How to normalize your audio to -14 lufs for youtube
- Step 1Get the audio track — If you have the audio stem (voiceover bounce, music mix), drop it. If you only have the finished video, the tool will read the video's audio stream — but it outputs audio, so you'll re-mux it into the video afterward, or extract first with video-to-wav.
- Step 2Select Spotify / YouTube (-14 LUFS) — Choose Spotify / YouTube (-14 LUFS) in the Loudness target dropdown —
loudnorm=I=-14:TP=-1:LRA=11, matching YouTube's reference. (Apple -16, Amazon -14/-2 TP, EBU -23 are the others.) - Step 3Pick an output format — Export WAV or FLAC to mux back into your video losslessly (best quality); MP3/M4A (192 kbps) for a reference copy. No separate bitrate control on this tool.
- Step 4Run the two-pass normalisation — Click run. Pass one measures the track; pass two corrects to -14 with the -1 dBTP limit. A typical video's audio processes in seconds to tens of seconds depending on CPU.
- Step 5Read the source loudness — The report shows the source Integrated LUFS, Range, True peak and Threshold — e.g. -9 LUFS explains a hot upload YouTube was pulling down. Note the figures.
- Step 6Mux the normalised audio back and upload — Replace the audio in your video editor with the normalised file, export the video, and upload. Check YouTube Studio's content-loudness reading after processing finishes — it should be near 0 dB (i.e. ~-14).
Preset for YouTube
YouTube and Spotify share the -14 LUFS reference, so they share a preset. Filters from lib/audio/audio-engine.ts.
| Preset (UI label) | Target | True peak | Use for |
|---|---|---|---|
| Spotify / YouTube (-14 LUFS) | -14 LUFS | -1 dBTP | YouTube videos and music releases |
| Apple Podcasts (-16 LUFS) | -16 LUFS | -1 dBTP | If the same audio is also a podcast |
| Amazon Music (-14 LUFS, -2 TP) | -14 LUFS | -2 dBTP | Amazon Music cross-posts |
| EBU R128 broadcast (-23 LUFS) | -23 LUFS | -1 dBTP | Broadcast/TV deliverables only |
How YouTube treats your upload loudness
YouTube turns loud uploads down toward -14 but does not boost quiet ones, so -14 mastering matters most for quiet creators.
| You upload at | YouTube does | Content-loudness in Studio | Fix |
|---|---|---|---|
| -9 LUFS (hot) | Turns it down ~5 dB | Shows about -5 dB | Normalise to -14 to keep your own limiting |
| -14 LUFS (on target) | Leaves it | Shows about 0 dB | Nothing — perfect |
| -20 LUFS (quiet) | Leaves it quiet | Shows about 0 dB but low | Normalise UP to -14 yourself |
| Clipping master | Plays clipped + may turn down | Negative | Re-master; normalisation can't undo clipping |
Output formats for a video workflow
Output is always audio. Choose lossless to mux back into the video without a second audio generation.
| Format | Encoder | Bitrate | Use for |
|---|---|---|---|
| WAV | pcm_s16le | uncompressed | Mux back into the video losslessly |
| FLAC | flac | lossless | Lossless, smaller than WAV |
| M4A (AAC) | aac | 192 kbps | AAC audio for the video container |
| MP3 | libmp3lame | 192 kbps | Reference copy / quick check |
Cookbook
Creator scenarios. Report figures are SOURCE measurements (the audio before correction). Mux the normalised output back into your video before upload.
Hot music-video master at -9 LUFS
A music video bounced at -9 LUFS gets turned down ~5 dB by YouTube. Normalising to -14 yourself keeps your mastering limiter in charge instead of YouTube's.
Source report: Integrated : -9.0 LUFS Range : 5.1 LU True peak : -0.2 dBTP Preset: Spotify / YouTube (-14 LUFS) -> -5 dB Result: -14 LUFS, peaks <= -1 dBTP. Export WAV, re-mux.
Quiet talking-head voiceover at -20 LUFS
YouTube won't boost a quiet upload, so a -20 LUFS voiceover stays quiet next to louder channels. Normalising up to -14 fixes it at source.
Source report: Integrated : -20.2 LUFS Range : 7.4 LU Preset: Spotify / YouTube (-14 LUFS) -> +6.2 dB Result: -14 LUFS. Check noise floor; denoise if needed: /audio-tools/ai-noise-reducer
Extract, normalise, re-mux a finished video
You only have the rendered MP4. Extract the audio, normalise to -14, and re-mux in your editor so the picture is preserved.
Step 1: video-to-wav -> extract audio losslessly
/audio-tools/video-to-wav
Step 2: loudness-normalizer, Spotify / YouTube (-14), WAV
Step 3: replace audio track in your editor, re-export MP4Dynamic cinematic edit hits dynamic mode
A trailer-style edit with whisper-quiet dialogue and explosive music has a wide LRA, triggering dynamic mode. Compress the music bus or level dialogue first.
Source report: Range = 17.5 LU (> LRA 11)
-> Dynamic-mode banner; output may sit off -14.
Step 1: audio-compressor on the mix to narrow LRA
/audio-tools/audio-compressor
Step 2: Spotify / YouTube (-14) -> clean linear -14.Consistent loudness across a YouTube series
Episodes mixed on different days drift in loudness. Normalise each to -14 so subscribers don't ride the volume between videos.
Vid 1 source: -11 LUFS -> -3 dB Vid 2 source: -17 LUFS -> +3 dB Vid 3 source: -14 LUFS -> ~0 dB All at -14 -> uniform channel loudness.
