| name | daw-master:mix-engineer |
| description | Use when polishing AI-generated audio — per-stem processing with SoX cleanup, EQ, compression, and remix. Processes stems or full mixes into polished WAVs ready for mastering. |
| version | 0.1.0 |
| author | Ported from bitwize-music-studio |
| license | MIT |
| metadata | {"hermes":{"tags":["audio","mix","polish","stems","sox","ffmpeg"],"related_skills":["daw-master:sox-engine","daw-master:ffmpeg-audio","daw-master:audio-analyzer","daw-master:mastering-engineer"]}} |
Mix Engineer
Polishes raw AI-generated audio (Suno, etc.) by processing per-stem WAVs with targeted cleanup, EQ, and compression using SoX + FFmpeg engines, then remixes into a polished stereo WAV ready for mastering.
Role: Per-stem processing, noise reduction, frequency cleanup, dynamic control, stem remixing.
Not role: Creative production, lyrics, generation.
Primary tools: sox-engine.transform, ffmpeg-audio.transform, audio-analyzer.analyze
Core Principles
- Stems first: When
stems/ directory is present, process each stem independently for targeted processing impossible on a full mix.
- Preserve performance: Mix polishing removes defects, not character. Be conservative.
- Non-destructive: All processing writes to
polished/ — originals never modified.
- Frequency coordination with mastering: Mix presence boost at ~3 kHz (clarity); mastering harshness cut at ~3.5 kHz — different center frequencies minimize cancellation.
Dependencies
sox-engine — single-track effects (EQ, compand, normalize, fades, highpass/lowpass/bandpass)ffmpeg-audio — multi-track mixing, loudnorm, advanced dynamicsaudio-analyzer — feature extraction for analysis decisions
No external Python packages beyond those already installed.
Path Convention
{audio_root}/artists/[artist]/albums/[genre]/[album]/
├── stems/
│ ├── 01-track-name/
│ │ ├── 0 Lead Vocals.wav
│ │ ├── 1 Backing Vocals.wav
│ │ ├── 2 Drums.wav
│ │ ├── 3 Bass.wav
│ │ └── ... (up to 12 stem types)
├── polished/ ← output
└── mastered/ ← mastering-engineer output
When to Use
- Cleaning up noise from AI-generated vocals or instruments
- Applying frequency-targeted EQ (mud cuts, presence boosts, high tames)
- Compressing stems to even out dynamics
- Removing clicks or transient spikes from generated audio
- Mixing stems into a polished full stereo mix
- Applying genre-specific processing chains (hip-hop, rock, EDM, folk, ambient)
API
analyze_mix_issues(album_path: str) -> Dict
Scan for stems or full mix and detect issues.
{
"source_mode": "stems" | "full_mix",
"tracks": [...],
"issues": {
"noisy_tracks": [...],
"muddy_tracks": [...],
"harsh_tracks": [...],
"clicky_tracks": [...],
"sub_bass_rumble": [...]
}
}
polish_audio(album_path: str, genre: str = "default", dry_run: bool = False) -> Dict
Process stems or full mix with genre-appropriate settings. Writes to polished/.
{
"mode": "stems" | "full_mix",
"tracks_processed": 5,
"output_dir": "/path/to/polished",
"commands": [...],
"success": bool
}
polish_album(album_path: str, genre: str = "default") -> Dict
End-to-end: analyze → polish → verify.
_generate_noise_profile(input_wav: str) -> str
Auto-generates SoX noise profile from quietest 100ms window.
_detect_clicks(audio_data, sr) -> List[float]
Detects transient clicks (> 6σ amplitude spikes).
Per-Stem Processing
Full chains for all 12 stem types (Vocals, Backing Vocals, Drums, Bass, Guitar, Keyboard, Strings, Brass, Woodwinds, Percussion, Synth, Other): references/stems.md
Quick reference:
| Stem | Key steps |
|---|
| Lead Vocals | NR → +2dB @ 3kHz → -2dB @ 7kHz → compress |
| Backing Vocals | NR → +1dB @ 3kHz → -2.5dB @ 7kHz → compress tighter |
| Drums | Click removal → compress gentle |
| Bass | Highpass 30Hz → mud cut 200Hz → compress |
| Guitar | Highpass 80Hz → mud cut 250Hz → +1.5dB @ 3kHz → compress |
| Keyboard | Highpass 40Hz → mud cut 300Hz → +1dB @ 2.5kHz → compress |
| Strings | Highpass 35Hz → +1dB @ 3.5kHz → wide stereo → very light compress |
| Brass | Highpass 60Hz → +1.5dB @ 2kHz → compress |
| Woodwinds | Highpass 50Hz → +1dB @ 2.5kHz → compress |
| Percussion | Highpass 60Hz → click removal → +1dB @ 4kHz → stereo width |
| Synth | Highpass 80Hz → +1dB @ 2kHz → stereo width → compress |
Genre Presets
| Genre | Vocal | Bass | Drums | High-mid |
|---|
| Default | 0 dB | 0 dB | 0 dB | –2 dB @ 7 kHz |
| Hip-Hop/Rap | +1 dB | +1 dB | +0.5 dB | aggressive |
| Rock/Metal | 0 dB | 0 dB | +0.5–1 dB | –2.5 to –3 dB |
| EDM/Electronic | 0 dB | +0.5–1 dB | +0.5–1 dB | lighter |
| Folk/Country | +0.5 dB | 0 dB | +0.5 dB | –1.5 dB |
| Ambient/Lo-Fi | 0 dB | 0 dB | 0 dB | –1.5 dB; lighter NR |
Handoff to Mastering
from daw_master.mastering_engineer import master_audio
master_audio(album_path, source_subfolder="polished", genre=genre)
Quality Standards
Before handoff: no clipping (peak < –0.1 dBFS), all samples finite, noise floor reduced, no artifacts, files in polished/.
