87% of the 4.2 million albums Eelco Grimm analyzed for the streaming service Tidal have a loudest track louder than -14 LUFS, the level Spotify normalizes music to (Grimm/HKU, Recommendation for Loudness Normalization by Music Streaming Services 2017). That single gap explains most of what follows: platforms turn loud masters down, broadcasters hold programmes at -23 or -24 LUFS, and regulators still take complaints about the parts of the chain that escape normalization. The FCC logged at least 1,700 loud-commercial complaints in 2024, more than double the approximately 825 of 2023 (FCC Notice of Proposed Rulemaking FCC 25-16, 2025), and California made streaming ads the next front with a law that became operative on July 1, 2026. We aggregated data from the European Broadcasting Union (EBU), the ITU, the Audio Engineering Society (AES), the FCC, the National Association of Broadcasters, Free TV Australia, ARIB, Spotify, Apple, the California Legislature, a 2026 peer-reviewed study of 120 Billboard hits, and chart measurements by HOFA College and iZotope. For the business side of the trade, see our mastering industry statistics.
A quick glossary for the numbers below: LUFS (Loudness Units relative to Full Scale) and LKFS are two names for the same ITU-R BS.1770 loudness measurement, and an LU (loudness unit) is a difference of 1 dB on that scale. “Normalization” means a player or broadcaster changes the playback gain so every programme lands at the same measured loudness, without re-mastering the file.
TL;DR
- EBU R 128 normalizes broadcast programme loudness to -23.0 LUFS, with plus or minus 1.0 LU allowed for live programmes (EBU R 128 v5.0, 2023).
- ATSC A/85 recommends -24 LKFS for US stations using a fixed dialnorm value (NAB CALM Act Summary of Best Practices 2013).
- EBU radio may distribute at -20.0 to -16.0 LUFS when it does not use -23 LUFS (EBU R 128 s3 Loudness in Radio, 2023).
- Spotify adjusts tracks to -14 LUFS, and Premium users can choose -11 LUFS (Loud) or -19 LUFS (Quiet) (Spotify for Artists, Loudness normalization).
- Apple Podcasts asks for around -16 LKFS with a tolerance of plus or minus 1 dB, measured with ITU-R BS.1770-5 (Apple Podcasts for Creators, Audio requirements).
- AES TD1008 sets -16 LUFS for music, -18 LUFS for speech and -17 LUFS for mixed streams (Radio World on AES TD1008, 2022).
- Billboard Year-End hits averaged -8.0 LUFS in 2010 and -8.8 LUFS in 2020, versus -12.5 LUFS in 1990 (Tint and Dragulin, Bulletin of the Transilvania University of Brasov 2026).
- Average dynamics (peak-to-loudness ratio) of those hits fell from 12.6 LU in 1990 to 9.3 LU in 2010 and 2020 (Tint and Dragulin 2026).
- The FCC received at least 1,700 loud-commercial complaints in 2024, up from approximately 750 in 2022 (FCC 25-16, 2025).
- California SB 576 extends the loud-ad rule to video streaming services from July 1, 2026 (California Legislature, Chapter 336, Statutes of 2025).
- ITU-R BS.1770, the measurement behind every target above, reached its sixth edition, BS.1770-5, in November 2023 (ITU-R Recommendation BS.1770).
1. Broadcast Loudness Targets and Compliance Tolerances
Broadcast loudness is a regulated or quasi-regulated number almost everywhere, and the striking part is how tight the tolerances are. EBU R 128 allows only plus or minus 0.2 LU of measurement error in quality control around its -23.0 LUFS target (EBU R 128 v5.0, 2023). A window that small is a rule for automated file checks rather than for ears: once a file is normalized, compliance is a matter of meter accuracy.
