Karps Music — KARPS and cyan MUSIC stacked beside an original ink-drawn listening robot with rectangular LED sensors and headphones
Skip to content
All guides

For Creators / Guide

How Loud Should Your Music Be for YouTube and Spotify?

For Spotify, the official artist guidance recommends a master at −14 integrated LUFS, with true peak below −1 dBTP, or below −2 dBTP if the master is louder than −14 integrated LUFS.12 Do not turn that into a universal YouTube or TikTok rule: we did not verify an official numeric upload target for either in the documentation searched for this guide.

Engraved audio workspace contrasting musical dynamics with controlled signal peaks

Open Karps Refine to audition its processing on your audio. An account is required. Free offers two previews per UTC day, up to 30 seconds each, rather than a full-track download. Premium or Pro is required for a full-track job, which uses one credit.

Understand LUFS, peaks and loudness matching

LUFS describes a loudness measurement, not a file format or a quality score. Integrated loudness covers the program being measured; it is different from the highest peak. FFmpeg's official loudness-processing documentation treats integrated loudness and maximum true peak as separate targets.14

True peak is a separate check on peak level, expressed as dBTP. A track can need more average level while having peaks that limit how far it can be raised. Spotify gives an example of a quiet track that cannot be lifted all the way to its normal reference because the necessary gain would leave insufficient headroom.12

For editing, the practical distinction is simple: “the loudest instant is near the limit” and “the whole song feels loud” are not the same observation. A peak-normalized export can still have quiet sustained passages. Conversely, forcing every passage toward a high average level can sacrifice the contrast you liked in the arrangement.

Do not judge from a waveform's apparent thickness. Measure the actual export with a loudness-capable tool, then listen. Keep the measurement context with the result: the full track and a chorus excerpt are not interchangeable programs.

What the platforms actually document

Platform or mode Verified official guidance How to use it
Spotify Normal Playback reference of −14 LUFS; artist mastering guidance of −14 integrated LUFS.12 A useful Spotify reference, not an absolute creative rule.
Spotify Loud / Quiet −11 LUFS and −19 LUFS playback settings respectively.12 Listener settings can change playback; your file has not acquired three different masters.
YouTube The linked help page documents Stable volume, not a numeric music-upload LUFS target.9 Do not label −14 LUFS an official YouTube delivery requirement on this evidence.
TikTok No official numeric target verified in this review. Do not copy an unsourced LUFS figure into a delivery checklist.

Spotify's page also notes that its web player and third-party devices do not use loudness normalization. Device and playback context matter; a reference level is not a promise that everyone hears the same result.12

YouTube says Stable volume reduces variation between quiet and loud parts, is not available for all videos, and is turned off for YouTube Music and official music videos.9 That feature is not the same thing as a published mastering specification. Do not assume a platform will repair your quiet intro or preserve every transition you designed.

The sensible common goal is a clean, intentional master with controlled peaks and an appropriate balance. Use platform-specific requirements where they are actually documented, and keep recommendations separate from requirements.

Why slowed + reverb edits can seem quiet or muddy

Start by identifying the complaint. Is the entire file quiet, does only the vocal disappear, or does the bass feel big on headphones but unhelpful on a phone? Those symptoms call for different decisions.

Slowing in the browser changes pitch with speed, and added reverb layers a tail over the original. Before blaming the final volume, compare the dry version. If the vocal is clear without reverb but buried with it, reduce the effect rather than raising the whole file.

Similarly, if kick and bass overlap continuously, more gain may emphasize the problem. Try a gentler speed and less reverb. Our slowed + reverb settings guide gives you a structured way to compare those choices before reaching for finishing processing.

Spotify's troubleshooting guidance cautions that tonal balance, dynamics and playback systems can affect perceived loudness. Its normalization does not make every arrangement feel identical.13 A loudness number cannot tell you whether a singer is intelligible or a bass line serves the song.

What Refine does, and what it does not

Karps Refine is a separate, authenticated upload workflow, not the public no-sign-up effect maker. It accepts MP3 or WAV sources up to eight minutes, analyzes the song, and applies a finishing chain based on the issues its analysis identifies.

