"For one thing, dB SPL alone doesn't tell you the nature of the sound you're hearing and this is essential in understanding restaurant noise. Rock concerts can be as loud as jackhammers, but not many people pay money to listen to a jackhammer. An analogy could be measuring a liquid when preparing a recipe. A cup of wine and a cup of milk measure the same, but they don't have the same effect on the taste of the dish."
– Tobi Szuts, Research Scientist, Meyer Sound
Not long ago, diners visited restaurants bringing an empty stomach and good company. Now many also carry a sound level meter or a smartphone app to sample ambient noise levels; others plan ahead by checking the noise ratings in restaurant reviews before making a reservation.
But do simple sound level measurements truly reflect the comfort levels for diners' conversations? How accurate are these numbers? And what do they tell us about the auditory experience?
These questions—of increasing concern to restaurant owners along with their architects and interior designers—are unpacked by three sound experts at Meyer Sound: Senior Scientist Roger Schwenke, Senior Acoustic Engineer Pierre Germain, and Research Scientist Tobi Szuts. All three contributed to the Libra and Constellation acoustic systems recently installed at Bay Area restaurants to help diners converse easily over a comfortable background buzz.
Schwenke: It points out the growing importance of the restaurant noise issue, and raises public awareness regarding how your auditory experience affects enjoyment of a meal. It's hard to enjoy your evening when you can't understand what's being said across the table. The readings can be useful to some extent if the measurements are done properly and consistently with good equipment. Some reviewers are probably doing this better than others.
Germain: Accuracy of equipment is the first concern. For example, a smartphone is obviously not a precision calibrated instrument designed for measurement, and on top of that, different apps often show widely varying results even on the same phone—sometimes by as much as 10-20 dB. That said, some can be reasonably accurate if used consistently and with care. As one example, readings should be done at ear level and with the arm fully extended, to minimize the effect of your body and the table. If you put the phone on the table, your reading could go up by 3 dB, and up 6 dB if it's on the table next to a wall.
Germain: It's important to be aware of the factors that impact the published noise ratings. For example, where was the reviewer sitting? If he was next to the kitchen and picking up noise from fans and refrigeration, that might affect the reading. But that in turn might depend on what settings he was using in the measurement app.
Schwenke: Most apps will have fast and slow response settings. The fast setting has a 125-millisecond averaging time and is most useful for observing the minimum and maximum levels. The slow setting has a one second averaging time and is more stable estimate of the average when measuring the constantly varying sounds in a restaurant.
The other important setting is the weighting, either A or C. Weighting curves are applied to either emphasize or limit different parts of the frequency spectrum, and are indicated by dBA or dBC. The A weighting curve is patterned off the sensitivity of the ear, and emphasizes the speech frequencies while limiting the contribution of lower frequencies to the overall measured level. When we are concerned with speech intelligibility, the A curve is more useful and is generally the best for this application.
Germain: That's true, and we should also point out that there are circumstances where the A curve might be misleading. For example, if you're sitting next to a fan or refrigeration unit that produces a low drone, it may have minimal impact on speech intelligibility, but it could prove annoying—particularly if it's cycling on and off. This would show better with C weighting than with A.
Schwenke: The reading will indicate the sound pressure level—abbreviated as dBA SPL—in that space at that moment. This is a measurement of the average acoustic pressure created by the sounds relative to the threshold of hearing, which is 0 dB SPL. Normal conversation is 50 to 60 dB SPL, a passing truck may hit 85 dB SPL, and a jackhammer is around 110 dB SPL. Restaurants seem to be all over the map, from 60 to 100 dB SPL, but in any case that number is only part of the story.
Szuts: For one thing, dB SPL alone doesn't tell you the nature of the sound you're hearing and this is essential in understanding restaurant noise. Rock concerts can be as loud as jackhammers, but not many people pay money to listen to a jackhammer. An analogy could be measuring a liquid when preparing a recipe. A cup of wine and a cup of milk measure the same, but they don't have the same effect on the taste of the dish.
Germain: Also, some sounds are inherently more annoying or distracting at the same measured level. If a couple at the table next to you is having a heated argument at 72 dB SPL, it will be more distracting than ocean surf or classical music at the same level.
Szuts: Our ears are much more sophisticated devices for interpreting speech than any microphone. We have two ears, and our outer ears are acoustic devices that make us sensitive to the distance and direction of sounds. That allows us to sense the direction and distance of a sound source. It gives our ears an unmatched advantage in distinguishing speech from irrelevant noise. This phenomenon is called spatial release from masking.
Szuts: I'm not sure that's a fair appraisal. Acoustics in restaurants is a difficult matter, in large part because it's a moving target. For example, you may want livelier acoustics around the bar, but with mo