The two big regional families differ by just 1 LU: Europe at -23 LUFS and the US and Australia at -24 LKFS. The more meaningful difference is in what is measured. A/85, as summarized in the NAB CALM Act best practices, lets stations correct short-form content only when it misses the target by more than plus or minus 2 dB, and it ties the target to “dialnorm”, the dialogue-level metadata value carried in the AC-3 (Dolby Digital) stream that tells the TV how loud the speech is. Australia’s Free TV OP-59 keeps the same -24 LKFS target but asks for plus or minus 1 LU.
| Metric | Value | Source |
|---|---|---|
| EBU R 128 target programme loudness | -23.0 LUFS | EBU R 128 v5.0, 2023 |
| EBU R 128 tolerance for live programmes | plus or minus 1.0 LU | EBU R 128 v5.0, 2023 |
| EBU R 128 tolerance for measurement error in quality control | plus or minus 0.2 LU | EBU R 128 v5.0, 2023 |
| ATSC A/85 target for US stations using a fixed dialnorm value | -24 LKFS (dialnorm set to 24 in the AC-3 encoder) | NAB CALM Act Summary of Best Practices 2013 |
| Deviation from the station target that triggers correction of short-form content (US) | more than plus or minus 2 dB | NAB CALM Act Summary of Best Practices 2013 |
| Australian TV loudness target and tolerance | -24 LKFS, within plus or minus 1 LU | Free TV Australia OP-59 Issue 4, 2018 |
| EBU radio distribution level when -23 LUFS is not used | between -20.0 and -16.0 LUFS | EBU R 128 s3 Loudness in Radio, 2023 |
| Maximum loudness-to-dialogue ratio for cinematic content under R 128 s4 | no greater than 5 LU | Telos Alliance (vendor) summary of EBU R 128 s4, 2023 |
Context note: the R 128 radio supplement is the EBU’s own admission that a single -23 LUFS level does not fit every distribution channel, since it permits a higher distribution level when -23 LUFS is not used (EBU R 128 s3). The cinematic supplement, EBU R 128 s4, targets a different problem: films and drama series whose dialogue sits far below the overall programme loudness, which makes speech hard to follow at home. Most recent available data for the US and Australian practice documents: NAB 2013 and Free TV Australia 2018.
2. Streaming Platform Normalization Targets
Streaming did not adopt the broadcast number. Spotify normalizes music to -14 LUFS, 9 LU louder than the -23 LUFS broadcast target in section 1 (Spotify for Artists, Loudness normalization; -14 minus -23 equals 9). The EBU itself concedes the point: its streaming supplement notes that streaming services typically use a higher target loudness than broadcast standards (EBU R 128 s2, 2023). The Spotify loudness normalization page also shows a quieter detail: positive gain is capped by headroom, so a soft, dynamic master is not always lifted all the way to the target.
The AES position, published as TD1008, splits the difference by content type: music at -16 LUFS, speech at -18 LUFS and live mixed formats at -17 LUFS (Radio World, Streaming Audio Loudness Guidelines Explained, 2022). Speech gets the lower number because spoken word is perceived as louder at the same measured level (Telos Alliance explainer, 2025). Apple applies a similar logic to podcasts: its audio requirements page asks creators to deliver around -16 LKFS so that Sound Check does not have to change the playback level. Podcasters and live streamers who route voice through OBS face the same arithmetic in real time, which is why a loudness meter matters as much as any voice changer for OBS.
