The short answer
The middle mix arrives at −14 LUFS, right where Spotify, YouTube and Tidal play music. But the middle hides a wide spread: 49.6% are louder than −14 LUFS (a streaming service would turn them down) and 22.5% are quieter than −16 LUFS.
- −14 LUFSmedian integrated loudness of the mixes that arrive
- 49.6%arrive louder than −14 LUFS, Spotify's reference
- 16.4%arrive at −9 LUFS or louder, already club-loud before mastering
- 13 dBmedian peak-to-loudness ratio (PLR), the song's dynamic cushion
- 6.6%have a PLR of 8 dB or less, so they arrive already heavily limited
- 47.6%arrive as MP3 or M4A, files that are already lossy
How loud do mixes arrive?
Each column is a whole LUFS value. The highlighted column is −14, the level Spotify, YouTube and Tidal use as their reference.
Integrated loudness of the mix, in LUFS (songs per value)
3,082 mixes, 18–26 October 2026. Quieter to the left, louder to the right.
Show the numbers
| LUFS | Songs |
|---|---|
| ≤−27 | 82 |
| −26 | 20 |
| −25 | 22 |
| −24 | 26 |
| −23 | 45 |
| −22 | 50 |
| −21 | 67 |
| −20 | 76 |
| −19 | 75 |
| −18 | 107 |
| −17 | 122 |
| −16 | 232 |
| −15 | 275 |
| −14 | 354 |
| −13 | 274 |
| −12 | 313 |
| −11 | 219 |
| −10 | 218 |
| −9 | 198 |
| −8 | 137 |
| −7 | 71 |
| −6 | 51 |
| −5 | 20 |
| −4 | 10 |
| ≥−3 | 18 |
- Half of the mixes sit between −16 and −11 LUFS (the middle 50 %).
- One in ten arrives at −21 LUFS or quieter: raw mixes that have not been touched by any limiter.
- About one in six (16.4%) arrives at −9 LUFS or louder: songs somebody already pushed hard.
- 47.0% fall in the band from −16 to −12 LUFS, close enough to streaming levels that a master changes the tone more than the level.
What this means for you: a mix already at −14 LUFS has used up its headroom for streaming; mastering it louder only makes it denser, because the service will turn it back down. A mix at −20 LUFS has room, and a master will lift it. The guide on streaming loudness has the target for each platform.
- How loud to master for Spotify: what happens to louder and quieter songs
- LUFS for Spotify, Apple Music and YouTube: every platform's level
How much dynamics do they have?
PLR (peak-to-loudness ratio) is the true peak minus the integrated loudness, in dB. It is how far the loudest moment rises above the average. A heavily limited song has a low PLR; a natural, open mix has a high one.
Peak-to-loudness ratio of the mix, in dB (songs per value)
Lower is more compressed or limited; higher is more open.
Show the numbers
| PLR (dB) | Songs |
|---|---|
| ≤5 | 16 |
| 6 | 15 |
| 7 | 48 |
| 8 | 123 |
| 9 | 176 |
| 10 | 287 |
| 11 | 358 |
| 12 | 452 |
| 13 | 412 |
| 14 | 389 |
| 15 | 280 |
| 16 | 214 |
| 17 | 121 |
| 18 | 99 |
| 19 | 32 |
| 20 | 19 |
| 21 | 13 |
| ≥22 | 28 |
- The median is 13 dB; half the mixes sit between 11 and 15 dB.
- 12.3% have 9 dB or less, the territory of songs already limited for loudness; 6.6% have 8 dB or less.
- 38.8% have 14 dB or more, open mixes with room for a master to work.
A mix that arrives with a PLR under about 9 dB has little left to give a mastering tool: the loudness is already spent, and a second pass of limiting mostly adds distortion. If your PLR is that low, go back to the mix and turn the master-bus limiter off.
Phones, computers and regions
The device is the one the visitor used on OBRA, not necessarily the one the song was made on, but it separates two groups.