Edge cases and what actually happens
Tool outputs audio, not video
Audio onlyThis is the most common surprise: loudness-normalizer always outputs an audio file, even if you fed it a video. To publish, mux the normalised audio back into your video in your editor (or extract first with video-to-wav, normalise, then re-mux). The audio file alone won't carry the picture.
YouTube doesn't boost quiet uploads
ExpectedYouTube turns loud content down toward -14 but does not turn quiet content up. A -20 LUFS upload stays quiet relative to louder creators. That's why normalising up to -14 yourself matters — you can't rely on YouTube to lift you.
Wide loudness range triggers dynamic mode
By designCinematic edits with big quiet-to-loud swings exceed the target LRA (11 LU), so FFmpeg uses per-window dynamic correction and shows a yellow Dynamic mode banner; output may sit slightly off -14. Narrow the range first with the audio-compressor for a clean linear pass.
Report shows the source, not -14
ExpectedThe report is the source measurement (e.g. -9 or -20), which explains how YouTube was treating your upload. To confirm the output hit -14, re-drop the normalised file and read its new source reading.
Lossy audio re-encoded again
Quality lossExporting MP3/M4A re-encodes lossily; if your input was already lossy that's two generations. For a video mux, export WAV or FLAC so the audio you put back into the video isn't degraded before YouTube's own encode.
Music bed clips after the boost
SupportedA loud music bed pushed up to -14 has its inter-sample peaks limited to -1 dBTP by the loudnorm pass, so it won't clip on YouTube's encode. If you want peak control without changing loudness, use the true-peak-limiter.
Clipped source can't be restored
Cannot fixIf the audio was recorded or bounced clipped, normalising changes its level but not the distortion — the destroyed peaks are gone. Re-bounce from your project at a safer level. Normalisation prevents clipping on adjustment, but can't reverse it.
Custom -13 or -12 target wanted
Not availableOnly the four presets exist; no free-text LUFS field. -14 (Spotify/YouTube preset) is YouTube's reference, so the preset matches the real target.
Video file over the tier limit
RejectedAudio limits apply: 50 MB / 30 min (Free), 200 MB / 120 min (Pro), 100 GB / unlimited (Pro + Media / Developer). loudness-normalizer needs Pro + Media. A long video can exceed the duration cap on lower tiers; extract just the audio (smaller) with video-to-wav or upgrade.
Frequently asked questions
What LUFS should I use for YouTube?
YouTube normalises content loudness to roughly -14 LUFS. Master your video's audio to -14 LUFS integrated with a -1 dBTP true-peak ceiling — the Spotify / YouTube preset here (loudnorm=I=-14:TP=-1:LRA=11) does exactly that — and YouTube delivers it at the intended level.
Will YouTube boost my audio if it's too quiet?
No — YouTube turns loud uploads down toward -14 but does not turn quiet ones up. A quiet -20 LUFS video stays quiet next to louder channels. That's why you should normalise up to -14 yourself before uploading; you can't rely on YouTube to lift it.
This tool gave me an audio file, not a video — what do I do?
loudness-normalizer always outputs audio. Mux the normalised audio back into your video in your editor (DaVinci, Premiere, etc.). For a cleaner workflow, extract the audio first with video-to-wav, normalise to -14, then re-mux — that way the picture is untouched.
How do I check YouTube's content-loudness reading?
After your video processes, open it in YouTube Studio and view Stats for nerds on the watch page (or the content-loudness figure if shown). A value near 0 dB means it's at YouTube's -14 reference; a large negative value means you uploaded too hot and YouTube turned it down — normalising to -14 first avoids that.
Is my unpublished audio uploaded to a server?
No. Everything runs in your browser via FFmpeg 8.1 WebAssembly — processed on your CPU and offered as a download. Nothing reaches a server, so unreleased videos and audio stay private.
Should I export WAV or MP3 to put back in my video?
Export WAV or FLAC so the audio you mux into the video is lossless — YouTube re-encodes on upload anyway, so you don't want to stack a second lossy generation before that. MP3/M4A (192 kbps) is fine only for a quick reference.
Why is my channel quieter than other creators?
Almost always because your masters are quiet and YouTube doesn't boost them. Normalise each upload's audio to -14 LUFS and your videos will sit at the same level as well-mastered channels. If hiss appears after boosting a quiet track, denoise first with ai-noise-reducer.
What's the Dynamic-mode warning on my cinematic edit?
Big swings between quiet dialogue and loud music give a wide loudness range (LRA). When it exceeds 11 LU, FFmpeg switches to dynamic correction and warns you; output may sit slightly off -14. Narrow the range first with the audio-compressor for a clean linear -14 pass.
Does normalising prevent my loud music from clipping?
Yes — the loudnorm pass includes a -1 dBTP true-peak limiter, so loud music or stingers stay under the ceiling even after the loudness adjustment, surviving YouTube's lossy encode. For peak-only control without loudness change, use the true-peak-limiter.
Can I set -12 or -13 LUFS for a louder channel?
No — the tool only exposes the four presets, and -14 is YouTube's reference. Going louder than -14 just means YouTube turns it back down, so the -14 preset is the right target anyway.
Can I normalise the audio of a long video?
Audio duration limits are 30 min (Free), 120 min (Pro), unlimited (Pro + Media / Developer); loudness-normalizer needs Pro + Media. For long videos on a capped tier, extract just the audio (smaller and shorter to handle) with video-to-wav, or split the audio with audio-splitter.
How do I keep a whole series at the same loudness?
Run every episode's audio through the same Spotify / YouTube (-14) preset before muxing. Because the target is fixed, all episodes land at -14 and subscribers never have to ride the volume between your videos.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.