Level vs. 'Loudness'

Ever watched late night TV and been blown away by how much louder the shouting of your local used car salesman feels during the commercials compared to the actual level of the show you're watching?

"Save, save, SAVE!! Unheard of prices!.  No credit?  No worries.  No address?  Who cares?  EVERY DAY IS A GREAT DAY FOR SHOPPING AT CRAZY JOE'S QUALITY USED CARS!!!"

You might even have been dozing off during the show only to be traumatized awake by the salesman's shouting.  Why is his banter so loud compared to your TV show?  Well, in truth, it's not actually louder.... though it feels like it is.  Grasping why this is so can help you, as an audio engineer, create perceptively louder mixes without even reaching for a compressor. 

People think that when their audio hits 0dB on their D/A converters it should be officially 'loud,' but they're often mystified to find that, compared to many other mixes, it still feels quiet.  The meters are solid in the red, but it just isn't kicking like it should.

Audio level meters, be they RMS or peak detecting, show amplitude only.  They do NOT, however, tell you what frequencies make up this overall level, or the shapes of those frequencies.  This is why the HPF I wrote about in an earlier post is so important.  It removes unwanted LF energy that would otherwise push those meters up without giving you the benefit of perceived volume.

To demonstrate this point, take a signal generator, and make a 100Hz sine wave at any output level, and give it a listen.  Then switch to 1kHz at the same level -- a LOT more annoying, no?  It feels louder, but it's actual amplitude isn't any higher than the 100Hz tone.  Now, change that sine wave to a triangle wave and see which one feels louder.

What do we learn by doing this?  Simply that our perception of volume is based on the frequency and shape of the audio content as much as by actual output level being produced.

So back to your late night TV and the commercials.  The Federal Communications Commission (FCC) executes rules on how strong broadcast signals can be (absolute amplitude), so in order to produce the highest perceived level broadcasters run their audio through a brick-wall limiter before being broadcast.  This limiter isn't concerned with frequency content so much as the absolute level being transmitted.  The LF rumblings of traffic and ambient noise of the city in that detective show re-run uses up the available headroom during the show, but during the commercial there's nothing but the purely mid-range voice of the car salesman to assault your ears.  Same maximum amplitude for both (because of the station's limiter), but the frequency content is so vastly different that the perceived level is as well.

VU meters are critical in audio engineering (more about metering in a future post), so keep using and trusting them, but in the quest for 'loudness' you must make sure to carve away any and all unwanted content to make room for what your really want your listeners to hear.