What it does
Give it the measured integrated loudness and true peak, and the spec you must meet.
- Pass or fail. Offset in LU, loudness and true-peak checks against the chosen spec.
- Gain and limiter. The gain to reach the target, and whether it would push the peak over the limit.
- Delivery versus playback. Keeps broadcast specs apart from streaming platform normalization.
Where it helps
For checking files and programmes before delivery and understanding LU, LRA and true peak. Platform targets change and are company policies: verify them. It does not measure audio.
How to use it
- Add the skill folder to Claude as a Skill, or open the universal prompt file and paste it into ChatGPT, Gemini or any assistant (custom GPT, Gem, system prompt, or the start of a chat).
- Describe your problem in your own words. The skill asks the few questions that change the answer.
- Check the answer against your equipment manuals and, where it matters, a measurement.
Limits
A skill makes an AI assistant more careful, not infallible. It cannot see your equipment, it can be wrong, and standards and platforms change. Every skill lists its sources and what it does not know, and shows whether a person has validated it. Mains power, RF power and structural work always need a qualified person.
Changelog
26.10.2 2026-10-29
Corrected the EBU R 128 tolerances (+-0.2 LU in QC, +-1.0 LU where the target is not practically achievable), added live and Apple Podcasts presets, the ATSC A/85 dialogue-gating note and the primary document of each target.26.10.1 2026-10-29
First release. Status: to validate (not yet reviewed by a broadcast engineer).