The Loudness War: Dynamic Range over the years

The mean dynamic range (Album DR value) per year.

The mean dynamic range (Album DR value) per year. The error bars represent the standard error.

A lot has been said about the Loudness War. According to Wikipedia it is a “pejorative term for the apparent competition to master and release recordings with increasing loudness“. Sure, loud music should be loud, but the believe among engineers that “louder sounds better” has resulted in numerous loudness war victims. In order to make the albums as loud as possible, lots of compression and limiting has to be applied to the signal, decreasing the dynamic range of the record and in the worst case introducing digital clipping. The most famous example is Death Magnetic by Metallica. This video on Youtube clearly shows the difference between the CD master (which sounds awful) and the version from the Guitar Hero game, which has a much higher dynamic range. The latter sounds much more powerful and less distorted.

There’s much more to say about this, but for further reading you could go to Ian Shephard’s website (who also initiated Dynamic Range Day), read the wise words of Bob Katz or watch this AES talk by Bob Ludwig.

There are various ways to measure loudness of songs. Currently, EBU R128 is becoming a standard. Another interesting parameter is the DR-value, which expresses the perceived Dynamic Range of a song in dB’s. The Pleasurize Music foundation offers the DR-meter, which measures these values.

In order to see how the dynamic range of albums changed over the years, I analyzed all DR-values in the DR Database. I’ve chosen to focus on the period 1985-2012. The figure above shows the mean DR-value per year. It is clear that, overall, the Dynamic Range of albums has decreased in this period from approximately 13 dB to the 8-9 dB range. It is also interesting to see that the minimum mean value is reached in 2008, the year that Death Magnetic was released. In 2009, we see an increase of +/- 0.5 dB (a double-sided t-test shows that this difference is statistically significant at the 0.01 level). Maybe this has to do with the awareness created by this album with respect to the Loudness War?

Furthermore, it seems that the Dynamic Range does not further decrease over the last four years. Has the Loudness War come to an end?

Finally, the figure below shows the number of albums per year that were analyzed. In total, 14,680 albums were included in the analysis.

The number of albums analyzed per year.

The number of albums analyzed per year.

3 thoughts on “The Loudness War: Dynamic Range over the years

  1. Wilco

    As you say Jasper, a lot of things can be said about the loudness war. In my opinion you have to listen to the music (the source) and what it needs to be mastered. With the digital techniques that we use today, you can easily break down well recorded and mixed songs due to “over-compression” and too much limiting. Just as recording and mixing engineers do, mastering engineers should also use their ears.

    Reply
  2. Ian Shepherd

    Interesting post, thanks !

    I think we need to be careful drawing conclusions form this database, though. People submitting data are interested in the issue of the loudness war, and I think there’s almost certain to be a bias towards albums with limited DR.

    There are other quirks – many albums have been submitted more than once, by multiple users – and some data relates to “fan remasters” attempting to restore crest factor to music that has been already been released. Perhaps you’ve already compensated for this, though ?

    Despite these questions, the graph does show a clear trend, and I certainly think it clearly shows that the loudest albums have decreased in crest factor dramatically over the last decade.

    It would be interesting to try and analyse this further, on the basis of genre, for example – but the work required would be formidable…!

    Thanks for helping spread the word, and thanks for the DRD plug !

    Reply
  3. Jasper Post author

    @Wilco: Well said!

    @Ian: Thank you for your comments. They are certainly valid, and are also discussed in this thread on Hydrogenaudio.org: http://www.hydrogenaudio.org/forums/index.php?s=b12f4921c971af6ba64b23a0cd4930dd&showtopic=99557

    In this thread, I also posted the following graph of the distribution for the DR-values: http://www.themindgap.nl/wp-content/uploads/2013/02/dr_dist.png

    The distribution looks OK, although that does not prove the absence of bias, obviously. The ‘multiple submission’ problem was not accounted for, and the same goes for ‘fan remasters’. Definitely interesting things to investigate in the future (although I have the feeling that the number of ‘problematic’ albums is really small compared to the total size of the database).

    Do you know about this database, by the way? Who is maintaining it?

    Reply

Leave a Reply

Your email address will not be published.

Connect with Facebook