UitbreidenOS When to activate
- Cleaning up raw podcast recordings
- Applying noise reduction, EQ, and compression
- Normalizing audio levels across speakers
- Mastering final mix for distribution (-16 LUFS stereo, -19 LUFS mono)
- Creating FFmpeg or SoX processing pipelines
When NOT to use
- For music production or mixing
- For live audio engineering
- For sound design and foley
Instructions
- Assess raw audio. Check for background noise, plosives, sibilance, clipping, and level inconsistency.
- Apply noise reduction. Profile noise floor from silence segments; apply spectral subtraction (Audacity/SoX).
- Remove artifacts. De-ess sibilance (>6kHz), de-plosive low-frequency pops (<100Hz), trim dead air.
- Level and compress. Normalize each speaker track to -6dB peak; apply 3:1 compression with 10ms attack.
- EQ each voice. High-pass at 80Hz, cut boxy frequencies (200-400Hz), presence boost at 3-5kHz.
- Master final mix. Target -16 LUFS integrated (stereo) or -19 LUFS (mono) per podcast standard.
- Export and validate. MP3 128kbps stereo + WAV backup; verify loudness with
ffmpeg -i file.mp3 -af loudnorm -f null -.
Example
# Noise reduction + normalization with SoX
sox input.wav cleaned.wav noiseprof silence.prof
sox cleaned.wav output.wav noise reduction silence.prof gain -n -3
# Loudness normalization with FFmpeg
ffmpeg -i mixed.wav -af loudnorm=I=-16:TP=-1.5:LRA=11 -ar 44100 final.mp3