Term
Audio normalization
Audio normalization is the adaptive linear scaling of an audio signal to a target level — the tool first measures a reference value (peak, RMS or LUFS) and from that calculates the required gain.
Audio normalization — explained in more detail
Audio normalization raises or lowers the level of a recording so that a measured reference value lands exactly on a target. Unlike a fixed gain, it works adaptively: the tool scans the signal, determines the reference and computes the factor needed to hit the target. A quiet recording is boosted hard, an already loud one barely at all.
Three variants exist. Peak normalization refers to the loudest sample and lifts it to the target peak — typically -1 dBFS, via SoX norm -1. RMS normalization maps the average signal energy to a target value and is more stable when program material varies strongly in level. LUFS normalization measures perceived loudness per ITU-R BS.1770 — the broadcast standard with targets of -23 LUFS (EBU R128 for EU TV), -16 LUFS (podcast) or -14 LUFS (Spotify, YouTube).
What normalization is not: no change to dynamics, no EQ, no de-esser. Quiet and loud parts are scaled by exactly the same factor. The ratio between peak and average stays intact — only the absolute amplitude shifts.
Example / In practice
In ASR preprocessing for Whisper or Parakeet, sox in.wav out.wav norm -1 is a standard step: the peak is raised to -1 dBFS, so the model sees every recording at a similar level without clipping. In podcast mastering, LUFS normalization to -16 LUFS sits at the end of the chain so episodes play back at consistent loudness.
How it differs from related terms
Compression changes the ratio between quiet and loud passages and reduces dynamic range — normalization does not. Gain is a fixed offset without measurement; normalization is gain with measurement in front. Limiting softly catches peaks to prevent clipping and is often applied after normalization, but it is a separate, nonlinear step.
Entdecke mehr
AI Workflows by Keyword: How We Make Recurring Routines Enforceable
A typed keyword triggers a fixed AI routine — and every single step must be committed before the next one appears. Why that's the actual trick.
GlossarReasoning Effort
Reasoning Effort is a control parameter on modern reasoning models that sets how much internal step-by-step thinking (thinking tokens) a model spends before answering — higher levels raise quality but also latency and cost.
LexikonLLM Hallucinations — Causes and Remedies
Why LLMs confidently invent falsehoods, what types of hallucinations exist, and which remedies actually help — RAG, source enforcement, verification.