Loudness war

The phrase loudness war (or loudness race) refers to the music industry's tendency to record, produce and broadcast music at progressively increasing levels of loudness to create a sound that stands out from others. This phenomenon can be observed in many areas of the music industry, particularly broadcasting and albums released on CD and DVD. In the case of CDs, the war stems from a desire to create CDs that sound as loud as possible or louder than CDs from competing artists or recording labels. While the maximum amplitude of a CD cannot be increased, the overall loudness can be increased by reducing the dynamic range and distorting or clipping the waveform.

Introduction

Loudness

Initially, a loudness war started between FM stations competing for listener ratings. Subsequently, record labels began increasing the loudness of their releases, both on vinyl and on CD, resulting in a volume "arms race". The main reason for this practice is that when comparing two recordings with different levels, it is likely that the louder one will be regarded as sounding better. This can be attributed to the way in which the human ear responds to sound pressure at different levels: as our ability to respond to sound frequencies change according to differences in sound pressure level, the more the SPL increases, the greater the amount of low and high frequency content we perceive. Music with higher levels is easily heard and understood in noisy environments such as: riding a car, a train, or walking on a busy city street. Higher levels can also result in subjectively better sounding recordings on low quality reproduction systems such as: web audio formats, AM radio, mono television and telephones. Compounding this loudness problem is the fact that artists are more inclined to request that their mastered CDs match the loudness level of top contemporary CDs.

Opponents

This practice has been condemned by several recording industry professionals including Grammy Award-winning mastering engineer Doug Sax, noted recording/mastering engineer Barry Diament, and many others. It is also considered unnecessary by its opponents (which also include audiophiles and hi-fi enthusiasts) who claim the same effect can be achieved simply by turning up the volume on their playback equipment. In addition, most digital audio players do not have sufficient gain at full volume setting to play older recordings or classical recordings with sufficient loudness. Finally, if a CD is broadcast by a radio station, the station will apply its own signal processing. This further reduces the dynamic range of the broadcast material to closely match levels of absolute amplitude, regardless of the original record loudness. Opponents have also called for immediate changes in the music industry regarding the level of loudness. Some music bands are being petitioned to re-release CDs with less distortion. This may indicate a general public discontent to this practice, and a call to put an end to the "Loudness War".

Effects

Waveform image of a snare drum transient from a 1990 audio recording.

Clipping resulting from above waveform boosted by 9dB to achieve average volume level of recent popular music.

Waveform processed with distortion plugin and boosted by 9dB to achieve average volume level of recent popular music.

Waveform processed with a limiter plugin and boosted by 9dB to achieve average volume level of recent popular music.

The practice of increasing a CD's loudness to be louder than competing CDs often results in distortion. In the digital realm, this is usually referred to as clipping. Digital media cannot output signals higher than digital full scale (0dB), so whenever the peak of a signal is pushed past this point, it results in the wave form becoming "clipped". When this occurs, it can sometimes produce an audible "click". However, very often certain sounds like drum hits will reach their peak for only a very short time, and if that peak is much louder than the rest of the signal, this allows an engineer to amplify the recording simply by increasing the volume. The peaks of the drum hits will be clipped, but it will not be noticeable to the casual listener. However, if clipping occurs too much in a recording, or if certain signals are noticeably clipped, it can make the recording sound distorted, making it harsh and fatiguing to listen to. How much is too much is a matter of taste, but most pop CDs, and many classical and jazz CDs have some amount of digital clipping.

Analog media, on the other hand, "bend" the signal as it exceeds its saturation point (also marked as 0dB). Such distortion can be utilized in the digital realm as well, either by transferring audio processed with tape or valve saturation to a digital recording medium, or by using computer software to emulate the effect (this process is often referred to as "bitcrunching" or "soft clipping"). This analog distortion results in harmonics that can appear to the listener as a discreet "crackle" or "fuzz" within the sound. The effect can vary depending on the sound itself, as well as the amount and kind of distortion used. Because analog distortion doesn't "flatline" to the extent that clipping does, the results are less harsh-sounding and can result in a desirable "warmth" to the recording, at the cost of slightly less transient response. The amount of distortion increases the more a signal is overdriven, ranging from transparent to highly audible, and just like digital clipping, certain instruments or musical arrangements can better mask distortion than others.

In other cases, compression or limiting is used. While the resulting distortion is lessened from the final product this way, it has the side effect of significantly reducing transient response (most often heard as lessened drum impact), and, when taken to severe levels, can reduce the natural dynamics of other instruments within the recording. Loudness increasing techniques, however, do not always affect macrodynamics (the difference in volume between sections of a song). For example, loudness-based compressors (such as limiters) will only affect the signal in a "local" manner. Multi-band compression is commonly used to make a mix more uniform and easier to balance, more compatible with low-end equipment, or to achieve a certain "sound" or artistic effect. Slow-acting, broadcast-style compression, on the other hand, will be applied to the music to make the volume in different song sections more uniform. This can make the recording more suitable for background listening or noisy environments, but can also reduce the dynamic expressiveness of the song as a whole.

