The volume argumentHow loud should live music be in a pub?

Half past nine on a Saturday. The room is full, the duo is deep into their second set, and a regular you have known for years leans across the bar and says something you cannot hear. You lean in. He says it again. Still nothing, so you nod and pull him a beer he may not have ordered. Somewhere in there sits the question: how loud should live music be in a pub, and who in the building is qualified to say?

Not the person standing next to the speaker.

Who sets your ceiling, and it is not the mixer

Start with what is not a matter of taste. In Western Australia your noise sits under the Environmental Protection (Noise) Regulations 1997, which set an assigned level for entertainment premises of typically around 54 dB(C). That ceiling is not a setting on your mixer, and not judged by how the room sounds to you.

The consequence lands on your licence. Under section 117 of the Liquor Control Act 1988, a complaint of undue disturbance can be brought by police, a government agency, a local council, or three unrelated adults at different addresses. It must be served on you first, with time to fix the cause, so one furious neighbour cannot end your Saturday alone. If it runs its course, the Director of Liquor Licensing can vary your conditions or prohibit you from providing entertainment at specified times.

One part of Perth changes the arithmetic. The Northbridge Special Entertainment Precinct, in force from 6 February 2026, lets approved venues emit amplified music at 90 dB(C) in the core zone and 79 dB(C) in the frame, and puts the attenuation cost on new residential development. Two things get misread: approval is opt-in, venue by venue, and the precinct is Northbridge, not your suburb.

So how loud should live music be in a pub? Loud enough to talk over

Inside that boundary there is no published number. No Australian source sets an in-room decibel target for a pub, so a figure you find online is almost certainly someone else's jurisdiction or someone else's room. What you have instead is a test.

It takes ten seconds and no equipment. Stand four metres from a table, mid-song. Could two people there hold a conversation at a raised but comfortable voice? Leaning in and repeating themselves means you are over. Hearing the fridge means you are under. Run it in the first song, from three spots: the corner by the stage, the middle of the floor, and the doorway, which is what your street hears.

The test works because of what the level is for. A room feels alive when people talk over something rather than at each other across it. Too soft and the act is furniture, paid background nobody looks up for. Too hard and the tables stop talking. That is the costlier failure, because a group that cannot hold a conversation does not order another round. It finds its coats.

Too quiet and the act is furniture. Too loud and the room stops talking. The level you want is the one nobody mentions.

Most of what gets called a volume problem is a matching problem. That is why energy beats genre when you choose an act.

What the research says about turning it up, and why I would not

There is a finding here that points straight at your till. A bar experiment by Nicolas Guéguen and colleagues found that a higher than usual sound level went with customers drinking more, through arousal. It is the one study captured here that tests volume on its own, apart from tempo or genre.

Now the honest state of it. The record is uncertain about the year, approximately 2005 to 2007, and about the country, probably France. It measured drinks and nothing else: not comfort, not whether anybody came back, not where volume starts emptying chairs. Your risk sits at a neighbour's fence line and your margin with the group that stays for a fourth round. Neither was in the study.

Setting the level on a Saturday

Three things, none of which need a decibel meter to start.

1. Agree the ceiling when you book, not at 9.40. A performer told the limit beforehand builds the set around it. One asked to drop mid-set does it badly through no fault of their own, because the arrangement was pitched higher.
2. Listen from the far corner and the doorway, not from beside the desk, which is the least representative position in the building. In eight years as a professional musician I have never trusted my read of a room from the stage.
3. Walk out to the fence line once, at the heaviest point of the night. That is the only listening position your section 117 exposure cares about, and almost nobody stands in it.

Then track the signal that costs nothing: how often your bar staff ask a customer to repeat an order. If it is happening at most orders, the room has stopped being able to talk. The other layer that catches new hosts is whether you need a licence to play music in your bar.

If the level is a conversation you have with an act every week, have it once instead, at the booking. Have a look at how we run venue music for Perth rooms. No lock-in, and every test here works whether you call us or not.

The honest limits. Nothing captured here measures where volume starts costing a venue trade, so the four-metre test is craft rather than evidence. The 54 dB(C) figure is a typical assigned level, not a reading of your own licence, which may carry its own condition. The precinct's commencement date comes from press coverage rather than the government page, which had not been updated when it was captured. The volume and alcohol result is a single bar with no published sample size, no significance test, and an unconfirmed year and country. If a complaint has been served, that is a conversation with your lawyer, not with a blog post.