Audacity LUFS Export Disagrees with Youlean After Sample Rate Changed Mid-Chain - How to Fix

Problem: You normalized a fest trailer in Audacity (Effect → Loudness Normalization). The built-in meter or post-effect readout suggests −18 LUFS integrated. You open the exported MP3 in Youlean and see −14 LUFS—a 3+ LU gap. Marketing asks which number is “true” before Steam upload.

Who is affected now: Indies following the fourteen-tool LUFS listicle in May–June 2026—bouncing between Tool 2 (Audacity) and Tool 1 (Youlean) without a written measurement chain. The gap is usually not broken meters; it is different files, sample rates, or double loudness passes.

Fastest safe fix: Set Audacity project rate to your delivery spec (usually 48 kHz for video trailers) → mix and normalize once on the timeline → export 32-bit float WAV for the authoritative Youlean pass → encode MP3/AAC delivery only after receipt passes → log loudness_receipt_v1.json with measurement_file, project_rate_hz, and all three readings within ±0.5 LU.

Direct answer

Audacity and Youlean both implement BS.1770-style loudness, but they rarely measure the same bytes. Resampling on export (48 kHz project → 44.1 kHz MP3), lossy re-open, mono downmix, or Normalize twice (Audacity then Youlean) produces 3+ LU drift that looks like tool bugs. Pick one master WAV and one authority meter (Youlean or ffmpeg loudnorm JSON)—use Audacity for editing, not competing final numbers.

Why this issue spikes in May 2026

Listicle traffic — Teams run Tool 1 and Tool 2 in parallel without LOUDNESS_TARGETS.md.
Agency handoffs — Trailers arrive at 44.1 kHz while your game mix is 48 kHz.
MP3 for “quick check” — Youlean on lossy files vs Audacity on WAV before encode.
October fest locks — Wednesday smoke expects receipt-backed audio.
Measurement-chain bookmark list — 14 Free trailer loudness measurement chain agreement tools (resource) locks one WAV + authority meter; this help is the fix when numbers already disagree.

Pair with ffmpeg loudnorm channel-layout help when ffmpeg also disagrees (layout issue, not sample rate), 14 Free fest trailer LUFS tools (resource), and the measurement-chain agreement list.

Symptoms and search phrases

Audacity −18 LUFS; Youlean −14 LUFS “same” trailer.
Numbers change when you re-import MP3 into Audacity.
Project rate 48000 Hz; export dialog shows 44100 Hz.
Stereo timeline; Youlean reports mono integrated (collapsed).
Ran Normalize in Audacity, then Normalize again in Youlean.
ACX −3 dB peak headroom applied twice.
“Which LUFS do we put on the upload ticket?”

Root causes (check in order)

Export resample — project 48 kHz → delivery 44.1 kHz without dither policy documented.
Measuring MP3/AAC in Youlean while Audacity meter ran on pre-encode WAV.
Double loudness pass — Audacity Normalize + Youlean target adjustment.
Mono/stereo mismatch — export collapsed to mono; Youlean still shows stereo file differently.
Selection vs whole file — Audacity selection LUFS ≠ full mix integrated.
Clipped master — meters disagree on true-peak handling after clip.
Old Audacity build — Loudness Normalization options differ from 3.x docs online.

Beginner path (45 minutes)

Prerequisites: Audacity 3.x+, Youlean Loudness Meter (free), one trailer mix.

Open LOUDNESS_TARGETS.md from the listicle—pick trailer band (e.g. −16 to −14 LUFS integrated).
In Audacity: Edit → Preferences → Quality — set Default sample rate to 48000 (or your locked delivery rate).
Tracks → Resample only if source file rate differs—once, before edits.
Apply one loudness pass (Audacity Normalize or plan to use Youlean/ffmpeg later—not both).
Export WAV (32-bit float) → measure only that file in Youlean.
Fill receipt JSON—if within ±0.5 LU, encode MP3 for review links.

Fastest safe fix path

Step 1 — Lock project sample rate before editing

Tracks → Resample… if imported WAV is not at target rate.
Edit → Preferences → Quality → Default Sample Rate: 48000 Hz (fest video common).
Confirm status bar shows Project Rate: 48000 Hz before Normalize.

Outbound: Audacity Manual — Quality Preferences.

Step 2 — Single loudness authority (choose one lane)

Lane	Edit in	Measure in	Encode last
A — Youlean authority	Audacity (level only, no Normalize)	Youlean on WAV	ffmpeg or Audacity MP3
B — Audacity authority	Audacity Normalize to target	Re-open same exported WAV in Youlean for verify only	MP3 after verify
C — ffmpeg CI	Any DAW	ffmpeg two-pass `loudnorm` JSON	See channel-layout help

Pro tip: Lane A avoids fighting Audacity’s meter vs Youlean—use Audacity for cuts and fades; Youlean owns the number on the WAV master.

