Start a project →
Data · 2026

How loud are YouTube videos? 50 videos measured

I measured the loudness, speech pace and length of 50 recent long-form videos from 25 of the biggest channels on YouTube, then checked what YouTube's own player does with the loud and the quiet ones. The short version: big channels don't agree on a loudness, and mixing louder than about -14 LUFS buys you nothing.

By Kevin Tabares · Sep 27, 2026 · Updated · 7 min read

How loud are YouTube videos?

Across 50 recent long-form videos from 25 large channels, integrated loudness ran from -24.3 to -7.4 LUFS, with a median of -16.3. Half of the videos sat between -18.2 and -14.1. In a playback check, YouTube turned the loud ones down to -14 and left the quiet ones quieter than everyone else.

So there is no shared house standard among the biggest channels. What there is, is a player that evens part of it out.

What does YouTube do with loud and quiet videos?

YouTube's player shows what it does in "Stats for nerds", on the line "Volume / Normalized", which reports the content loudness of the audio it is playing and the target it aims for. I opened five of the measured videos on September 27, 2026 and read that line:

VideoLoudness of the file (LUFS)Player: content loudness / target
MrBeast, "Last To Leave Grocery Store"-7.4-14.0 dB / -14.0 dB
Joshua Weissman, "I Tried Fast Food That No Longer Exists"-12.3-14.0 dB / -14.0 dB
Veritasium, "Something is jamming GPS over Europe"-13.2-14.0 dB / -14.0 dB
Ali Abdaal, "What To Do With Your Life"-23.9-20.7 dB / -14.0 dB
Wendover Productions, "How the Channel Tunnel Works"-24.3-20.2 dB / -14.0 dB

The three loud videos all played at the -14 target: the extra loudness was simply taken away. The two quiet ones played at about -20, six decibels under the target, and the player did not raise them further (it showed "100%/100%"). A viewer who watches a Wendover video right after a MrBeast video hears it noticeably quieter.

Two limits on this check: it is five videos in one desktop browser session, and the player served its "DRC" audio version (the line read "100%/100% DRC"). Other devices and apps may behave differently. The pattern matched on all five, though.

What loudness should you mix to?

For a voice-led long-form video, around -14 LUFS integrated is the practical target. Louder than that gets turned down, so a harder limiter only costs you dynamics. Much quieter than that, and your video plays softer than the one before it in someone's queue. Two more things from the data:

How the music bed sits under the voice is a separate question, measured on three narrated channels in how loud background music should be under a voiceover.

How fast do top YouTubers talk?

The median across the 50 videos was about 179 words per minute, counted from each video's captions over its full length. Half of the videos fell between 164 and 191. The slowest was a Markiplier gameplay video at about 105, where long stretches are play without talk; the fastest, an Ali Abdaal video at about 291. In 43 of the 50 videos, the voice started within the first second.

Speech pace is a style, not a rule: the explainers from Kurzgesagt, RealLifeLore and Johnny Harris sat at 155-170, while tech and money channels such as Marques Brownlee, Linus Tech Tips and Graham Stephan sat at 185-202. For how the picture changes on top of that voice, see cut pacing measured on real channels.

How long are their long-form videos?

The median length was 28.4 minutes, and 33 of the 50 videos ran 20 minutes or longer, from 10.6 minutes (a Marques Brownlee follow-up) to 80.9 minutes (a DanTDM gameplay video). The measuring tool picks each channel's most-viewed recent long uploads, so this describes what is working for these channels right now, not their average.

The full table

Two videos per channel, ordered from the loudest channel to the quietest. Each value is shown for both videos, separated by a slash.