| Metric | Value | Source |
|---|---|---|
| Spotify playback levels: Normal / Loud (Premium) / Quiet (Premium) | -14 / -11 / -19 LUFS (ITU 1770) | Spotify for Artists, Loudness normalization |
| Spotify headroom kept when raising quiet tracks (lossy formats) | 1 dB; a -20 LUFS track with a -5 dB true peak is raised only to -16 LUFS | Spotify for Artists, Loudness normalization |
| Apple Podcasts delivery loudness | around -16 dB LKFS, plus or minus 1 dB (ITU-R BS.1770-5) | Apple Podcasts for Creators, Audio requirements |
| Apple Podcasts playback level for episodes with loudness or Sound Check metadata | -16 dB | Apple Podcasts for Creators, Audio requirements |
| AES TD1008 distribution loudness for track-normalized music | -16 LUFS | Radio World on AES TD1008, 2022 |
| AES TD1008 distribution loudness for speech | -18 LUFS | Radio World on AES TD1008, 2022 |
| AES TD1008 distribution loudness for mixed-format and sports streams | -17 LUFS | Radio World on AES TD1008, 2022 |
| Tidal album normalization on mobile: level of the loudest track of each album | -14 LUFS | audioXpress, 2017 |
Outlier note: normalization is not universal even inside one service. Spotify states that its web player and third-party devices such as TVs and speakers do not apply loudness normalization at all, so the same track can play at very different levels depending on the device. “True peak”, which appears in the Spotify row, is the highest level the waveform reaches between digital samples after conversion back to analog; the clipping risks it creates are a separate subject from loudness targets.
3. How Loud Commercial Music Is Mastered Compared With the Targets
The industry has standardized playback, not production. The 120 Billboard Year-End hits studied by Tint and Dragulin averaged -9.7 LUFS across 1990, 2000, 2010 and 2020, about 4.3 LU above a -14 LUFS streaming playback level (Bulletin of the Transilvania University of Brasov 2026; -9.7 minus -14 equals 4.3). The loudest period was 2010 at -8.0 LUFS, and 2020 eased only slightly to -8.8 LUFS. On services that normalize, extra loudness buys no extra playback level, yet the peer-reviewed analysis shows mastering habits barely moved.
Newer, smaller samples agree. HOFA College’s 2024 reference sample of 45 chart songs found most between -9 and -7 LUFS. In July 2025, iZotope’s chart analysis of nine top Billboard Hot Rock and Alternative songs found an average of -8.1 LUFS, so at a -14 LUFS playback level the average song is turned down by about 5.9 dB (-8.1 minus -14). The loudness is still made, then thrown away at playback.
| Metric | Value | Source |
|---|---|---|
| Mean integrated loudness of Billboard Year-End hits, 1990 / 2010 / 2020 | -12.5 / -8.0 / -8.8 LUFS | Tint and Dragulin, Bulletin of the Transilvania University of Brasov 2026 |
| Mean peak-to-loudness ratio, 1990 vs 2000 vs 2010 and 2020 | 12.6 LU vs 10.6 LU vs 9.3 LU | Tint and Dragulin 2026 |
| Mean loudness range (LRA) by decade, 1990 / 2000 / 2010 / 2020 | 6.6 / 5.5 / 5.8 / 6.0 LU | Tint and Dragulin 2026 |
| Typical integrated loudness of 45 chart songs released in 2024 | most between -9 and -7 LUFS | HOFA College, 2024 |
| Quietest and loudest of those 45 chart songs | -11.2 LUFS and -4.6 LUFS | HOFA College, 2024 |
| Average integrated loudness of 9 top Hot Rock and Alternative songs, July 2025 | -8.1 LUFS | iZotope chart analysis 2025 (vendor’s own measurements) |
| Loudest and quietest of those 9 songs | -5.1 LUFS and -11.9 LUFS | iZotope chart analysis 2025 (vendor’s own measurements) |
Context note: the peak-to-loudness ratio (PLR) is the gap between a track’s highest peak and its average loudness; a lower PLR means heavier compression. While PLR fell, the loudness range (how much loudness varies across the song) stayed between 5.5 and 6.6 LU in every decade. In plain terms, songs got denser, not flatter in structure. Sample caveat: the HOFA and iZotope samples are small (45 and 9 songs), and iZotope sells metering and mastering software.
4. Album vs Track Normalization: What 4.2 Million Albums Show
How a service normalizes matters as much as the target. In just 2% of all albums is the softest track as loud as the loudest one, so normalizing every track to the same level changes the artistic loudness balance of 98% of albums ever made (Grimm/HKU 2017). A ballad meant to sit quietly after a loud single gets pulled toward the single’s level.