Step 3 — Export WAV master for measurement

File → Export → Export Audio…

Format: WAV (Microsoft)
Encoding: 32-bit float (or 24-bit PCM if size matters)
Do not enable a second resample to 44.1 kHz unless delivery spec requires it
Filename: trailer_master_48k.wav

Never measure the MP3 for receipt pass/fail.

Step 4 — Verify in Youlean and optional ffmpeg cross-check

Drag trailer_master_48k.wav into Youlean.
Note Integrated LUFS, LRA, True Peak.
Optional CI cross-check:

ffmpeg -i trailer_master_48k.wav -af loudnorm=I=-14:TP=-1:LRA=11:print_format=summary -f null -

If ffmpeg and Youlean differ by > 0.5 LU, run ffmpeg channel-layout help before blaming sample rate.

Step 5 — Encode delivery formats after pass

Only after receipt promotion_allowed: true:

ffmpeg -i trailer_master_48k.wav -c:a libmp3lame -b:a 192k trailer_review.mp3

Re-open MP3 in Youlean for info only—expect small drift from lossy coding; receipt should cite WAV numbers.

Step 6 — Log loudness_receipt_v1.json

{
  "schema": "loudness_receipt_v1",
  "asset": "fest_trailer_stereo_v3",
  "measurement_file": "trailer_master_48k.wav",
  "project_rate_hz": 48000,
  "export_delivery_rate_hz": 48000,
  "measurement_chain": [
    { "tool": "Audacity", "role": "edit_only", "normalize_applied": false },
    { "tool": "Youlean", "role": "authority_meter", "integrated_lufs": -14.2 },
    { "tool": "ffmpeg_loudnorm", "role": "ci_cross_check", "integrated_lufs": -14.0 }
  ],
  "target_band_lufs": { "min": -16, "max": -14 },
  "agreement_lu": 0.2,
  "promotion_allowed": true
}

Attach to BUILD_RECEIPT audio row and Wednesday smoke.

Working dev path (proof table)

Check	Pass signal
Project rate locked	Status bar + receipt `project_rate_hz`
Single normalize pass	`normalize_applied` true in one chain row only
Authority file	`measurement_file` ends in `.wav`
Meter agreement	`agreement_lu` ≤ 0.5
Lossy drift documented	MP3 reading optional footnote, not gate
Targets documented	`LOUDNESS_TARGETS.md` committed

Verification checklist

[ ] Audacity project rate matches export_delivery_rate_hz in receipt.
[ ] Youlean integrated LUFS within signed target band.
[ ] ffmpeg summary within ±0.5 LU of Youlean on same WAV.
[ ] No MP3 used for promotion gate.
[ ] Trailer upload ticket cites WAV LUFS + receipt hash.
[ ] Marketing review MP3 labeled “lossy preview—not master meter.”

Prevention

Commit LOUDNESS_TARGETS.md with rate + LUFS bands—send to composers.
Ban “quick MP3 loudness check” for pass/fail in team Discord.
Template export preset: 48 kHz WAV master only.
One row in receipt: authority_meter": "Youlean" (or ffmpeg JSON in CI).
Pair with trailer frame audit before October lock.

Troubleshooting

Symptom	Fix
4+ LU gap	Different files (WAV vs MP3) or double Normalize
Youlean mono / Audacity stereo	Check export channels; avoid force mono
Drift only on MP3	Expected—gate on WAV
Audacity −23, Youlean −14	Selection meter vs full mix
Worse after Resample	Wrong rate chosen—re-import source
ffmpeg −inf	Channel layout—other help article

FAQ

Is Audacity wrong and Youlean right?
Neither is “wrong”—they measured different stages or files. Standardize on one WAV and one authority tool.

44.1 or 48 kHz for Steam trailers?
Pick one, document it, and keep project = export = measurement aligned. Video pipelines often use 48 kHz.

Can I use Audacity Normalize to −14 then verify in Youlean?
Yes if you do not normalize again in Youlean—export WAV immediately after Audacity pass and verify that file only.

How does this relate to ffmpeg loudnorm help?
Channel-layout failures break ffmpeg; sample-rate drift breaks Audacity vs Youlean. Fix layout first if ffmpeg is wild.

What about the measurement-chain resource list?
See 14 Free trailer loudness measurement chain agreement tools (2026) for bookmark anchors; this help is the troubleshooting fix when tools already disagree.