Help us improve
Share bugs, ideas, or general feedback.
From phantom
Professional mastering methodology for audio engineering. Covers the complete mastering chain (HPF through dither), corrective vs enhancement mastering, when to send a mix back, loudness targeting per platform, iZotope Ozone 11 workflow, and reference-based mastering. Use this skill whenever the user wants to master a mix, prepare audio for distribution, target a specific loudness standard, compare against a reference track, decide whether a mix needs more work or is ready for mastering, deliver for streaming (Spotify, Apple Music, YouTube), CD, or vinyl, or make any mastering decision. Also use when the user asks about LUFS, true peak, limiting, dithering, loudness normalization, or format-specific delivery requirements -- even if they don't say "mastering" explicitly.
npx claudepluginhub fadelabs/phantom --plugin phantomHow this skill is triggered — by the user, by Claude, or both
Slash command
/phantom:mastering-engineerThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> **Where this fits in the workflow:**
Guides audio mastering for streaming platforms with loudness optimization, tonal balance, and platform-specific targets. Analyzes WAV files and applies mastering to meet loudness standards.
Professional mixing methodology for audio engineering. Guides through pre-mix analysis, phase checking, gain staging, EQ decisions, compression selection, spatial processing, and automation. Encodes the decision-making process of a senior mix engineer backed by Phantom MCP measurement tools. Use this skill whenever the user wants to mix stems or tracks, balance a mix, make EQ or compression decisions, set up signal chains, choose compressor types, solve frequency conflicts between instruments, set up spatial processing (reverb, delay, panning), automate volume or effects, or compare their mix against a reference. Also use when the user mentions muddy mixes, harsh frequencies, buried vocals, kick/bass conflicts, or any mixing problem -- even if they don't say "mix" explicitly.
Provides FFmpeg commands for audio encoding (AAC/MP3/Opus/FLAC), EBU R128 loudness normalization, extraction from video, format conversion, volume/EQ adjustments, channel ops, and podcast/broadcast chains.
Share bugs, ideas, or general feedback.
Where this fits in the workflow:
/phantom:audio-diagnostician-- analyze stems first/phantom:session-architect-- set up the DAW session/phantom:mix-engineer-- mix with measurement-backed decisions/phantom:effects-engineer-- creative processing- You are here:
/phantom:mastering-engineer-- the final stage
Don't fix it if it isn't broken. The best mastering enhances what's already good -- it doesn't rescue a bad mix. If you're reaching for 4 dB of EQ, the mix needs work, not mastering. Know when to send it back.
Mastering is the last creative decision and the first technical one. You're simultaneously making the music sound its best AND ensuring it meets the technical requirements of every delivery platform. Both matter.
Run detect_problems and analyze_loudness on the mix first. Measurement, not opinion, drives this decision.
Deliver this assessment honestly. "Your mix needs a 6 dB cut at 300 Hz -- I can do 2-3 dB in mastering, but the rest needs to happen in the mix" is more helpful than silently trying to fix it.
Corrective first, enhancement second. Never enhance before correcting -- you'll be polishing problems.
Corrective mastering: fixing what the mix engineer missed or couldn't solve. Tonal imbalances, excessive dynamics, phase issues, noise. This is surgical, necessary work.
Enhancement mastering: elevating what's already good. Adding air, optimizing loudness for the target platform, subtle widening, warmth. This is the creative, satisfying part.
Nine stages in strict order. Each stage feeds the next -- skipping or reordering changes the result.
Remove sub-bass rumble below 20-30 Hz. This content is inaudible but wastes headroom and causes the limiter to react to energy you can't hear. A gentle HPF here gives you 1-2 dB of free headroom.
Linear-phase EQ for surgical cuts. Fix problems identified by analyze_spectrum -- resonances, tonal imbalances, mud. Narrow Q for problem frequencies, wider Q for broad tonal issues.
Run analyze_spectrum first. If the spectral centroid is unusually low (dark/muddy mix), you'll need a broad tilt or shelf. If there are narrow peaks, surgical notch cuts.
Glue, not squash. 1.5:1-3:1 ratio, 30-100 ms attack, auto or 200-300 ms release. Target 1-3 dB of gain reduction. If you're hitting more than 3 dB, the mix needs more compression in the mix stage, not the master.
VCA or Vari-Mu character. VCA for transparency, Vari-Mu for warmth and glue.
Frequency-specific control for problems that broadband compression can't solve. A bass note that rings too loud on certain hits. A harshness at 3 kHz that only appears in the chorus. Multiband lets you control one frequency range without affecting the others.
Use dynamic EQ over multiband compression when the problem is intermittent -- dynamic EQ only engages when the threshold is exceeded, leaving the signal untouched the rest of the time.
Shape the final tone. This is the enhancement stage -- air (gentle shelf above 10 kHz), warmth (subtle 100-200 Hz), presence (2-4 kHz). Wide Q, gentle boosts. If you're boosting more than 2 dB, reconsider whether this should have been a corrective cut elsewhere.
Make bass mono below 80-150 Hz. Low-end stereo content causes problems on every playback system -- vinyl needles skip, club subs cancel, headphones create an unstable image. Mono the sub.
Optional: gentle high-end widening above 8 kHz for air and spaciousness. Check with analyze_stereo -- correlation should stay above +0.3 after widening.
Optional warmth before the limiter. Add character before the limiter flattens it. Tube saturation for warmth, tape for glue. Light touch -- this is mastering, not mixing.
True peak ceiling at -1 dBTP (never 0 -- inter-sample peaks can exceed 0 dBTP even when the meter shows -0.1). Set the ceiling, then bring the threshold down until you reach your loudness target.
Run analyze_loudness after limiting to verify integrated LUFS and true peak meet your platform target.
For platform-specific loudness targets, see format-targets.md.
Only when reducing bit depth (24-bit to 16-bit for CD). Always the absolute last stage in the chain -- nothing after dither. TPDF for transparency, MBIT+ noise shaping for perceptually optimized dither (pushes dither noise into frequencies where hearing is least sensitive).
Never dither twice. Never dither when staying at the same bit depth. Never dither when going up in bit depth.
Run analyze_loudness to check integrated LUFS and true peak. Run compare_to_profile for genre-appropriate targets. Run compare_to_reference for A/B against a reference track.
Level-match before A/B comparison. Louder always sounds better -- remove the bias. Match the integrated LUFS of your master to the reference before comparing.
Match direction not destination. The reference is a compass, not GPS. If the reference has more high-end sparkle, add some sparkle -- but don't try to make your track sound identical.
Different references for different aspects. One reference for low-end balance, another for vocal clarity, another for overall loudness. No single reference is perfect in every dimension.
For platform-specific loudness and format specs, see format-targets.md.
Use match_to_reference for automated spectral/loudness/width matching if Matchering is installed. This gives you a starting point -- then adjust by ear. Apply the automated match at 50-70% strength, not 100%.
Use compare_to_reference to see per-dimension deviations between your master and the reference. This shows you exactly where your master differs -- spectrum, loudness, dynamics, stereo width -- so you can make targeted corrections.
For module-by-module Ozone 11 guidance (Equalizer, Dynamic EQ, Dynamics, Maximizer, Imager, Master Assistant, Match EQ, Exciter, Vintage modules, Codec Preview, Stem Focus), see ozone-guide.md.
Requires a Reaper MCP server for automated plugin insertion. See the setup guide for installation.
The Reaper stock plugins handle 80% of mastering tasks:
For compound Reaper operations that apply this skill's methodology, see reaper-recipes.md.
Available recipes:
Recipes require a Reaper MCP server (TwelveTake recommended). See setup guide for installation.