Everything you need to master your audio
Master your audio in 3 simple steps. No technical knowledge required.
LOUDCRAFT runs a studio-grade pipeline: input health analysis, genre detection, corrective EQ, adaptive dynamics, Immersive Spatial Engine, loudness normalization, and 10-point quality gate. Powered by a dedicated audio engine for professional results. Here's the quick workflow.
Drag a file or click to browse. Supports MP3, WAV, FLAC, AAC, OGG, M4A (up to 200MB).
Choose settings or leave defaults. Auto-detect handles everything automatically.
Download as WAV, FLAC, MP3, or AAC. Compare A/B before downloading.
Supported formats: MP3, WAV, FLAC, AAC, OGG, M4A. Maximum size: 200MB per file.
Batch mode: Drop multiple files at once to process them all with the same settings. Up to 20 tracks per batch.
Upload WAV or FLAC for best results. Lossy formats (MP3/AAC) have already lost information that cannot be recovered during mastering.
Leave on Auto-detect and the AI will analyze your track to choose the best processing style. You can override it if needed.
Uses Essentia.js ML model (11 audio features, 20 genre profiles). If confidence < 40%, it blends the top-2 genre parameters proportionally. Below 15%, it falls back to default moderate settings.
20 genres across 4 processing tiers. Each genre has tuned compression curves, spatial width, reflection patterns, and limiter settings.
Override genre only if auto-detection is wrong. Choosing the wrong genre intentionally (e.g., "metal" for a jazz track) will produce unnatural results.
LUFS = how loud your track will be. Leave at -14 for streaming platforms (Spotify, Apple Music, YouTube).
Integrated LUFS per ITU-R BS.1770. Two-pass loudnorm ensures precise targeting within 0.5 LU tolerance. True peak limiting at -1.0 dBTP (adjustable by genre tier).
| Preset | Target | Use Case |
|---|---|---|
| Streaming | -16 LUFS | Spotify, Apple Music, YouTube |
| Broadcast | -14 LUFS | TV, Radio, Podcast (default) |
| Loud | -11 LUFS | Club, DJ, competitive |
| CD Master | -9 LUFS | CD/Vinyl distribution |
Streaming services apply their own normalization. Mastering louder than -14 LUFS means your track gets turned down by Spotify/Apple, gaining nothing but losing dynamic range.
Controls how "wide" and immersive your music sounds. Default 100% is genre-optimized. Go higher for more effect, or set to 0% to skip spatial processing entirely.
Scales all spatial parameters proportionally via scaleSpatialParams(). Affects side gain, Haas delay, reflection amplitude, and stereotools width. Does NOT scale crossfeed, dynamics, or limiter (structural parameters). Range: 0% (bypass) to 150% (maximum).
Use 0% if your mix already has spatial processing (e.g., immersive audio production). Above 100% is best for headphone-focused listening.
When you click "MASTER YOUR TRACK", your audio goes through 8 processing stages automatically. Each stage is optimized for your detected genre.
10-step orchestrated pipeline with per-genre parameterization and input-adaptive processing. Steps 3b and 6b are sub-steps of their parent stages.
Checks file format, size, and probes duration, sample rate, channels.
AI genre detection + LUFS measurement. Runs 13-point health check and adaptive corrective EQ.
Detects song structure (quiet, verse, build, chorus, drop) using windowed energy analysis with adaptive dynamics per section.
Multiband compression with LRA-aware ratios. Preserves dynamics for quiet material, adds density for loud genres.
Immersive Spatial Engine. M/S stereo widening with genre-adaptive parameters and mono-safety protection.
Mastering EQ, Dynamic EQ, and M/S EQ polish chain. Harmonic enhancement for analog cohesion.
Loudness normalization to industry targets with true-peak safety. Exports as 24-bit/48kHz WAV.
10 broadcast-standard checks: true peak, LUFS, LRA, mono compatibility, clipping, harshness, transients, stereo width, short-term LUFS, PLR.
The pipeline adapts to your input. A dynamic jazz recording gets gentle treatment; a compressed EDM track gets aggressive spatial expansion. The dynamics adapter measures actual crest factor and LRA to scale compression, not just genre defaults.
The spatial engine makes your music feel wider and more immersive. It uses M/S (Mid/Side) processing to widen the stereo field while protecting vocals and mono compatibility, all optimized for your genre.
Studio-grade M/S processing with multiband compression, stereo field widening, and feature-based spatial intensity. Auto-derives spatial parameters from input mono correlation — no manual configuration needed.
Splits audio into 4 frequency bands (sub / low / mid / high) using high-order linear-phase crossover filters.
Parallel compression per band. Each genre tier has proprietary threshold, ratio, and envelope settings tuned for optimal dynamics preservation.
