My statement paper: The loudness war Introduction ! Over the last couple of decades an interesting phenomenon has introduced itself into music. Ever since the cd was introduced songs started to get louder & louder (Friedemann Tischmeyer). Even before the advent of the cd, it was observed that louder-cut vinyl records were picked more often in the jukebox, but fortunately there’s a limit to how loud a vinyl can be cut without making it unplayable, so even the loudest-cut records managed to retain quite reasonable dynamics. Unfortunately, digital recording removed such constraints — a CD, for example, is playable regardless of the amplitude of the encoded audio data — and that simple technical freedom facilitated the 'war' that has been raging with (arguably) ever more musically destructive power over the last 30 years (Emmanuel Deruty). This is why vinyl is preferred by so many audiophiles, including the writers of this text. Motives & Background The loudness war or loudness race is a pejorative term for the apparent competition to master and release recordings with increasing loudness (Stanstudio). The first occurrence of this phenomenon was reported with 7 inch single recordings. At first it seems that louder music sounds better, because of the nature of human hearing. However, it is believed that increasing the loudness of a song has a major downside, i.e. the dynamics of the recording will decrease. The loudness war started with the idea that a song that is mixed louder will stand out from the rest and thus will gain more attention. When comparing two recordings with different levels, listeners are likely to rate the louder one better sounding. One way to explain this is the psycho-physiological effect illustrated by the Fletcher-Munson (Phon) curves. These curves describe how the human hearing system responds to different sound pressure levels. They are also called equal-loudness contours. In order to characterize this frequency dependency, some psycho-acoustical experiments have been conducted. In these experiments sinusoidal sounds were presented to a large group of (normally-hearing) young adults. First, a sinusoidal sound was presented to the listeners and it was determined for which intensity level the sound was just audible. The experiment was repeated for a large number of frequencies in the audible range, in this way the Fletcher-munson, as shown in figure 1, curves were determined (Geurts L.). Music with a higher average level is also more easily heard and understood in noisy environments such as in cars, trains, etc… On lower quality sound systems, it may also sound better because of its higher signal to noise ratio. This compensates the more limited distortion-free dynamic range of those systems (Stanstudio). This basically means that it lies in the ear's tendency to hear more bass and high frequencies at higher loudness levels, thus making the song sound more lively and punchy. ! Figure 1 Fletcher-munson curves ! Once a maximum amplitude of a recording is reached, loudness can digitally still be increased even further through signal processing techniques. At a certain loudness, increasing introduces a trade-off: if loudness is increased even further, the quality of the recording will suffer. A lot of albums have already fell victim to the loudness war, the most famous example is probably Metallica’s Death Magnetic album, which is mastered incredibly loud with peaks beyond the point of digital clipping. Later, a version was released through the Guitar Hero game that didn’t have that extreme dynamic range compression. The difference in waveform is enormous, as seen in figure 2. Examples of this can easily be found on Youtube. ! ! Figure 2 Metallica - The Day That Never Comes CD version vs. Guitar Hero version But luckily there’s also a counter movement coming from some very influential artists, e.g. on the introduction of Jack White’s latest album on vinyl, ‘Lazaretto’, he says “Bob Ludwig, who mastered this album, mastered the vinyl without any compression, it’s analog to analog, mastered tape to tape without any use of compression while the digital version of the album does contain standard compression”. It is clear from his intonation that this is a big advantage of the vinyl album. Bob Dylan has called a lot of modern records “atrocious”, “they lack definition of vocals”, etc… And Daft Punk have also made it clear that their latest, Grammy Award winning album ‘Random Access Memories’ made use of as little dynamic range processing as possible. This album has been unanimously perceived as a sonically great album. But it is important to ask if the loudness has really increased and thereby led to a reduction in the dynamics of records over the past years. Research Loudness One of the main questions concerning the loudness war is: has the loudness of music