The Grimm report also explains why album normalization at -14 LUFS needs little positive gain: most albums already exceed that level at their loudest track (first row of the table below). Tidal adopted the approach for its mobile apps (audioXpress, 2017), and AES TD1008 later recommended album normalization where practical. For anyone producing narrated or described audio for streaming, the same logic of keeping speech consistent across a whole programme applies; see our audio description statistics for streaming TV.
| Metric | Value | Source |
|---|---|---|
| Albums whose loudest track is louder than -14 LUFS (of 4.2 million analyzed) | 87% | Grimm/HKU 2017 |
| Albums whose softest track is louder than -14 LUFS | 50% | Grimm/HKU 2017 |
| Listeners preferring album normalization (38 subjects, shuffled 24-song playlist): overall vs blind comparison | 80% vs 71% | Grimm/HKU 2017 |
| Recommended level of the loudest album track, mobile vs stationary (living room) players | -14 LUFS vs -18 LUFS or lower | Grimm, AES Convention 147 Paper 10268, 2019 |
| Typical loudness variation between songs on popular albums | 2 to 3 LU | Radio World on AES TD1008, 2022 |
| Typical level of the 13% of albums whose loudest track is quieter than -14 LUFS (mostly classical, jazz or 1980s releases) | -15 or -16 LUFS, playing 1 to 2 dB low | audioXpress, 2017 |
Context note: in the listening test, the tracks were chosen for extreme loudness differences of up to 10 LU, so the preference for album normalization held even where the quiet songs were very quiet. Most recent available data: the 2017 Tidal catalog analysis, presented at the AES Convention in 2019.
5. Loud Commercials: Complaints and Regulation
Normalization covers the programme; ads are where viewers still notice loudness. FCC loud-commercial complaints more than doubled in a year, from approximately 825 in 2023 to at least 1,700 in 2024 (FCC 25-16, 2025; 1,700 divided by 825 is about 2.1). That jump preceded the first CALM Act rulemaking since the rules were last updated in 2014, opened as MB Docket 25-72 (FCC 25-16).
The notice also points to a design limit: A/85 governs average loudness rather than maximum loudness, which advertisers may exploit with loud openings, and the FCC asks whether simple averaging should change (Federal Register, March 11, 2025). It asked about streaming but proposed no streaming rules. California moved first: SB 576 bars video streaming services that serve California residents from sending ads louder than the program they accompany.
| Metric | Value | Source |
|---|---|---|
| FCC loud-commercial complaints, 2022 and 2023 | approximately 750 and approximately 825 | FCC 25-16, 2025 |
| FCC loud-commercial complaints, 2024 | at least 1,700 | FCC 25-16, 2025 |
| Complaints filed from the date the CALM Act rules took effect (December 13, 2012) to the FCC’s June 2013 report | 15,850 | TV Technology, reporting an FCC report, 2013 |
| Complaints filed in December 2012 alone, the first month of the rules | 4,777 | TV Technology, reporting an FCC report, 2013 |
| Licensed commercial TV stations covered by the rules (December 31, 2024) | 1,385 | FCC 25-16, 2025 |
| CALM Act rulemaking adopted / comments due / reply comments due | February 27, 2025 / April 10, 2025 / April 25, 2025 | FCC 25-16, 2025; Federal Register 2025 |
| California SB 576: approved / operative for streaming ads | October 6, 2025 / July 1, 2026 (Chapter 336) | California Legislature, SB 576 |
Context note: of the 15,850 early complaints, about 10,700 were referred to the FCC Enforcement Bureau and 5,150 were incomplete (TV Technology, 2013). SB 576 creates no private right of action, so viewers cannot sue under it directly. Loud audio is also a hearing-health issue for people already exposed to high sound levels at work; our musician hearing loss and noise exposure statistics cover that side.
6. How the Loudness Standards Evolved
Every target in this article rests on one measurement method, and it keeps changing. ITU-R BS.1770 has gone through six editions since July 2006, the latest being BS.1770-5 in November 2023 (ITU-R BS.1770). November 2023 was a busy month: the EBU also published R 128 version 5.0 and new versions of its streaming, radio and cinematic supplements (EBU R 128).