History

(Note: Some of these examples are explained using RMS (Root Mean Square) power values. In reference to CD audio, these values are based on the calculation of the average of CD audio sample values with digital full scale used as a reference.)

The practice of focusing on loudness in mastering can be traced back to the introduction of the compact disc itself, but has also been said to exist when vinyl was the primary released recording medium. For example, many Motown records pushed the limits of how loud records could be made, and record labels there were "notorious for cutting some of the hottest 45s in the industry." However, because of the limitations of the vinyl format, loudness and compression on a released recording were restricted in order to make the physical medium playable — restrictions which do not exist on digital media such as CDs — and as a result, increasing loudness levels never reached the significance that they have in the CD era.

The stages of CD loudness increase are often split over the three decades of the medium's existence. Since CDs were not the primary medium for popular music until the tail end of the 1980s, there was little motivation for competitive loudness practices then. CD players were also very expensive and thus commonly exclusive to high-end systems that benefited less from higher recording levels during this period. In addition, for the first four or five years of the CD, there was absolutely no way for a recording to be compressed in the digital domain, as signal processing in the digital domain did not exist for mastering engineers.

As a result, the two common practices of mastering CDs involved either matching the highest peak of a recording at, or close to, digital full scale, or referencing digital levels along the lines of more familiar analog VU meters. When using VU meters, a certain point (usually -6 dB, or 50% of the disc's amplitude on a linear scale) was used in the same way as the saturation point (signified as 0db) of analog recording, with several dB of the CDs recording level reserved for amplitude exceeding the saturation point (often referred to as the "red zone", signified by a red bar in the meter display), because digital media cannot exceed 0db. The RMS level of the average rock song during most of the decade was around -18 dB.

At the turn of the decade CDs louder than this level began to surface, and CD volumes became more and more likely to exceed the digital limit as long as such amplification would not involve clipping more than approximately two to four digital samples, resulting in recordings where the peaks on an average rock or beat-heavy pop CD hovered near (usually in the range of -3dB) 0db but only occasionally reached it. Guns N' Roses's 1987 album Appetite for Destruction is an early example of this, with RMS levels averaging -15 dB for all the tracks.

In the early 1990s, some mastering engineers decided to take this a step further, and treat the CDs levels exactly as they would the levels of an analog tape and equate digital full scale with the analog saturation point, with the recording just loud enough so that each (or almost every) beat would peak at or over 0 dB. Though there were some early cases (such as Metallica's self-titled "Black Album" in 1991), albums mastered in this fashion generally did not appear until 1992. Alice in Chains's Dirt, Rage Against the Machine's self-titled debut and Faith No More's Angel Dust are some examples from this year. The loudness of CDs during this period varied massively depending on the philosophies of the engineer and others involved in the mastering process. 1994 was the year in which this style of "hot" mastering became commonplace, though exceptions, such as the album Superunknown by Soundgarden from the same year, still existed. The most common loudness for a rock CD in terms of RMS power was around -12 dB. Overall, most rock and pop CDs released in the 1990s followed this method to a certain extent.

The concept of making CDs "hotter" began to appeal to people within the industry, due in part to how noticeably louder CDs had become, and also in part to the notion that customers preferred louder CDs. Engineers, musicians and labels each developed their own ideas of how CDs could be made louder. While the increase in CD loudness was gradual throughout the 1990s, some opted to push the format to the limit, such as on Oasis' widely popular album (What's the Story) Morning Glory?, which reached -8 dB on many of its tracks — a rare occurrence, especially in the year it was released (1995). In 1997, Iggy Pop assisted in the remix and remaster of the 1973 album Raw Power by his former band The Stooges, creating an album which, to this day, is arguably the loudest rock CD ever recorded. It has an RMS of -4 dB in places, which is rare even by today's standards.

The standards of loudness would reach its limit in the 2000s. -10 dB has been the standard for the past several years, but this is often pushed to -9 dB, and occasionally higher. Quieter exceptions to today's standards are rare.

Interpretations

Views regarding the impact of the loudness war are heavily subjective. Proponents of louder CDs claim that consumers prefer louder CDs and that they are better for most busy listening environments. Many hold the opinion that only a handful of albums (such as the Red Hot Chili Peppers 1999 release Californication, a CD with such excessive amounts of high-frequency digital clipping that audio enthusiasts have made complaints) are examples worth considering, while other, more fanatical types believe any CD where digital full scale is frequently utilized (or when mastering processes are used solely to prevent such when attempting the same volume) should be considered unacceptable. Likewise, many listeners do not notice the effects, some consider it a minor annoyance, and others find themselves completely unable to listen to albums mastered in a loudness-based fashion.

