How loud is too loud? How loud should I send my mixes to you? Why does my mix sound quiet? Why don’t you master to a higher LUFs? Are LUFs even real; can they hurt me?
Questions pertaining to loudness are some of the most common questions that people ask me about. Many recording professionals will even admit to uncertainty at times regarding loudness. For me, loudness has been something that I have put great effort into understanding, and that I consider to be one of the most wonderful parts of music production. For me, the question “why are some tracks in my iTunes library quieter than others?” was what initially sparked my interest.
To talk about loudness, it if important to first understand what people perceive as loudness in recorded music. A good place to start then is to talk about what people perceive as loudness in a very simple context: a recording of a guitar strumming a chord. When we record a guitar, the loudest part of the recording will typically be the initial strum of the guitar, which we will call the transient. As the chord rings out, as the strings lose energy, the guitar gets quieter, until eventually it fades to silence. How loud will this recording be? How do we measure loudness? There are 2 major areas that we can look at here. The first will be ’peak volume’ - which will be that transient mentioned earlier: the initial hit. The second will be the average volume over time. If we were to by some means make the decay of the guitar louder, without increasing the volume of the transient, what does that mean? The peak volume would stay the same, but the average volume would be increased. Would we perceive one guitar as being louder than another, when the peak volume is the same?
The answer to that question is typically ‘yes, we would perceive the guitar with the higher average volume as louder’. In context of music production, this means that if we say universally ‘peaks can only exist this loud’ (which is more or less limited from a technological standpoint to 0db), what we perceive as difference in volume from track to track is almost entirely then from the average volume of the track. Increasing the average volume of a track while maintaining a constant peak value is typically done through use of compressors and limiters.
So does louder mean better? Do we universally want the decay of the guitar strum to be louder? What about on a fully mixed song? How does this apply?
To a degree, I think many people can intuitively understand the change of feeling from a guitar that rings on forever, maintaining a high volume, to a guitar that decays quicker and falls to silence. Sometimes we want the guitar to ring on, for the feeling it creates. Other times, the ebb and flow of the guitar fading out might create the mood that you are seeking. This concept applies directly to a fully mixed song. Changing the perceived volume of a song, again while assuming our peaks maintain a constant level, will change the feeling that we provide to the listener.
If we believe that different music should provide different feeling, then we should acknowledge that having some difference between ‘average volume’ should be acceptable.
So lets talk numbers. One of the most common ways of measuring loudness is a measurement called LUFs (Loudness Units relative to Full Scale). That is to say, how loud is the program material relative to the full scale, where full scale is the ’peak’. If you are more familiar with RMS volume as your ‘average’ scale, for the purpose of this post you can consider them to be the same (ish). If we go back in time to the peak of the loudness war, the goal of producers was to have their song be ‘louder’ than other songs, so that when it got played on the radio or in the club, it would catch the attention of the listener, at the cost of ’letting the guitar chord decay’. It would not be uncommon for music to hit values of -6LUFs. This means that the difference between the loudest point and the quietest point of the guitar chord is about 6dB. This difference is actually not that significant, and we lose a lot of the feeling of a guitar chord decaying in this example. Again, you can use your imagination, that the guitar chord stays fairly constant in volume, versus trailing out.
Since the peak of the loudness war, many have suggested that values closer to -14LUFs typically allow listeners to have a more ‘natural’ listening experience to what is typically recorded. This is more than twice the total range that has sometimes been utilized. In my opinion, it is not a hard and fast rule that we should always try and maintain volume around here, but it is a good benchmark or starting point. Sometimes we might want to give something more room to breathe. The amount of drama and tension that can be built up by allowing more dynamic range, a byproduct of lower average volumes, can be incredibly powerful. Conversely, there are times when we want the sound to be in your face, aggressive, and loud.
Interesting to note, that many modern streaming services, Spotify, Tidal, Apple Music, Deezer, etc., have algorithms built into them that will regulate volume. The benefit to the listener, is they will be able to hit play on a playlist and not have to worry about having to change volume between songs without risking blowing their eardrums when a song produced louder comes on. Most of the algorithms that these services use aim to target music to around -14LUFs (not exactly).
What ends up happening to a track that was produced to -8LUFs, then? What the algorithms typically do, is drop the whole track volume (peaks and all) by 6db. This means that relative to the full scale, where 0 is the max, the song is now at -14LUFs, but the peaks are at -6dB instead of 0. As a result, you have lost about 6db of drama and tension and feeling that you could have had, relative to a song that has its peaks at 0, and average volume at -14LUFs. This is not always a bad thing. Again, the decision made on loudness has tonal impact. You may still want your guitar to ring out in that way. However, increasing loudness for the sake of being more perceived where people listen has been greatly diminished by these algorithms.
So how loud should music be? In my opinion, there is no hard and fast rule. I have mastered music anywhere from -20LUFs to -8LUFs. As an experiment, I once brought a song to -3LUFs, but that sounded terrible. If the intended release is streaming platforms, I would suggest always trying to stay close to -14LUFs, so that the streaming service algorithms don’t have to adjust your music, and that you know how it will sound. That being said, there are cases to be made for going louder.
I would also like to note again that the peaks are always going to max at 0dB in context of recorded music. The kick drum in your dance track will only ever hit 0dB. People dance to the ebb and flow of music, not just blaring loudness. You do not have to make your song louder just because other artists do. Your song can hit just as hard, if not harder, when you give it some room to breathe.
And as a final note, I realize that I mentioned ‘how loud should I send my mix to you?’ At the beginning of this post, and then completely ignored that point. I usually recommend peaks being between -2dB and -3dB, and LUFS being what ‘feels right’ for you. Understanding, of course, that the louder your song is, the less control I have. If the intent is for streaming, -17LUFs with peaks at -3dB would be already -14LUFs when I bring the peaks closer to 0dB just by applying gain. Different mastering engineers have different opinions, though, so always check with engineer you’re sending to if you are unsure!