Also, note that this article is extremely technical and the majority of music listeners will simply not care or roll their eyes at the technical mumbo jumbo audiophiles use. Except in the absolute worst cases, the issues of loudness and clipping may be largely unobtrusive unless you know what to listen for and have high-end speakers or at least decent-quality headphones. (The effects are particularly annoying on high-quality headphones, perhaps unsurprisingly; conversely, on all but perhaps the highest-end car stereos, the loudness war's effects will probably remain comparatively subtle except in the absolute worst cases.note )
It is important to distinguish between brickwall limiting
, an extreme form of dynamic range compression that prevents the amplitude from exceeding a certain threshold, and digital clipping, which actually introduces digital distortion into the signal by removing the peaks and troughs from a waveform. Another important thing to note is that compressingnote music to a lossy format like AAC or MP3 can introduce digital clipping through intersample peaks. This can introduce clipping to heavily brickwalled material that previously did not clip at all and (needless to say) makes already clipping material sound even worse, especially at lower bitrates. (MP3 is often considered to do a worse job compressing loud music than other lossy formats such as AAC and Ogg Vorbis do, which is one reason the format has a negative reputation among audiophiles.) All of these conspired to turn audiophiles away from digital audio altogether, with many of them adopting vinyl, which helped start the "Vinyl Revival" in the late '00s.
The absolute peak of loudness started slowly creeping up in 1995 when Owen Morris mastered Oasis' (What's the Story) Morning Glory? to -8 dB RMS. Since then the tendency has been to make records louder. The resulting loudness war is due to a variety of factors, such as commercial concerns, stupid executives, following the leader or listeners / musicians who are unaware of this phenomenon, can't tell the difference and don't care, actually LIKE how it sounds, or are taking advantage of it for artistic purposes, or (especially as of the 2010s) amateur production jobs from self-produced bedroom projects or small-time producers with no formal training.
One of the most egregious aspects of the industry's reliance on increasing loudness is that hundreds of albums that originally had good dynamic range are now being "digitally remastered" with almost completely brickwalled peak levels.note
By contrast, one of the sadder aspects that Nick Southall highlighted was the belief that if you master the songs loud, they'll be played more on the radio. It doesn't work like that: Radio stations (as well as TV stations) have their own compressors and equalisers to squash everything up to the same volume, with the result that any CD will get hit with this a second time for broadcast and an already hyper-compressed CD will just sound like shit squared. The existence of technologies such as ReplayGain and iTunes Radio's Sound Check additionally means that the volume of pieces played on internet radio is now frequently normalised to the same level, meaning that the only effect loudness war stunts will have on material played through these sources is making it less punchy. Mastering engineer Bob Katz' comments on this have been widely reported and discussed, and some sources believe that this will lead to the less widespread use of the practice in future recordings. (Indeed, some studies have suggested that the trend has already waned, with the average loudness of commercial releases peaking in around 2005.)
This is the main reason why people say vinyl records are "higher quality" (besides personal taste reasons such as the crackle and hum of records). The inherent quality of CDs is far better than records, but since "records are for audiophiles", there is far less incentive for audio engineers to trade-off quality for loudness on records. Additionally, vinyl records have a smaller dynamic range, which actually serves to nullify the ability to pull off loudness war stunts, even though it seems counterintuitive that this would be the case. While it's commonly believed that it's impossible to press a low-dynamic-range master to vinyl, this isn't strictly true; however, the format's limitations are of average loudness (as contrasted with digital formats, whose limitations are of peak loudness), meaning that if you want to press a low-dynamic-range master to vinyl, you will need to lower the volume to do so. If you tried to press a DR3 recording with the same loudness as it would have on CD, it would usually just throw off the needle or make the record unplayable. Since loudness war stunts will have little to no effect on a vinyl record's average loudness, this reduces the incentive to pull off such stunts; a DR10 master would stand out over a DR3 one on vinyl even more than it would on other formats, meaning that the only incentive for labels to press loudness war masters to vinyl is simply being too lazy to master the album separately (which, unfortunately, still happens fairly often). The Irony in all this, of course, is that digital formats like the CD finally made it possible to make audio as quiet as you wanted without any analog hiss obscuring it, but with a lot of equipment out there accommodating the audio levels of the War, exploiting this quality will often make things simply too quiet to hear.
For further information, Imperfect Sound Forever is required reading. More information is available online, including these two videos that do a great job of providing abbreviated explanations. There's also a 20-page forum thread dedicated to poorly mastered albums for examples. See also this, to show things are more complicated than they seem. A website allows you to use offline software to measure the dynamic range of a CD, express it as a number and add it to its ever-growing database.
