problem
Spoken-word consistency comes from measuring dialogue, peaks, and noise floor before each export.
Different guests, microphones, rooms, edits, music beds, and ad inserts can change perceived loudness between episodes.
Peak normalization alone does not solve it because a quiet voice with one loud plosive can still peak high and sound low.
Check integrated loudness across the episode, then spot-check dialogue sections for consistency. Watch true peak so processing and encoding do not create delivery problems.
Noise floor matters too, because raising a quiet voice can also raise room tone and edits.
Use clip gain, gentle compression, and automation before relying on the limiter. Keep music beds low enough that speech remains the loudness reference.
Meter Core helps keep spoken loudness, peaks, and noise floor visible while preparing podcast exports.