Audio Mixer Strategist

Your agent for audio processing workflows — covering mixing, mastering, format conversion, and audio manipulation using FFmpeg and command-line audio tools.

When to Use This Agent

Choose Audio Mixer Strategist when:

• Mixing multiple audio tracks or stems into a final output
• Applying audio effects (normalization, EQ, compression, reverb)
• Converting between audio formats (WAV, MP3, FLAC, AAC, OGG)
• Processing podcast audio (noise reduction, leveling, splitting)
• Automating batch audio processing workflows

Consider alternatives when:

• You need video editing — use a video processing agent
• You need music generation — use a music AI agent
• You need real-time audio streaming — use a streaming architecture agent

Quick Start

Copy
# .claude/agents/audio-mixer.yml
name: Audio Mixer Strategist
model: claude-sonnet
tools:
  - Read
  - Write
  - Edit
  - Bash
  - Glob
  - Grep
description: Audio processing agent for mixing, effects, format conversion, and batch audio workflows using FFmpeg

Example invocation:

claude "Create an FFmpeg command to mix a voice track and background music — voice should be -3dB louder, music ducked during speech, with final output normalized to -14 LUFS for podcast distribution"

Core Concepts

Audio Processing Pipeline

Stage	Operation	FFmpeg Filter
Input	Load audio files	-i input.wav
Normalize	Level matching	loudnorm=I=-14:TP=-1
EQ	Frequency adjustment	equalizer=f=100:t=h:w=200:g=3
Compress	Dynamic range control	acompressor=threshold=-20dB:ratio=4
Mix	Combine multiple tracks	amix=inputs=2:duration=longest
Fade	Fade in/out	afade=t=in:st=0:d=2
Output	Encode and save	-c:a libmp3lame -q:a 2

Common FFmpeg Audio Commands

Copy
# Convert WAV to MP3 at 192kbps
ffmpeg -i input.wav -c:a libmp3lame -b:a 192k output.mp3

# Normalize audio to -14 LUFS (podcast standard)
ffmpeg -i input.wav -af loudnorm=I=-14:TP=-1:LRA=11 output.wav

# Mix voice + music with ducking
ffmpeg -i voice.wav -i music.wav \
  -filter_complex "[1]volume=0.3[bg];[0][bg]amix=inputs=2:duration=first" \
  output.wav

# Extract audio from video
ffmpeg -i video.mp4 -vn -c:a copy audio.aac

Configuration

Parameter	Description	Default
output_format	Target audio format (wav, mp3, flac, aac)	mp3
sample_rate	Audio sample rate	44100
bit_depth	Bit depth for WAV output	24
loudness_target	Target LUFS level	-14
channels	Output channel count (mono, stereo)	stereo

Best Practices

1. Process in lossless format, encode to lossy only at final output. Work with WAV or FLAC during editing and mixing. Converting to MP3 between processing steps compounds quality loss. Encode to the distribution format only as the final step.

2. Use loudnorm filter for broadcast/podcast normalization. The loudnorm filter implements EBU R128 loudness standard, which is required for podcast platforms (Spotify: -14 LUFS, Apple Podcasts: -16 LUFS, YouTube: -14 LUFS).

3. Apply effects in the correct order. Standard order: noise reduction → EQ → compression → normalization → encoding. Reversing the order (compressing before EQ) produces different, often inferior results.

4. Always use the two-pass loudnorm for accurate normalization. Single-pass loudnorm estimates loudness but can overshoot. Two-pass (-filter:a loudnorm=I=-14:dual_mono=true:print_format=json first pass, then apply measured values) produces precise results.

5. Preview with short clips before processing large files. Test FFmpeg filter chains on a 30-second sample before processing a 2-hour file. This catches filter errors and lets you tune parameters quickly.

Common Issues

Output audio has clipping or distortion after mixing. When combining multiple tracks, their levels sum and can exceed 0 dBFS. Reduce individual track volumes before mixing, or use dynaudnorm or acompressor after mixing to control peaks.

FFmpeg filter syntax errors are hard to debug. Complex filter graphs with multiple inputs and outputs are sensitive to syntax. Build filter chains incrementally — add one filter at a time and verify output. Use -filter_complex_script for complex graphs to keep commands readable.

Audio quality is poor after format conversion. Default codec settings may use low quality. Explicitly set quality: -c:a libmp3lame -q:a 0 (highest VBR quality) or -c:a libmp3lame -b:a 320k (highest CBR). For AAC, use -c:a aac -b:a 256k or the higher-quality libfdk_aac codec.