Skip to content
Elephant Audio
Loudness and delivery reference
Standard reference

EBU R 128 s2 — Loudness in Streaming

You deliver at −23 LUFS, so why does streaming say −14?

EBU R 128 s2 recommends that programmes be produced to R 128 and streamed unchanged at −23.0 LUFS, with loudness metadata indicating the actual programme loudness. Where metadata is not employed, it permits an interim Distribution Loudness Level higher than −23.0 LUFS, in the range −20.0 to −16.0 LUFS.

Source

Transcribed from the published standard, read 2026-09-23. Check it yourself

−23.0 LUFS
streamed unchanged
the recommended default
−20.0 to −16.0
interim distribution window
in LUFS, only without metadata
2023
document year
Loudness in Streaming

The question this document exists to settle

Anyone delivering both broadcast and streaming hits the same apparent contradiction. European broadcast normalises to −23.0 LUFS. Consumer streaming platforms normalise playback toward figures more than eight units louder. Both are correct, and nothing in R 128 itself explains how they relate.

s2 is the EBU's own answer, and it is not the answer most articles give. It does not say to master louder for streaming.

What it actually recommends

Transcribed from the numbered clauses, 2026-09-23:

  • d) that programmes shall be produced or prepared according to EBU R 128 and EBU Tech 3343.
  • e) that programmes should be streamed unchanged, that is at −23.0 LUFS.
  • f) that Loudness Metadata should be used, correctly indicating the actual Programme Loudness.
  • g) that, unless metadata is employed to manage device gain and the dynamic properties of content, the Distribution Loudness Level may be higher than −23.0 LUFS prior to streaming — and in that case the interim value should be in the range of −20.0 to −16.0 LUFS.

⚠️ The louder window is a fallback, not a target

Read clause g) carefully, because the ordering matters. The recommendation is to stream unchanged at −23.0 LUFS and carry metadata. The −20.0 to −16.0 LUFS window applies only where metadata is NOT being used, and the document calls that value interim.

So it is not a streaming loudness target. It is what a broadcaster may do when it cannot rely on the device to handle gain and dynamics, and wants to keep control of the result rather than leave it to whatever the player does.

Why the conflict exists at all

The document sets out two streaming situations with, in its words, conflicting requirements. One is playback with sufficiently high gain and headroom and low background noise — it names television receivers and smart speakers — where a dynamic programme reproduces faithfully at its original level.

The other is playback with lower gain and headroom, or higher background noise, and it names smartphones and personal music players. That is the listening situation that pushes distribution louder, and it is the reason a single number cannot serve both.

The practical consequence for anyone delivering to both worlds: produce once to R 128, and treat the streaming level as a distribution decision made downstream rather than a mix decision made at the master.

Sources

Common questions

Should a programme be mastered louder for streaming?

Not according to s2. It recommends producing to R 128 and streaming unchanged at −23.0 LUFS with metadata carrying the actual programme loudness. A louder distribution level is a fallback for when metadata is not used, and the document describes that value as interim.

Does the −20.0 to −16.0 LUFS window replace −23 LUFS?

No. It is the Distribution Loudness Level — what may be applied before streaming, downstream of production — and only where metadata is not employed. Production still targets −23.0 LUFS under R 128.

How does this relate to Spotify's −14 LUFS?

They are different layers. Spotify normalises playback for consumer music. s2 is the European broadcaster's position on how broadcast programmes should reach a streaming device. Neither is a mastering instruction, and neither cancels R 128.

Delivering to broadcast and streaming?

Produce once to the broadcast target and let distribution be a downstream decision. We deliver with the measurements attached so whoever handles the stream is not guessing.

Studio credit

Elephant Audio

Audio production · Sound design · Music production

LahoreEngineering sound since 2018

Last updated: