Term
dBFS, Headroom & Clipping
Three related concepts for digital audio levels: dBFS is the scale with 0 dBFS as the hard ceiling, headroom is the safety margin below the maximum, clipping is the hard cut-off when a signal exceeds it.
dBFS, Headroom & Clipping — explained in more detail
dBFS (Decibels relative to Full Scale) is the logarithmic scale for digital audio levels. Its reference is the numerical maximum a sample can take at the given bit depth — this ceiling is 0 dBFS. Real-world levels sit below it and are stated as negative values: -6 dBFS is half the amplitude, -20 dBFS a much quieter signal. Unlike dBSPL, dBFS is dimensionless and defined purely relative to the bit-maximum.
Headroom is the distance between the loudest point of a recording (peak) and 0 dBFS. It serves as a safety buffer: common mastering practice leaves -1 dBFS to -3 dBFS of peak headroom to absorb rounding errors from resampling, lossy encoders or mix-bus summing without later stages tipping into clipping.
Clipping happens the moment a signal would exceed 0 dBFS. Since no value above the maximum can be represented, the waveform is cut off hard — it goes flat at top and bottom. The result sounds rough, harsh and distorted. The original information is lost beyond recovery; clipping is destructive, not reversible.
Example / In practice
Clean dictation recordings typically sit between -12 dBFS and -3 dBFS peak. When they are fed to ASR models such as Whisper, clipping is especially damaging: distorted vowels look like broken phonemes to the model and lead to misreads. Software normalization to a -1 dBFS peak is therefore the standard move — hot level without digital overflow.
How it differs from related terms
dBSPL measures acoustic sound pressure in the air and is not a digital concept. Peak is the highest instantaneous value, RMS is average energy over a window, LUFS is perceived loudness under broadcast standards. Headroom is normally stated against peak level, not against RMS or LUFS.
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.