Mid/Side encoding per band. Center content (vocals) is preserved while side channels receive genre-adaptive spatial enhancement.
All 4 processed bands mixed back into a single stereo stream with normalized gains.
Multi-tap room simulation. Genre-specific ambience profiles create the sense of a physical acoustic space.
Binaural crossfeed for headphone speaker simulation. Prevents extreme stereo separation from sounding unnatural on headphones.
Psychoacoustic HRTF-based processing that simulates vertical spatial cues. Makes audio feel "taller" in the soundfield.
Safety net preventing intersample peaks above the broadcast limit. Genre-specific limiting for transparent loudness control.
The ISE is a psychoacoustic stereo simulation — not an object-based format requiring a decoder. It creates spatial immersion through M/S processing, HRTF-inspired filters, and early reflections. Most dramatic on headphones. Vocals are protected via multiple safety layers including dynamic and spatial safeguards, mono compatibility checks, and quality gate verification.
After mastering, 10 automatic checks verify your track meets professional broadcast standards. A score of 100/100 means everything passed perfectly.
EBU R128 / AES broadcast compliance. Each check scores 0-10 points (fail/warn/pass). Total: 0-100. Thresholds calibrated to broadcast and streaming platform requirements.
Score: 10 points per check (pass = 10, warn = 5, fail = 0). A score of 80+ is release-ready.
If mono compatibility shows a warning, reduce spatial intensity from 100% to ~80%. This usually fixes it while keeping most of the spatial effect. Alternatively, "Skip Spatial" entirely for mono-critical playback (club PAs, phone speakers).
After mastering, LOUDCRAFT measures several technical aspects of your track. Here's what each metric means and why it matters for your music.
Key metrics measured by the Quality Gate and how they map to commercial mastering standards.
LUFS measures perceived loudness (not peak level). Streaming platforms normalize to specific targets:
LOUDCRAFT uses three LUFS measurements:
Integrated LUFS — Average loudness of the entire track. This is what platforms use for normalization. Short-Term LUFS — Loudness in 3-second windows. Detects if peak sections are too loud. Momentary LUFS — 400ms snapshots for instant peaks.
These are reproduction targets, not mastering targets. Master for the sound you want, then check if the result is competitive. Commercial pop/hip-hop typically lands at -7 to -10 LUFS, while jazz/folk works at -12 to -18 LUFS intentionally.
True Peak measures the actual signal peak after digital-to-analog conversion, which can exceed 0 dBFS due to intersample peaks. The broadcast standard is -1.0 dBTP.
Why it matters: streaming platforms encode to AAC/Ogg Vorbis. If True Peak exceeds -1 dBTP, the codec conversion creates audible distortion — clicks, pops, and harshness that weren't in your master.
LOUDCRAFT uses a 4x oversampled limiter (192 kHz) to catch intersample peaks with a two-pass safety net.
PLR = True Peak minus Integrated LUFS. It measures density — how compressed and "filled in" your track is.
Two tracks can measure exactly -14 LUFS but sound radically different. The one with lower PLR has more constant energy — the "loudness" you hear in commercial masters comes from density, not volume.
Commercial genre ranges: EDM/Trap 6-8 dB PLR, Pop/Rock 8-10 dB, Jazz/Folk 10-14 dB. If your PLR shows WARN, your track is very dense — intentional for aggressive genres, but a red flag for acoustic music.
The frequency distribution of your track affects both loudness and clarity:
Sub-bass (20-50 Hz) consumes headroom without adding perceived loudness. LUFS uses K-weighting, which still counts sub-bass energy. Excess sub forces the overall level down.
Presence (2-5 kHz) is where your ear is most sensitive. A track with strong presence sounds louder than one without, even at identical LUFS. LOUDCRAFT applies a conditional +0-2 dB boost at 3 kHz when this zone is weak.
Harshness (2-5 kHz excess) — too much presence creates fatigue. The Health Check monitors the 2-5 kHz ratio and flags tracks above 0.35.
20 genres organized into 4 processing tiers. Each tier defines a processing philosophy that scales from minimal intervention to maximum enhancement.
| Characteristic | Delicate | Conservative | Moderate | Aggressive |
|---|---|---|---|---|
| Compression | Minimal | Gentle | Balanced | Heavy |
| Spatial Width | Narrow | Moderate | Wide | Maximum |
| Room Size | Large / airy | Medium | Medium-tight | Tight |
| Limiting | Conservative | Conservative | Standard | Aggressive |
The spatial engine adapts width per song section:
| Section | Width Multiplier | Effect |
|---|---|---|
| Quiet (intro/outro) | 70% | Narrower, more intimate |
| Verse | 85% | Moderate width |
| Build (pre-chorus) | 100% | Full default width |
| Chorus | 115% | Wider, more expansive |
After mastering, click "COMPARE A/B" to hear original vs. mastered side-by-side with real-time visualizations. Switch instantly between versions to hear the difference.
