Why We're Talking Loudness
Releasing music is complicated. One of the reasons it's complicated is that it can be really hard to know how loud to release a song. If you make your song too quiet, many listeners might check out and skip to the next song because they feel it sounds "boring". But if you make your song too loud, the audio quality really suffers.
This is pretty easy to solve for those with money: hire a great (loudness-war-aware) mastering engineer and let him/her decide. But for those of us who do our own mastering, usually for budget reasons, this is something we have to get a handle on.
A Brief History Lesson
For better or worse, when you hear two songs side-by-side, the louder one tends to sound better. It pops out more, it seems to have more layers, and it somehow just pleases the ear more. You can experiment with this at home, even with two copies of the same song. It's just part of how we hear music.
When records were predominant as the media format for music, engineers had to keep levels quiet to prevent turntable needles from skipping grooves, and to increase how much playback time could fit on a record. And when cassettes were at their height, most engineers aimed for a happy medium of volume, facilitated by analog gear designed with headroom in mind. Due to this, music was distributed to the consumer without much reason to compress the dynamic range.
And dynamic range is important: when your drums can pop out of your song, the mix just sounds bigger and better. There need to be loud instruments for soft instruments to sound quiet by comparison. And there need to be quiet sections in a song for a loud section to feel powerful. To a point, dynamic range is a wonderful thing that makes music sound louder and better when you control the playback volume. Without quiet, there can be no loud.
The CD was the first digital music media format with widespread adoption. And in digital, volume works fundamentally different: instead of gradually sounding worse with added volume, as with cassettes and analog gear, digital gear sounds perfectly clear with added volume until it reaches its max, called 0 decibels full-scale, or 0 dBFS for short. It is impossible to store a digital sound louder than 0 dBFS. When engineers began working in the digital domain, they lost the reference point of 0 dB in the analog domain, a reference point created with lots of headroom for louder sounds to poke through.
The Loudness War
When working with CD as the final destination for music, it quickly became common practice to "normalize" the volume of a CD, bringing the volume of the entire album up until its loudest moment became exactly the loudest possible volume the CD could contain: 0 dBFS.
Engineers quickly noticed that a song on a normalized CD popped out more from the speakers than a song from a non-normalized CD. Louder could sound better, and could be better at grabbing the attention of listeners. A louder song would pop out more than other songs when heard on the radio.
Surprising to some, soft acoustic music, especially music without drums, seemed to sound louder than percussive music. How could that be? Without strong transients in the audio, normalizing music to the loudest possible volume brought the average level of a folk song higher than the average level of a rock song. Artists, record labels, and engineers couldn't have this, releasing rock albums quieter than folk albums, so they started pushing limiters harder in the mastering chain, to achieve a louder perceived sound at the expense of dynamics and with the addition of distortion.
New technologies only encouraged this race to loudness: multi-band compression became a dangerous tool for loudness, and look-ahead brick-wall limiters allowed music to be made louder than ever before.
When one artist at one label had a louder sounding album, other artists and labels scrambled to produce even louder albums. And the loudness war was born, reducing the dynamic range and increasing the distortion of most all recorded music over the last twenty years and counting, getting worse every year.
You really need to hear it to understand. This video explains it far better than I ever could. And if you want to make a difference, signing this petition at Change.org will help.
Hope for Relief
For decades, radio stations have managed the loudness of all the songs played on the air: for them, it's a delicate balance of playing music as loud as possible while not incurring fines from the government. Unfortunately, the audio quality of radio stations is quite bad, with very compressed, very distorted music as the end result.
Things got easier for some when a few media players began including functionality for adjusted volume normalization. Proper design could scan an album for its average volume level, then turn playback volume down to achieve a target average volume level. If CD #1 was on average 3 dB louder than CD #2, then CD #2 might be turned down 10 dB for playback, and CD #1 might be turned down 13 dB for playback. This was great for minimizing volume variation while playing music from your computer, if you used such a media player. But it did nothing to encourage artists and record labels to release quieter music.
I don't know the full history of which streaming service went first, but as of this writing, YouTube, Spotify, Apple Music, and Tidal all stream with normalized volume. What this means is that if you release a song at a somewhat conservative -12 LUFS or a very loud -10 LUFS or a punishingly loud -8 LUFS, it will sound the same volume through these streaming services after their systems work their magic, robbing you of any volume playback advantages from pushing your music too hard into a limiter. Hopefully this is enough to get the attention of record labels, towards the goal of having all artists release music with less limiter-induced distortion and more dynamic range.
Where You Come In
You've made it this far. You know a fair bit about how music started becoming louder, and why that's not a good thing. But if you don't have proper metering, and if you haven't established a measurable loudness goal for your own music, you're still in the dark. Your releases could still end up far too loud, with all the consequences of loud music, or too quiet to foster much audience engagement.
What can you do to keep your music from sounding bad? For the love of donuts, don't push it hard with a limiter! That's the "Keep it simple, stupid" version.
And if you want to learn how to better strike that fine balance, you'll definitely want to read my post explaining LUFS and loudness targeting for mastering. This article provides the "why", that article provides the "how".