In this article
Two voiceover files can show the same maximum peak and still sound quite different in level. A single loud syllable can determine the peak reading, while the rest of the performance is what the listener hears for most of the recording.
Start with a recording worth keeping
Record a test at the actual distance you’ll use for the read. Listen for room noise, plosives and changes in level across a full paragraph. If a word has clipped or the room is too noisy, turning up the file will bring those problems along with it. Our guide to improving a recording room is a good place to investigate a recurring issue.
Know what the normalise command does
Peak normalisation applies the same gain change to the whole file to reach a target peak. The loudest sound moves and so do the quietest ones. The balance between them stays the same.
A loudness measurement looks at the signal over time and helps explain why a tightly compressed voice can sound louder than a dynamic one with the same peak. When a brief specifies loudness and true peak, check both using the producer’s delivery requirements. The EBU R128 recommendation is one example of a specification for broadcast programme audio.
Decide how much processing the job needs
If the team wants to edit against music or mix several performers, ask what they want left for post-production. If they ask for a finished read, listen before and after any compression at comparable playback levels. Pay particular attention to whether processing has changed the delivery or made breaths and room noise more noticeable.
Listen to the delivered file
Open the export, check the full beginning and end, and listen through a quiet and a loud passage. Confirm the channel layout, format and filename alongside the level checks. The audition recording guide covers how to fit that listen into the time you have for a submission.
One response to "Peak normalisation and loudness in voiceover"
Simple to understand – extremely useful to know!!