation Number 128 (EBU R128) is a recommendation to European broadcasters regarding loudness. It largely incorporates the ITU-R BS.1770 standards for measuring loudness, including its definition of loudness unit (LU) and loudness unit relative to full scale (LUFS), true peak, foreground loudness, and dialog loudness. It recommends a standard of –23 LU/LUFS (with a deviation of –1.0 LU if exacting normalization is not practical, such as with live situations), with a maximum permitted true peak level of –1 dBTP. Also, it recommends matching the loudness of a commercial with the “foreground loudness” of its associated program. It recommends noting a recording’s maximum LU/LUFS in its digital metadata. Whereas these are the main recommendations, there are others, mostly that define how the meters for making these adjustments should be fashioned. There is also “loudness meter test material” that has been released and updated by the EBU for ensuring that meters used for this purpose are calibrated accurately.
ATSC A/85
This standard is comprised of recommended practices developed by the Advanced Television Systems Committee (ATSC). It involves many of the same topics as ITU-R BS.1770 while providing a comprehensive background of information surrounding relevant topics.
CALM Act
The Commercial Advertisement Loudness Mitigation (CALM) Act was passed by Congress and signed by President Barack Obama into U.S. law on December 2, 2010. The CALM Act adopts parts of the ATSC A/85 standard into law. This law concerns audio used for television commercials. It limits the loudness of audio used during commercials, making it in line with the loudness of its associated program. It accomplishes this by linking the commercial’s loudness with the program’s using an anchor. The anchor is usually either the dialog of the program or what is called the foreground loudness of the program.