Normalize the narration, then mix the video
NarrateHQ's assembled output applies an internal level target and crossfade to make generated sections easier to join. The public preview normalizer also uses an R128 loudness pass with a true-peak ceiling. These are production helpers, not a promise that the final video mix meets every platform or genre expectation.
Once narration is combined with music, effects, or a limiter, measure the final program. The video mix, not the isolated TTS file, is what the audience hears and what a platform can normalize.
- Check integrated loudness and true peak with a meter on the final mix.
- Leave headroom for music and sound effects instead of maximizing the voice alone.
- Listen on headphones and a small speaker for harsh consonants and buried words.
Why a single target is not enough
Perceived loudness changes with delivery speed, pauses, room tone, and competing tracks. A numeric target helps establish a repeatable starting point, while editorial listening catches problems the meter cannot explain.
PRIMARY REFERENCE
YouTube's official loudness guidance ↗