Every public-address system is ultimately assessed on one thing: can people understand the announcement. Level and coverage are means to that end, not the end itself.
Where intelligibility is lost
Intelligibility degrades from three directions, reverberation smearing consonants, background noise masking them, and loudspeaker placement that puts more energy into surfaces than into listeners. Adding power addresses none of the three, and often makes the first one worse.
What actually improves it
In practice the wins are unglamorous.
- Bring loudspeakers closer to listeners so less gain is needed.
- Aim them at people rather than walls.
- Use more, quieter sources instead of fewer, louder ones.
- Cross over sensibly so the mid-range that carries speech is not being asked of a driver struggling to reproduce it.
Specify it, then measure it
Specify the target before installation, measure it after, and the conversation with the client stays factual rather than subjective.