- −13 LUFSmedian of the 1,430 mixes opened on a phone
- −14 LUFSmedian of the 1,629 mixes opened on a computer
Mixes opened on a phone are a little louder: 56.2% are above −14 LUFS, against 43.5% on a computer. A plausible reason is that phone users more often bring a finished export or a mobile-app file that is already limited. The data cannot prove it.
| Region | Songs | Median LUFS | Louder than −14 | Median PLR (dB) |
|---|---|---|---|---|
| North America | 430 | −13 | 59.5% | 12 |
| Europe | 1,099 | −14 | 45.3% | 13 |
| Africa | 452 | −13 | 57.3% | 13 |
| Asia-Pacific | 686 | −14 | 46.5% | 12 |
| Latin America | 232 | −14 | 44.4% | 13 |
| Middle East | 63 | −14 | 49.2% | 13 |
The regions look more alike than different: medians sit at −14 or −13 LUFS in every region with a sample large enough to read, and PLR medians are 12 and 13 dB. North America and Africa come out about one LU louder than Europe and Asia-Pacific. With one-LU rounding and these sample sizes, treat that as a hint and not a ranking. The Middle East row (63 songs) is too small to read.
Which files do people bring, and what do they take away?
File format of the uploads
4,560 uploads. Every kind of file counts, including video and files the tool could not read.
WAV is the biggest group at 45.0%, but 47.6% of uploads are MP3 or M4A. A lossy file has already lost some detail, and it is the weakest starting point for a master. If you still have the project, export a WAV.
Format of the downloaded masters
1,934 downloads.
93.6% of downloads are 24-bit WAV, the default. Sixteen-bit files and FLAC together make up a few per cent.
Which of the three masters people listen to
Share of listening selections (Natural, OBRA, Loud). A person can switch many times per song, so this counts selections, not final choices.
OBRA's balanced master is selected most often (38.3%), then Natural (32.6%) and Loud (29.1%). The spread is narrow: people audition all three.
Method, and what the data cannot say
- What is counted: one measurement per song dropped into the Auto or Reference mode of OBRA Audio Mastering between 18 and 26 October 2026 (UTC days), 3,082 in total. The same song can be dropped twice in different sessions and would count twice.
- How it is measured: the browser computes integrated loudness (ITU-R BS.1770-4, K-weighted and gated) and true peak; PLR is the true peak minus the integrated loudness. Both are rounded to whole numbers before they are counted.
- How it is collected: aggregate event counts in Vercel Web Analytics, cookieless. Only the two rounded numbers, the country, the device type and the file format are kept. No audio, no file name, no account and no person are in the data. Regions with fewer than 100 songs are flagged where they appear.
- Who is in it: people who searched for free online mastering and chose to use it. That skews to producers who do their own mixing and release independently, many in countries where the tool ranks well. It is a large sample of that group, not a sample of all music.
- What it misses: visitors who block analytics are not counted, and the measurement is of the mix before mastering, not of released songs. Loudness comes from the files people have, which may already have been through a limiter. The numbers say how loud mixes arrive, not how loud they should be.
Genres are not in the data: the tool does not label them, so this page does not claim any breakdown by genre.
Use the data
You are welcome to quote any number on this page, with a link to it. A suggested citation:
OBRA (2026). How loud is independent music in 2026? 3,082 mixes measured in LUFS. https://obra.art/mastering/data/indie-loudness-2026, accessed on the day you read it.
- Download the data (CSV): every table on this page, in one file
- Measure your own mix: free LUFS and true peak meter, in the browser
- Master your song free: two songs a day, no account
Data window 2026-10-18 to 2026-10-26. Prepared on 27 October 2026.
Frequently asked questions
How loud is the average song before mastering?
In this sample the median mix is −14 LUFS integrated, the mean is −14.1, and half sit between −16 and −11 LUFS. About 49.6% are louder than −14 LUFS.
What PLR do independent mixes have?
The median peak-to-loudness ratio is 13 dB, with the middle half between 11 and 15 dB. 6.6% are at 8 dB or less, which means they are already heavily limited.
How was this measured?
In the visitor's browser, with integrated loudness per ITU-R BS.1770-4 and a true-peak reading, rounded to whole numbers and counted as anonymous aggregates. No audio leaves the device.
Can I use these numbers?
Yes. Quote them with a link to this page. The full tables are in a CSV file linked above.
Is this all independent music?
No. It is a large sample of people who use a free online mastering tool, mostly independent producers who mix their own songs. It says how loud their mixes arrive, not how loud music in general is.