Streaming guidance moved from a wide window to fixed numbers. AES TD1004 (2015) allowed a range of -20 to no more than -16 LUFS for streaming (Current, 2015); TD1008 replaced it, its revision working group included Spotify, Apple and YouTube, and it became the AES77 standard in July 2023 (Grimm Audio, 2021, updated 2023). Japan’s broadcast guideline is now reaching beyond TV: ARIB TR-B32 version 1.7, dated October 9, 2026, adds loudness management for internet distribution.
| Metric | Value | Source |
|---|---|---|
| ITU-R BS.1770 editions, first and current | BS.1770-0 (July 2006) to BS.1770-5 (November 2023), six editions | ITU-R Recommendation BS.1770 |
| EBU R 128 first version and current version | 1.0 (2010) to 5.0 (November 2023) | EBU R 128 publication page |
| EBU R 128 relative gate change in version 2.0 (August 2011) | from -8 LU to -10 LU | EBU R 128 publication page |
| EBU R 128 s2 streaming supplement versions | 1.0 (August 2020), 2.0 (June 2021), 3.0 (November 2023) | EBU R 128 s2 publication page |
| AES streaming recommendation TD1004 range (2015) | -20 to no more than -16 LUFS | Current, 2015 |
| AES TD1008 upgraded to an official AES standard | AES77, July 2023 | Grimm Audio, 2023 update |
| ARIB TR-B32 (Japan) first version and latest version | 1.0 (March 28, 2011) to 1.7 (October 9, 2026) | ARIB TR-B32 overview page |
Context note: a “relative gate” is the rule that leaves passages far quieter than the average out of the measurement, so long soft sections do not drag the measured loudness down; moving it from -8 LU to -10 LU let more quiet material count. For how automated mastering tools respond to these targets, see our AI audio mastering adoption statistics.
Summary: Streaming Audio Loudness Normalization by the Numbers
| Metric | Value | Source |
|---|---|---|
| EBU R 128 target programme loudness | -23.0 LUFS | EBU R 128 v5.0, 2023 |
| ATSC A/85 target for US stations using a fixed dialnorm value | -24 LKFS | NAB CALM Act Summary of Best Practices 2013 |
| Australian TV loudness target and tolerance | -24 LKFS, plus or minus 1 LU | Free TV Australia OP-59, 2018 |
| EBU radio distribution level when -23 LUFS is not used | -20.0 to -16.0 LUFS | EBU R 128 s3, 2023 |
| Spotify playback levels: Normal / Loud / Quiet | -14 / -11 / -19 LUFS | Spotify for Artists |
| Apple Podcasts delivery loudness | around -16 dB LKFS, plus or minus 1 dB | Apple Podcasts for Creators |
| AES TD1008 music / speech / mixed distribution loudness | -16 / -18 / -17 LUFS | Radio World on AES TD1008, 2022 |
| Tidal album normalization: level of the loudest track on mobile | -14 LUFS | audioXpress, 2017 |
| Mean loudness of Billboard Year-End hits, 1990 vs 2010 vs 2020 | -12.5 vs -8.0 vs -8.8 LUFS | Tint and Dragulin 2026 |
| Mean peak-to-loudness ratio, 1990 vs 2010 and 2020 | 12.6 LU vs 9.3 LU | Tint and Dragulin 2026 |
| Typical loudness of 45 chart songs released in 2024 | most between -9 and -7 LUFS | HOFA College, 2024 |
| Average loudness of 9 top Hot Rock and Alternative songs, July 2025 | -8.1 LUFS | iZotope chart analysis 2025 |
| Albums whose loudest track is louder than -14 LUFS | 87% of 4.2 million | Grimm/HKU 2017 |
| FCC loud-commercial complaints, 2022 vs 2024 | approximately 750 vs at least 1,700 | FCC 25-16, 2025 |
| Complaints in the first months of the CALM Act rules (December 2012 to June 2013) | 15,850 | TV Technology, 2013 |
| California SB 576 operative date for streaming ads | July 1, 2026 | California Legislature, Chapter 336 |
| Current edition of the ITU loudness measurement standard | BS.1770-5, November 2023 | ITU-R |
| AES TD1004 streaming loudness range (2015), replaced by TD1008 | -20 to no more than -16 LUFS | Current, 2015 |
Methodology and Sources
We read every primary document or publisher page listed below in October 2026. All values are quoted as the source prints them; the four derived figures (the 9 LU gap between Spotify and EBU R 128, the 4.3 LU and 5.9 dB gaps between measured masters and -14 LUFS, and the 2.1x complaint growth) show their arithmetic in the text.