increased over the years? Yes, recordings are getting louder, there is enough evidence to proof this. Figure 3 displays the RMS volume values (RMS means root mean square, which basically is a kind of mean value) of 4500 records between 1970 and 2010. The y-axis is expressed in dB. ! Figure 3 RMS values of numerous records between 1970 and 2010 Another example of this can be seen in figure 4, this picture displays the loudness of the top 5000 songs from 1950 up to 2010. ! Figure 4 Loudness of 5000 songs from 1950 to 2010 On average today’s recordings are roughly 5dB louder than they were in the seventies (Emmanuel Deruty). Another way of measuring the loudness is looking at the crest factor. The crest factor is the ratio between the peak and RMS values, it indicates how extreme the peaks are in a signal(Yokogawa) and is basically the unit to express dynamic range. Generally speaking, more compression means a lower crest factor and the lower the crest factor, the louder the recording is on average. Compressing a signal does not increase the amplitude of the peaks, it decreases the amplitude of the peaks, thus decreasing dynamics. The volume of the entire signal can thus be increased without clipping. Figure 5 displays the crest factor of 4500 records between 1970 and 2010. ! Figure 5 Crest factor of numerous records between 1970 and 2010 It is clearly visible that the crest factor did heavily decrease the past 20 years, so we can conclude that music compression has increased a lot. Dynamics What are dynamics Something that is dynamic is constantly changing. Dynamic range is the ratio between the largest and smallest possible values of a changeable quantity e.g. sound(Digital audio essentials). In analog audio the dynamic range is the difference between the low-level thermal noise and the high-level signal saturation resulting in increased distortion and, if pushed higher, clipping. In music, dynamic range is the difference between the quietest and loudest volume of an instrument, part or piece of music(Bruce Fries, Marty Fries). Another way of describing the dynamic range is how variable the mobile music level is. It can be proven that comparing a record’s RMS value during a certain window period is not a reliable method to measure the dynamic range. E.g. an isolated peak in an otherwise flat RMS curve would distort the measure, giving a false impression of significant RMS mobility(Emmanuel Deruty). Instead of using the RMS values, the distribution/variability of the RMS values can be computed. In this way the ‘spread’ of the RMS is measured. This has to be done with certain window lengths. This method only proves to be useful if the utilised window length is sufficient. Following picture compares the RMS variability to the window length, the used songs are ‘smells like teen spirit’ by Nirvana and ‘fuk’ by Plastikman. Fuk is a minimalistic techno track and thus is expected to have a lower dynamic range than smells like teen spirit. This is only visible after taking a sufficient window length as can be seen on figure 6. ! Figure 6 Comparison RMS variability vs. window length between Fuk & Smells Like Teen Spirit ! Has the dynamic range decreased? The amount of compression and limiting used in mastering drastically increased between 1990 and 2000(Emmanuel Deruty). There is no denying that there is a trend in using more compression and thus having less dynamic compression. However some studies also indicate that this is not as big as a problem we might believe. In ‘Perceptual Effects of Dynamic Range Compression in Popular Music Recordings’ the writers state the following: The belief that the use of dynamic range compression in music mastering deteriorates sound quality needs to be formally tested. In this study normal hearing listeners were asked to evaluate popular music recordings in original versions and in remastered versions with higher levels of dynamic range compression. Surprisingly, the results failed to reveal any evidence of the effects of dynamic range compression on subjective preference or perceived depth cues. Perceptual data suggest that listeners are less sensitive than commonly believed to even high levels of compression. A signal processing specialist, Dr Damian Tardieu, has also done a study in which he finds that it is in fact not such a problem as others might believe. However, they both agree that effects like ear fatigue have become apparent together with this trend. Which we find already a good enough argument to prevent this. We find it hard to imagine that it is desired that one cannot listen to an entire album without discomfort. A database exists where the dynamic range of albums can be found, http://dr.loudness- war.info/. By walking up to the amplifier and raising the volume, the dynamic range is not affected and the crest factor still remains the same.