Normalize audio

• Podcast, −16 LUFS: the level Apple recommends for podcasts, and a good level for any spoken audio heard on a phone.
• YouTube & streaming, −14 LUFS: the level YouTube and Spotify play at. They turn louder uploads down, so going louder gains nothing.
• Broadcast, −23 LUFS: EBU R128, the standard for European TV and radio. US broadcast (ATSC A/85) uses −24.
• Peak, −1 dB: turns the whole file up or down until its highest peak sits at −1 dBTP, which is what “normalize” means in Audacity and most editors. The loudness ends up wherever that puts it.

Normalizing turns the whole file up or down by one amount. Quiet and loud passages keep their distance from each other, so a whisper stays a whisper; nothing is compressed or equalized.

When that gain would push the loudest peaks over −1 dBTP, a look-ahead limiter turns down just those peaks, for a few milliseconds each, and the result shows by how much. A few dB goes unnoticed on speech. Past about 6 dB, drum hits and plosives can lose their snap; a quieter target keeps them.

Normalizing won't even out a quiet guest and a loud host. The audio enhancer adds gentle compression for that, and sets podcast loudness in the same step.

LUFS (loudness units relative to full scale) measures loudness the way ears hear it: low rumble counts for less, and pauses don't drag the number down. AudioRinse measures it to ITU-R BS.1770-4, the standard behind every platform's loudness target.

True peak is the highest point of the waveform, including the points between samples. Converting to MP3 or AAC can push those in-between peaks over 0 dB and clip, which is why platforms ask for 1 dB of headroom, and why AudioRinse checks the peaks at 4× the sample rate instead of only at the samples.