Web Audio API pipeline with click-free crossfade switching, real-time FFT analysis, and independent stereo metering for accurate A/B comparison.
A = original audio, B = mastered, D = Diff mode (plays only what mastering changed: B minus A).
| Visualizer | Shows | Look For |
|---|---|---|
| Particle Frequency | Spectrum as floating particles | Balanced energy distribution |
| Waveform | Time-domain oscilloscope | Clean signal, no flat-topping |
| Frequency Spectrum | 64-bar log-scale analyzer | Smooth roll-off, no harsh peaks |
| Vectorscope | L/R Lissajous phase display | Signal in top half = good mono |
| Correlation Meter | -1 to +1 phase meter | Above 0.3 = safe |
| Mid/Side Spectrum | Separate M and S content | Mid dominates, side adds width |
| Chromatic Wheel | Key/harmonic detection | Identified musical key |
| Mastering Radar | 6-axis spider chart A vs B | Balanced polygon = well-mastered |
| Loudness Timeline | RMS over time | Consistent loudness |
| Spectrogram | Time-frequency heat map | Richer harmonics in mastered |
| 3D Waterfall | Perspective spectrogram | Frequency changes over time |
| Parametric EQ | 8-band interactive EQ | Fine-tune the master |
| Peak Meters | L/R peak levels | Balanced channels, no clipping |
| Metrics Table | Original vs Mastered values | Delta column shows improvement |
8 draggable bands to fine-tune your master after processing. Drag to move frequency/gain, scroll for Q width, double-click to reset a band. "Export with EQ" bakes your EQ changes into a new file.
Loop Region: Click and drag on the Track Overview to loop a section. Press C to clear.
Stereo Width: Slider from 0% (mono) to 200% (extra wide). Use 0% to check mono compatibility.
Spectrogram Comparison: Side-by-side original vs. mastered spectrograms. Click to enlarge.
Re-Master: Change settings and reprocess without re-uploading your file.
Download Formats: WAV (lossless), FLAC (lossless compressed), MP3 320kbps, AAC 256kbps.
Available on the Compare page:
| Key | Action | Note |
|---|---|---|
| Space | Play / Pause | |
| A | Switch to Original | Not in blind mode |
| B | Switch to Mastered | Not in blind mode |
| D | Switch to Diff mode | Hear what changed |
| X / Y | Switch in Blind mode | Blind test only |
| 1 / 2 | Pick X or Y as mastered | Blind test only |
| T | Toggle Blind Test | |
| L | Toggle Loudness Match | |
| E | Toggle EQ Bypass | |
| C | Clear Loop Region | |
| ← | Seek back 5 seconds | |
| → | Seek forward 5 seconds |
The Admin page provides system diagnostics and management tools in 6 panels:
| Panel | Description |
|---|---|
| System Status | Engine version, Node.js, Python, analyzer status, memory, disk usage, uptime |
| Test Runner | Run the full test suite (237 tests) from the browser with real-time output |
| Job Manager | View, download, compare, or delete all processed jobs. Bulk cleanup. |
| Quick Process | Process files with full control. Supports analyze-only (dry-run) mode. |
| Genre Reference | Detailed view of all 20 genre parameters (compression, spatial, limiter, reflections) |
| Config Inspector | Full dump of all internal configuration values |
Two common causes: (1) Auto-detection picked the wrong genre — try manually selecting the correct genre. (2) Your track was already mastered — re-mastering a mastered track (double-mastering) degrades quality. Upload the pre-mastered mix instead.
WAV or FLAC for best results. They preserve full audio quality. MP3 and AAC work but have already lost information that can't be recovered. The output is always 24-bit/48kHz WAV (with optional export to MP3/FLAC/AAC).
The spatial engine widened your stereo image, which can reduce mono correlation. Try reducing spatial intensity to 80% or lower. If the track will primarily play on mono systems (phone speakers, club PAs), consider enabling "Skip Spatial".
The Immersive Spatial Engine (ISE) is a psychoacoustic stereo simulation using M/S processing, HRTF-inspired elevation filters, and early reflections to create the perception of spatial immersion. It is a stereo processing technique, not an object-based format. The effect is most dramatic on headphones.
8 pre-mastering diagnostics: DC offset, spectral tilt, clipping detection, phase correlation, bass phase coherence, crest factor, resonance detection, and noise floor. Issues found trigger automatic corrective EQ before the main processing begins.
All uploaded and processed files are automatically cleaned up after 24 hours. Download your mastered files before they expire. The compare page shows a countdown timer for remaining time.