Phantom’s tools are organized by workflow stage. Each tool accepts audio files and returns measurement data your AI assistant can reason about.
Pre-mix Analysis
Understand your material before touching faders.
| Tool | What it measures |
|---|---|
| analyze_spectrum | Frequency balance — centroid, rolloff, contrast, flatness, dissonance |
| analyze_loudness | Integrated LUFS, true peak, loudness range, short-term and momentary |
| analyze_dynamics | RMS, peak, crest factor, dynamic range, complexity |
| analyze_phase | Phase coherence per frequency band, polarity, inter-channel delay |
| detect_problems | Clipping, DC offset, ISP, hum, sibilance, mud, harshness, resonances |
| full_diagnostic | Comprehensive single-file analysis (runs all pre-mix tools) |
| batch_diagnostic | Parallel analysis on up to 50 files |
During Mixing
Make informed decisions while mixing.
| Tool | What it measures |
|---|---|
| analyze_stereo | Stereo width, balance, mid/side ratio, correlation, panorama |
| analyze_masking | Per-octave frequency overlap between two stems |
| multi_stem_masking | Frequency collision severity across multiple stems |
| compare_phase | Phase relationship comparison between two sources |
Mastering Prep
Compare and match to targets.
| Tool | What it measures |
|---|---|
| compare_to_profile | Compare against genre target specs (EDM, hip-hop, acoustic, etc.) |
| compare_to_reference | A/B spectral and loudness analysis against a reference track |
| match_to_reference | Automated spectral, loudness, and width matching recommendations |
Remediation and Processing
Fix problems and apply corrections. These tools require the processing extra: pip install phantom-audio[processing].
| Tool | What it does |
|---|---|
| fix_audio | Auto-fix detected audio problems with corrective EQ and filtering |
| apply_processing | Apply custom audio processing chains to files |
Utilities
| Tool | What it does |
|---|---|
| separate_stems | Separate audio into vocals, drums, bass, and instruments via Demucs |
Pro tip
You don’t need to call tools by name. Describe what you want to know — “Is my mix too loud?”, “Are the vocals masking the guitar?”, “Compare this to a professional reference” — and your AI assistant will pick the right tool.