The implementation checks volume, peaks, distortion, bass, tonal balance, low-end and stereo issues. Optional noise reduction is enabled by the user rather than being an automatic promise to remove everything unwanted. The chain includes corrective processing, a gain stage aimed at −14 LUFS and oversampled peak limiting.

That −14 figure is a Karps processing target, not proof of a mandatory YouTube target. Source dynamics, corrective processing and peak control affect the delivered result; do not assume every output measures exactly the same or that the service certifies compliance with every distributor.

Refine is also not a replacement for fixing an excessive effect upstream. If you intentionally buried the vocal in reverb, first make a better-balanced edit. A finishing process cannot know which artistic choice you regret.

Listen to the differenceExisting Karps Refine demonstration: before and afterUse headphones at a comfortable volume.

This pair comes from the existing Karps demonstration files. It is not a fresh measurement of your source, a promised amount of improvement, or evidence that every file needs the same processing. Focus on what you hear in the voice, bass and transient edges, not merely which player sounds louder.

Use the preview without confusing it with a full export

  1. Sign in, then open Refine.
  2. Upload your best available MP3 or WAV, not an unnecessarily re-encoded copy.
  3. Enable noise reduction only if you want to audition that treatment.
  4. Wait for analysis and processing, then compare the available result.
  5. Check the voice, bass and peaks before choosing a full-track workflow.

On Free, the preview is up to 30 seconds and starts around 30% into the song, adjusted to fit shorter files. There are two previews per UTC day. It is not a freely selectable full-length render and does not include a full-song download.

Premium and Pro process the full track for one credit per job. Keep your original locally and compare the result before replacing anything. The protected page may take a signed-out reader through authentication; that is expected, not an anonymous browser-preview promise.

If you need a numeric verification report, use an independent loudness meter on the delivered export. The Refine interface presents checks rather than a detailed numerical measurement report.

Avoid three fixes that create another problem

Over-limiting for a number

If each extra push makes drums flatter or the chorus less open, stop and compare the previous version. FFmpeg's compressor documentation explains that excessive compression can produce a dull or pumping result.14 A platform reference is not a reason to remove every contrast from a performance.

Treating clipping as extra energy

Listen for rough edges on loud syllables and kick hits. Return to a clean source where possible rather than expecting downstream correction to recreate everything that was lost. Spotify's troubleshooting page notes that encoding distortion can make a track seem louder without improving its quality.13

Solving phone translation with bass boost

Audition the actual phone speaker at a sensible level. Can you follow the melody, lyric and groove? Do not keep adding low end merely because the phone does not reproduce the same experience as headphones. Make tonal choices while checking multiple playback systems, not one device alone.

Finish with a release check, not a magic target

Listen to the exported file from start to finish. Check the loudest chorus, the quietest section and transitions where the treatment changes. Compare references at a similar listening level so louder playback does not automatically win.

Then check the audio inside the final video or distribution deliverable. A good standalone export is not proof that a later editor preserved it. Keep your approved master, settings and project separately so you can trace a problem without repeatedly reworking compressed copies.

If you are building a channel, our creator workflow guide connects production checks to rights and publishing. For finishing, try Refine's account-based preview. Aim for an edit that stays clear and intentional, not an unsupported promise to hit one number for every platform.

Sources

Sources checked October 27, 2026.

  • 9 YouTube: control video volume and Stable volume
  • 12 Spotify for Artists: loudness normalization
  • 13 Spotify for Artists: why a track can seem quieter
  • 14 FFmpeg audio filter documentation

Frequently asked questions

Does Spotify recommend -14 LUFS?

Yes. Its artist guidance recommends -14 integrated LUFS and a true peak below -1 dBTP, or below -2 dBTP for masters louder than -14 integrated LUFS. This is not a universal requirement for every platform.

Is -14 LUFS an official YouTube upload requirement?

This guide did not verify an official numeric YouTube music-upload target. Its linked help page documents Stable volume, not a universal -14 LUFS specification.

What is TikTok's official LUFS target?

We did not verify a public official numeric target in the documentation searched. This guide does not present an unofficial figure as a TikTok requirement.

Is Refine free without signing up?

No. Refine requires an account. Free accounts get two previews per UTC day, up to 30 seconds each, not a full-song download. Full-track processing requires Premium or Pro and one credit per job.