Remasters

One Of Us, 1981 release

One Of Us, 2005 remaster

Recently many classic albums have been re-released in remastered form. In its most basic form, the current remastering practice is to boost the bass and treble frequencies (which does actually create a perceived improvement - see the Fletcher-Munson diagram) and to boost the average level with the aid of compression.

The two screenshots here, of ABBA's One Of Us, demonstrate the effect.

The first image is taken from the original Polydor CD release, where there is no clipping or distortion, and a good deal of headroom.

The second image is taken from the 2005 remaster (part of the Complete Studio Recordings box set). There is a heavy amount of compression, and many of the dynamics and "snap" of the original track have been lost as a result.

The Popular Examples section contains several other remastered albums that have been remastered in this fashion.

Other formats

At present the loudness war tends to only affect audio CDs and consequently any MP3 or other digital music files produced from them.

Recent recordings released on vinyl do not undergo the same kind of loudness-based mastering. This is partly due to technical limitations of the format and partly due to vinyl now being a niche market product favoured by a small number of hi-fi enthusiasts — similar to the CD's role in the mid-1980s.

Some SACD and DVD-Audio releases are affected as well. However nearly all DVD-Audio discs also contain a Dolby Digital (AC3) or DTS soundtrack to allow the disc to be played in a DVD-Video player that does not have DVD-Audio playback capability. Dolby Digital has a defined and calibrated reference playback level and the DTS track will also follow this. It therefore is beneficial that the high resolution DVD-Audio soundtrack will be produced at the same reference level - and this indeed is normally the case.

As these new high resolution formats are marketed largely at audiophiles, attempts to apply loudness mastering to them would almost certainly be counterproductive as the target audience is likely to be highly critical of sound quality and dynamics.

Possible solutions

• The adoption of a common alignment level and enforcement by audio standards organizations. For example, the European Broadcast Union standards call for 18dB of headroom.

• Shift all compression to the listener's end, by incorporating selectable compression and limiting settings in consumer equipment. However, this would require new listening devices must be in place before compression at source is eliminated.

• Many digital music formats now have the ability to analyse the volume of a given piece of music and either store this as metadata to allow the player to compensate on playback, or to losslessly alter the music file to match the volume to a standard level.

• The iTunes software has a feature called Sound Check, which calculates the perceived loudness of audio tracks as they are imported into the music library, and adjusts accordingly on playback. Windows Media Player has a similar feature (called simply Volume Leveling), and other formats such as FLAC calculate the perceived loudness as part of the encoding process, allowing the player to adjust the volume on a per-track basis.

File:Oneofus 1981 rg.png

One Of Us, 1981 release with Replay Gain

One Of Us, 2005 remaster with Replay Gain

• In 2001, Replay Gain was created as a new proposed standard for digital audio formats. Replay Gain scanning software calculates the perceived loudness of an individual track or album and stores the gain data in the file's tags so that a Replay Gain compliant player may adjust the playback level to a standard level. When gain is adjusted like this, "hot" compressed tracks can often sound slightly quieter than older less compressed material. The most effective demonstration of this can be performed by comparing an original CD release to a remaster. See examples.

It should be noted, however, that volume levelling (such as that used by Replay Gain) can only reduce the volume of "loudness mastered" audio so that it is not proportionately louder than the listener's other music. It cannot restore dynamics or undo clipping, so the ultimate solution may lie with a new digital media entirely.

• The new DualDisc format is a double-sided disc with CD audio on one side (usually mastered for maximum loudness) and DVD-Audio on the other. This offers a compromise: the CD audio side can be played in the car or transferred to an iPod or similar device and thereby gain some of the perceived benefits of compression, and the DVD-Audio side can be played at home to allow listeners to experience the full dynamic range of the recording.

• HDCD (High Definition Compatible Digital) is a technology that uses dynamic range compression when mastering a CD. When played back on a normal CD player the sound will be compressed and "loud". When played back on a CD player equipped with HDCD decoding, a dynamic range expander is brought into play, reversing the compression applied at mastering. Were it to be more widely adopted, HDCD would allow CDs to be made louder for the average consumer while maintaining dynamic range when played on an HDCD player. Unfortunately, however, relatively few CD players have HDCD decoding. There are also several licensing restrictions when using the technology; for example, the decoded signal may not be made available via a digital output.

Examples

• Front Line Assembly - Artificial Soldier

See also

• Programme levels
• Headroom
• Alignment level
• Audio quality measurement
• Audio noise measurement
• Loudness
• Replay Gain

External links

• EBU Recommendation R68-2000
• EBU Recommendation R117-2006 (against loudness war)
• AES Convention Paper 5538 On Levelling and Loudness Problems at Broadcast Studios
• AES: An Integrated Approach to Metering, Monitoring, and Levelling Practices (PDF)