菜单
Audio engineering — mastering, mixing, EQ, compression, loudness standards, synthesis, podcast production, music theory, spectrum analysis.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/Tibsfox/gsd-skill-creator --skill audio-engineering命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | audio-engineering |
| description | Audio engineering — mastering, mixing, EQ, compression, loudness standards, synthesis, podcast production, music theory, spectrum analysis. |
| activationKeywords | ["audio","mastering","mixing","EQ","equalization","compressor","compression","limiter","loudness","LUFS","true peak","EBU","podcast","waveform","spectrum","synthesis","synthesizer","FM synthesis","subtractive","granular","modular","BPM","tempo","key detection","music theory","chord","scale","sox","ableton","DAW","reverb","delay","noise reduction","normalization","dithering","sample rate","bit depth"] |
| type | skill |
| category | media |
| status | stable |
| origin | tibsfox |
| modified | false |
| first_seen | "2026-04-11T00:00:00.000Z" |
| first_path | examples/skills/media/audio-engineering/SKILL.md |
| superseded_by | null |
Built on research from 30 Music cluster projects, 360 PNW musicians deep-dived (S36/SPS), Ableton Live research (ABL), Deep Audio (DAA), Dead Frequencies (DFQ), and High Fidelity amplifier analysis (HFR/HFE).
Expert-level audio engineering covering mastering, mixing, loudness standards, synthesis, podcast production, music theory, and spectrum analysis. Works alongside the ffmpeg-media skill for codec/format operations.
| Platform | Target LUFS | True Peak | Standard |
|---|---|---|---|
| Spotify | -14 LUFS | -1 dBTP | AES streaming |
| Apple Music | -16 LUFS | -1 dBTP | Sound Check |
| YouTube | -14 LUFS | -1 dBTP | ITU-R BS.1770 |
| Podcast (Apple) | -16 LUFS | -1 dBTP | Apple spec |
| Podcast (Spotify) | -14 LUFS | -1 dBTP | Spotify spec |
| Broadcast TV | -24 LUFS | -2 dBTP | EBU R128 |
| Broadcast US | -24 LKFS | -2 dBTP | ATSC A/85 |
| CD master | -9 to -12 LUFS | -0.3 dBTP | Red Book |
| Film/Cinema | -24 LUFS | -1 dBTP | SMPTE RP 200 |
# Measure integrated loudness (LUFS) with ffmpeg
ffmpeg -i input.wav -af loudnorm=print_format=json -f null - 2>&1 | grep -A20 "Parsed_loudnorm"
# Full EBU R128 scan
ffmpeg -i input.wav -af ebur128=peak=true -f null - 2>&1 | tail -20
# Loudness normalization to -14 LUFS (two-pass for accuracy)
# Pass 1: measure
ffmpeg -i input.wav -af loudnorm=I=-14:LRA=11:TP=-1:print_format=json -f null - 2>&1 > /tmp/loudnorm.json
# Pass 2: apply (use measured values from pass 1)
ffmpeg -i input.wav -af loudnorm=I=-14:LRA=11:TP=-1:measured_I=-18.5:measured_LRA=9.2:measured_TP=-0.5:measured_thresh=-28.3 output.wav
# Normalize peak to -1 dBFS
sox input.wav output.wav gain -n -1
# Compressor (threshold -20dB, ratio 4:1, attack 5ms, release 50ms)
sox input.wav output.wav compand 0.005,0.05 -20,-20,-10,-10,0,-6
# 3-band EQ (low shelf +3dB at 200Hz, mid cut -2dB at 2kHz, high shelf +1dB at 8kHz)
sox input.wav output.wav bass +3 200 equalizer 2000 1q -2 treble +1 8000
# Noise reduction (profile then reduce)
sox noisy.wav -n noiseprof /tmp/noise.prof
sox noisy.wav clean.wav noisered /tmp/noise.prof 0.21
# Generate tone (440Hz sine, 3 seconds)
sox -n -r 44100 -c 1 tone.wav synth 3 sine 440
# Spectrum analysis (generate spectrogram PNG)
sox input.wav -n spectrogram -o spectrum.png
Input → EQ (corrective) → Compression → EQ (tonal) → Stereo Width → Limiting → Dithering → Output
# Full mastering chain: EQ → compression → limiting → loudness normalization
ffmpeg -i mix.wav -af "\
equalizer=f=80:t=h:w=100:g=2,\
equalizer=f=3000:t=h:w=1000:g=-1.5,\
equalizer=f=12000:t=h:w=2000:g=1,\
acompressor=threshold=-18dB:ratio=3:attack=10:release=100:knee=6,\
alimiter=limit=-1dBFS:level=false,\
loudnorm=I=-14:LRA=11:TP=-1\
" -ar 44100 -sample_fmt s16 mastered.wav
# Dithering (16-bit with triangular dither for CD)
ffmpeg -i master_24bit.wav -af "dither=method=triangular" -sample_fmt s16 -ar 44100 cd_master.wav
| Band | Range | Character | Common Uses |
|---|---|---|---|
| Sub-bass | 20-60 Hz | Felt, not heard | Kick fundamental, sub bass |
| Bass | 60-250 Hz | Warmth, body | Bass guitar, kick punch, vocal warmth |
| Low-mid | 250-500 Hz | Muddiness zone | Cut here to clean up mixes |