There are algorithms that can be used to repair clipped audio to a certain extent; none of them are perfect, but they will generally produce end results that sound, at least to most ears, better than the commercially released versions with clipping. More info on one of them is available here. (Note that the next page of the forum thread in this link contains rather disorganised information on Adobe Audition's declipper, which is much more sophisticated and produces much better-sounding results, but does not have the benefit of being free software. iZotope RX, also not free software, has an even more sophisticated algorithm, and is fairly straightforward to use; the forum thread linked here doesn't currently go into it, but iZotope's own manual provides all the basic information about how to use it.)
(Alternatively, a quick, cheap partial solution, using a program like Nero Wave Editor, is to simply reduce the bass using the graphic equalizer, since bass-boosting is usually part of the loudness enhancement process. If done properly the bass-reduced version will sound only marginally 'thinner' than the original while having more peak fluctuations; the "Normalize" function can also be used to adjust the volume of sections of the song, although care must be taken to avoid sudden jumps in volume between sections. Another potential quick solution is to run the song through a high pass filter, which mimics the effect of pressing an album to vinyl. A CD run through a high pass filter with the right settings will be virtually indistinguishable from a vinyl rip from a comparable-sounding master and will wind up with substantially higher dynamic range scores than the original recording. A third possible solution is to use a phase rotation filter - 90 degrees is, of course, the optimal amount [180 degrees would just invert the waveform, naturally]. Note that the clipping will still be present if any of these solutions are used; it will probably, however, be substantially less annoying and fatiguing to listen to.)
One potential way to find non-loud versions of songs is that video clips posted to video services like YouTube often avoid the loudness issue as they are mastered separately. In many cases, versions of albums that are specially mastered for iTunes (which are often advertised as being such) also have more dynamic range (although frequently you can only buy these in lossy versions, which carry their own problems; fortunately, the compression algorithm used to sell iTunes music in .m4a format is very, very good, to the point where the vast majority of people will be unable to tell the difference from a lossless sourcenote ). Failing that, people will Keep Circulating the Tapes of whichever version is the least clipped or even look for places to rip masters from (full-band Rhythm Games are one source) so they can try their hand at mixing themselves. Some (though not all) file-sharing websites also allow users to share fan-produced remixes/remasters of albums (if the unclipped stems aren't available, these are usually produced with one of the above methods, most commonly a declipper). This is not strictly legal, but it is another common source of fans circulating the metaphorical tapes, as many feel these remasters to be better.
A note about the measurements often used in this article and its subcategories: Two frequently used measurements of a record's loudness are its ReplayGain score and its TT Dynamic Range Meter score (written like "DR10" or "DR5"). A large negative ReplayGain score indicates a very loud record. This isn't necessarily a guarantee that the record will be heavily compressed (nor is a smaller negative score or a positive score a guarantee that there will be no compression), but most of the time it is correlated with this trope. The Dynamic Range Meter score measures the dynamic range of the recording more directly, though CD and vinyl scores sourced from the same master can vary (vinyl scores tend to be higher due to the nature of the format). Generally, the higher the Dynamic Range Meter score, the more dynamic a recording is, though that doesn't necessarily mean it'll be free from other artefacts of the war such as clipping. (A DR5 recording might have been compressed with an algorithm that doesn't lead to clipping, for example, while a DR9 recording may still have clipped peaks). The only foolproof way to determine whether a recording is clipped is to look at the zoomed-in peaks yourself (and it is worth noting that the clipped peaks of a vinyl or cassette that was sourced from a clipped master will appear diagonal rather than horizontal due to the nature of analogue playback; many listeners have also noted that the effect sounds less harsh on analogue formats for the same reason). A database of TT Dynamic Range Meter logs can be found here, though it is not complete (feel free to send in your own logs if you like; the plug-in is available for foobar2000 here).
Ironically, the popularity of vinyl among audiophiles who have a desire for more "dynamic" sound has caused mastering engineers to include more extreme bass frequencies in modern vinyl pressings on the assumption that they'll be played on high-grade turntables with magnetic cartridges, which can cause problems with the cheap new suitcase record players, such as those sold by Crosley, with ceramic cartridges as the mass market started to return to vinyl for physical releases with the decline of the CD. The existence of these kinds of cheap record players during vinyl's heyday was the reason mastering engineers EQ'd masters in the first place. This is Older Than They Think. An early pressing of 1969's Led Zeppelin II was withdrawn for exactly this reason.
This trope might be on the wane as the record industry has noticed public backlash toward excessive loudness in recordings. The change from CD to streaming has also caused the industry to reassess loudness in digital masters, as Spotify has volume normalization turned on by default to avoid giving overly loud recordings an unfair advantage. Consequently, releases starting in the '10s have backed off extreme levels of compression in masters. Lossy audio encoders also improved over the years, and at higher bit rates are virtually indistinguishable from CDs to most listeners, so technology marching on might kill this trope. As hard drive space increased, it made sense to use the highest bit rates, which made lossily compressed files sound more natural.