ChannelLoudness (LUFS)True peak (dBTP)Words / minLength (min)
MrBeast-7.4 / -10.4+2.8 / +3.4189 / 18243 / 37
SmarterEveryDay-13.2 / -11.8+1.1 / +3.5210 / 15749 / 47
Vox-13.0 / -13.2+1.1 / +0.7191 / 15742 / 51
Dude Perfect-13.4 / -13.0+1.4 / +2.3144 / 11519 / 17
Veritasium-13.2 / -15.4-0.9 / -0.4169 / 17734 / 48
Sidemen-13.4 / -16.1+0.9 / -0.3178 / 16851 / 46
Colin and Samir-14.3 / -15.4-1.3 / +0.6212 / 20212 / 30
penguinz0-13.1 / -16.7+3.6 / +0.4180 / 18218 / 13
jacksepticeye-14.1 / -17.3+0.6 / -0.0189 / 13132 / 53
Kurzgesagt-15.4 / -16.0-1.9 / -2.2161 / 17015 / 14
Marques Brownlee-15.9 / -16.7+0.7 / +0.4202 / 20117 / 11
RealLifeLore-16.2 / -17.2-0.6 / -0.7164 / 15539 / 30
LEMMiNO-15.9 / -17.6+0.7 / +0.7164 / 14129 / 28
Markiplier-17.4 / -16.6-1.2 / -0.2105 / 11413 / 17
Mrwhosetheboss-17.0 / -17.3+0.6 / +0.5177 / 19225 / 17
Johnny Harris-18.2 / -16.3+0.9 / +0.3158 / 16040 / 33
DanTDM-16.1 / -18.6+0.5 / +0.2172 / 18233 / 81
Linus Tech Tips-18.6 / -16.3-2.4 / -1.6191 / 18524 / 20
Joshua Weissman-23.1 / -12.3+0.1 / +2.7227 / 21825 / 33
Graham Stephan-18.1 / -17.4+0.6 / +0.8198 / 19016 / 15
Simple History-24.3 / -14.1-11.3 / -0.5166 / 16434 / 20
Grian-21.9 / -19.5-7.2 / +0.3177 / 16623 / 29
OverSimplified-20.1 / -21.5+0.2 / -0.2184 / 18930 / 34
Ali Abdaal-20.5 / -23.9-6.1 / -3.0248 / 29025 / 20
Wendover Productions-24.3 / -23.5-4.1 / -4.4182 / 18123 / 17

How was this measured, and what was left out?

For each of 25 channels, the tool took the two most-viewed long uploads among the recent ones and downloaded the audio YouTube serves. Loudness is ITU-R BS.1770 integrated loudness and true peak (the measure behind LUFS), cross-checked with FFmpeg's own loudness filter on two videos, which gave identical values to the decimal. Words per minute come from each video's captions. 44 of the 50 videos were published in 2026; LEMMiNO, OverSimplified and SmarterEveryDay upload rarely, so their picks are from 2020, 2021 and 2025. Three channels I meant to include (Casey Neistat, Binging with Babish and Gemini Tay) could not be read by the tool and are not in the data.

I also measured cut rates and faces on screen, and I am not publishing them. On gameplay footage the cut detector counted camera movement as cuts (nine "cuts" in 30 seconds of a single continuous Markiplier shot), and the face detector missed small facecams. A number I can't stand behind doesn't go on this page. The cut rates that did hold up, on narrated documentaries, are in the pacing data.

More on what changed this year on the platform: what changed on YouTube in 2026.

FAQ

What LUFS should I mix YouTube videos to?

Around -14 LUFS integrated is a practical target for voice-led long-form. In a 2026 playback check, YouTube turned louder videos down to -14, so mixing hotter bought nothing, while videos mastered near -24 still played at about -20, quieter than the rest.

How loud are popular YouTube videos?

Across 50 recent long-form videos from 25 large channels, integrated loudness ran from -24.3 to -7.4 LUFS, with a median of -16.3. A quarter were louder than -14.1 and a quarter quieter than -18.2, so there is no shared standard among big channels.

Does YouTube normalize loudness?

In our check it turned loud videos down: MrBeast, Veritasium and Joshua Weissman videos measured -7.4 to -12.3 LUFS in the file and showed a content loudness of -14 dB in the player. Quiet videos near -24 showed about -20 and were not raised further.

How fast do YouTubers talk?

The median in the 50 videos measured was about 179 words per minute, with half between 164 and 191. The slowest was a Markiplier gameplay video at about 105, the fastest an Ali Abdaal video at about 291.

Related guides

Data
How loud should background music be under a voiceover?
Data
How often should you cut in a YouTube video?
News
What changed on YouTube in 2026
Editing
Dopamine editing vs retention editing