| Mid | 500-2000 Hz | Body, presence | Vocal clarity, guitar body |
| Upper-mid | 2-4 kHz | Presence, bite | Vocal intelligibility, guitar attack |
| Presence | 4-6 kHz | Definition, edge | Consonant clarity, string attack |
| Brilliance | 6-12 kHz | Air, shimmer | Cymbals, vocal air, acoustic sparkle |
| Ultra-high | 12-20 kHz | Air, sparkle | Subtle sheen (careful: sibilance) |
| Source | Threshold | Ratio | Attack | Release | Knee |
|---|---|---|---|---|---|
| Vocals | -18 to -12 dB | 2:1 to 4:1 | 5-15 ms | 40-80 ms | Soft |
| Drums (bus) | -15 to -10 dB | 3:1 to 6:1 | 10-30 ms | 50-100 ms | Hard |
| Bass | -15 to -8 dB | 3:1 to 8:1 | 10-30 ms | 100-200 ms | Hard |
| Acoustic guitar | -20 to -12 dB | 2:1 to 4:1 | 10-25 ms | 100-150 ms | Soft |
| Mix bus | -20 to -15 dB | 1.5:1 to 2:1 | 10-30 ms | 100-300 ms | Soft |
| Podcast | -20 to -15 dB | 3:1 to 5:1 | 5-10 ms | 50-100 ms | Soft |
| Type | How It Works | Character | Classic Synths |
|---|---|---|---|
| Subtractive | Oscillator → Filter → Amplifier | Warm, analog, rich | Minimoog, Prophet-5, Juno-106 |
| FM | Operators modulating each other's frequency | Metallic, bell-like, bright | DX7, FM8 |
| Wavetable | Morphing between stored waveforms | Evolving, complex, modern | PPG Wave, Serum, Vital |
| Granular | Tiny audio grains layered and scattered | Atmospheric, textural, ambient | Granulator, Pigments |
| Additive | Sum of individual sine wave partials | Precise, organ-like | Kawai K5, Razor |
| Physical modeling | Mathematical model of physical instrument | Realistic, expressive | Chromaphone, Pianoteq |
| Sample-based | Recorded audio, pitch-shifted and layered | Realistic, natural | Kontakt, Sampler |
C
F G
Bb D
Eb A
Ab/E
| Name | Numerals | Example in C | Use |
|---|---|---|---|
| Pop | I-V-vi-IV | C-G-Am-F | 80% of pop music |
| Blues | I-IV-V | C-F-G | Blues, rock |
| Jazz ii-V-I | ii-V-I | Dm7-G7-Cmaj7 | Jazz standard |
| Andalusian | i-VII-VI-V | Am-G-F-E | Flamenco, dramatic |
| Canon | I-V-vi-iii-IV-I-IV-V | C-G-Am-Em-F-C-F-G | Pachelbel, ballads |
| Minor blues | i-iv-V | Am-Dm-E | Minor blues |
# Record from default mic (sox)
sox -d -r 44100 -c 1 -b 16 recording.wav
# Record with ffmpeg (specify ALSA device on Linux)
ffmpeg -f alsa -i default -ar 44100 -ac 1 recording.wav
# 1. Noise reduction
sox recording.wav -n trim 0 0.5 noiseprof /tmp/noise.prof
sox recording.wav clean.wav noisered /tmp/noise.prof 0.21
# 2. Normalize + compress + EQ for voice
ffmpeg -i clean.wav -af "\
highpass=f=80,\
lowpass=f=12000,\
equalizer=f=3000:t=h:w=1000:g=2,\
acompressor=threshold=-20dB:ratio=4:attack=5:release=50,\
loudnorm=I=-16:TP=-1\
" -ar 44100 podcast_ready.wav
# 3. Export MP3 for distribution
ffmpeg -i podcast_ready.wav -c:a libmp3lame -b:a 128k \
-metadata title="Episode Title" \
-metadata artist="Show Name" \
-metadata album="Podcast Name" \
-metadata genre="Podcast" \
episode.mp3
# 4. Generate waveform for show notes
ffmpeg -i episode.mp3 -filter_complex "showwavespic=s=1920x200:colors=0x1a1a2e" -frames:v 1 waveform.png
# Set all metadata
ffmpeg -i episode.mp3 -c copy \
-metadata title="EP 42: The Memory Architecture" \
-metadata artist="GSD Podcast" \
-metadata album="Getting Shit Done" \
-metadata track="42" \
-metadata date="2026" \
-metadata comment="LOD-tiered memory system deep dive" \
tagged.mp3
# Install aubio for beat/pitch detection
# apt install aubio-tools
# BPM detection
aubiotempo input.wav
# Pitch/key detection
aubiopitch -i input.wav -p yinfft
# Onset detection (transient markers)
aubioonset input.wav
# Generate stats (includes RMS, peak, DC offset)
sox input.wav -n stats 2>&1
| Format | Sample Rate | Bit Depth | Use |
|---|---|---|---|
| CD | 44.1 kHz | 16-bit | Consumer playback |
| DVD | 48 kHz | 24-bit | Video soundtrack |
| Hi-Res | 96 kHz | 24-bit | Audiophile streaming |
| Studio | 96-192 kHz | 32-bit float | Recording/mixing |
| Podcast | 44.1 kHz | 16-bit | Voice distribution |
| Phone/VoIP | 8-16 kHz | 16-bit | Voice calls |
# Downsample from 96kHz/24-bit to 44.1kHz/16-bit with dither
sox input_96_24.wav -r 44100 -b 16 output_441_16.wav dither -s
# Same with ffmpeg
ffmpeg -i input_96_24.wav -ar 44100 -sample_fmt s16 -af "dither=method=triangular" output.wav