- EBU R 128 publication page and version history and EBU R 128 v5.0 PDF, European Broadcasting Union, 2023.
- EBU R 128 s2, streaming supplement, EBU R 128 s3, Loudness in Radio and EBU R 128 s4, cinematic content, EBU, 2023.
- Telos Alliance summary of EBU R 128 s4, 2023, and Telos Alliance, Understanding Loudness for Streaming Audio, 2025 (vendor publications).
- ITU-R Recommendation BS.1770 version list, International Telecommunication Union.
- NAB CALM Act Summary of Best Practices, National Association of Broadcasters, 2013.
- Free TV Australia OP-59 Issue 4, October 2018.
- ARIB TR-B32 overview and version history, Association of Radio Industries and Businesses.
- Spotify for Artists, Loudness normalization on Spotify.
- Apple Podcasts for Creators, Audio requirements.
- Radio World, Streaming Audio Loudness Guidelines Explained, May 2022 (reporting AES TD1008).
- Current, New loudness standards released for streaming content, November 2015 (reporting AES TD1004).
- Grimm Audio, AES td1008 ends the loudness war, September 2021 with July 2023 update on AES77.
- Eelco Grimm, Recommendation for Loudness Normalization by Music Streaming Services (HKU, for Tidal), 2017, and the AES Convention 147 paper announcement, 2019.
- audioXpress, Tidal Implements Album Loudness Normalization, November 2017.
- Tint and Dragulin, From Data to Hits: Understanding Mixing and Mastering Techniques that Make a Song Chart-Worthy, Bulletin of the Transilvania University of Brasov, Series VIII, 2026.
- HOFA College, Mastering for Spotify, Apple Music and more: the loudness of the pros, September 2024.
- iZotope, Hot Rock and Alternative chart analysis, July 2025 (vendor’s own measurements).
- FCC 25-16, CALM Act Notice of Proposed Rulemaking, MB Docket 25-72, February 2025, and the Federal Register notice, March 11, 2025.
- TV Technology, CALM Complaints Total 15,850, July 2013 (reporting an FCC report to Congress).
- California SB 576, Chapter 336, Statutes of 2025.
- Data watch: TV Technology’s monthly complaint counts for December 2012 to May 2013 add up to 15,693, 157 fewer than the 15,850 total it reports, so we cite only the total and the December figure. Sources disagree on when AES TD1008 appeared: Grimm Audio described it as available in September 2021, while the 2025 Telos Alliance explainer dates it to early 2022. The Brasov study does not state a mean integrated loudness for its 2000 songs, so that year is missing from our loudness comparison, and iZotope’s printed average loudness range (4.8 LU) does not match its own per-song values, so we did not use it. We could not verify an official loudness target for Apple Music Sound Check, YouTube or Netflix from documents we could read, so none appears here. The Grimm catalog data dates from 2017 and is the most recent large-scale album analysis we found. We found no final FCC order in MB Docket 25-72 as of this update, and no enforcement data yet under California SB 576, which became operative on July 1, 2026.
Last updated: October 24, 2026. We update this roundup quarterly, and the next refresh is expected when the FCC acts on MB Docket 25-72 and the first enforcement reports under California SB 576 are published.