@audio/loudness-replaygain

ReplayGain 2.0 — track gain to the −18 LUFS reference

npm install @audio/loudness-replaygain

import replaygain from '@audio/loudness-replaygain'

Track gain per ReplayGain 2.0, which adopted BS.1770 integrated loudness as its measurement basis: gain = -18 LUFS − measured. Delegates the loudness measurement to @audio/loudness-lufs (ITU-R BS.1770-4 K-weighted gated loudness) and subtracts from the −18 LUFS RG2 reference.

replaygain(channels, { fs: 48000 })   // { gain, lufs, peak }

| Param | Default | | |---|---|---| | fs | 48000 | Sample rate, Hz | | weights | BS.1770-4 Table 1 | Per-channel gain array. Defaults by channel count assuming SMPTE/WAV/Web Audio order: 1/2ch all 1.0, 4ch [1,1,1.41,1.41], 5ch [1,1,1,1.41,1.41], 6ch [1,1,1,0,1.41,1.41] (LFE excluded). Other counts default to 1.0 per channel — pass explicitly for layouts in a different order |

Accepts Float32Array (mono) or Float32Array[] (multichannel). Returns { gain, lufs, peak } — gain in dB to reach −18 LUFS, lufs the measured integrated loudness, peak the linear sample peak — or null for silence / fully-gated input (same conditions under which loudness-lufs returns null).

peak is the maximum absolute sample across every channel, including any the loudness measurement excludes (LFE), since it describes what clips on playback rather than weighted loudness. It is 0..1 for unclipped material and can exceed 1 for sources that overshoot. RG2 tags carry sample peak, which is what this is; for 4x-oversampled dBTP use @audio/loudness-truepeak.

The pair is what makes the gain safe to apply: a player raising a track by gain dB should clamp so that peak stays below full scale.

Use when: computing per-track/per-album normalization gain for players and tag writers targeting the RG2 −18 LUFS reference — gain and peak map directly onto the REPLAYGAIN_TRACK_GAIN / REPLAYGAIN_TRACK_PEAK tag pair.

Per the ReplayGain 2.0 specification (BS.1770-based loudness, −18 LUFS reference), built on @audio/loudness-lufs.

Part of @audio/loudness — the loudness family umbrella. Documented from the reference implementation.

MIT © audiojs