How loud is it – when we turn up the volume, to really enjoy that physical bass and get the live experience. In plain, objective dB numbers – how loud.
People are curious to know how loud it is when they experience the systems in Room2. So now I have measured sound pressure levels for different music at relevant listening volumes, in Room2.
I have done this exercise before, many times – measuring how loud it is. There is even an article from a few years back in time, where I show capacity requirements for movies and music. This is an updated study, with more information, which can be used to guideline requirements for SPL capacity in sound systems for music. And satisfy your curiosity.
Sound pressure level – SPL – is not loudness
Perception of loudness in music relates to measured sound pressure level in some peculiar ways. Music is not a constant tone or simple noise signal, and the nature of the signal affects how we perceive loudness. Its spectral distribution across the frequency range, crest factor – difference between mean level and highest peak level, and contributions from decaying reflected energy in the room, all that matters.
This means that if we measure 90dB SPL, this can be perceived as suitable volume on one piece of music, and too loud on something else.
Sound pressure is measured in dB rms, and there are different standardized frequency weighting curves for noise measurements, such as dB-A, dB-C. dB-Z means no filtering. To get the rms value from a complex waveform, it is necessary to do some processing, including integration across some time interval. This means the rms value does not represent the loudest short peaks of the sound.
Sound pressure meters must be calibrated, and they should be configurable, so that one can choose at least between fast or slow integration time, and selectable for A/C/Z frequency weighting. For music, we use Z weighting – full range, no filtering.
For our exercise here, we also want to know the level of the highest peaks. This gives us more information about the music signal, and we can use this information to find requirements for SPL capacity.
Calibrated listening level
I always know how loud I play, which is a huge advantage in both testing and demo situations.
My systems are level calibrated, so that the listening volume is equal for same master volume setting. If I change my speakers, I recalibrate, and the volume control then gives the same playback volume, regardless of the new speakers sensitivity. If I go to a different room, with a different system, I know it will be equally loud, for the same master volume setting.
This means that 0dB on the master volume will always play equally loud, on the same piece of music. It does not mean that all music is equally loud, for the same master volume setting.
The master volume operates on a dB scale, where 0dB is calibrated to 82dB rms for a -20dBFS rms pink noise signal, on each speaker.
A demo session in Room2
We start with Carla Bruni at -20dB master volume:
The black line is the rms SPL, with fast integration time, no filter (z-weighting). We see it is not fixed, it changes in level. The red line is peak SPL, the highest registered peak since the reading started.
And continue with September In Montreal at 0dB:
From there on, a variety of music played quite loud. How about some Vestbo Trio, at +6dB:
Jøkleba, Mayshia, +6dB:
More examples
Infected Mushrooms, Avratz, +6dB and +8dB:
This looks very loud, but since the loud peaks are bass transients at low frequency, it is not perceived as uncomfortable.
Drums from Chesky Records, +8dB:
Ever experienced a drum set up-close – this recording manages to reproduce some of that powerful tactile impact, when played back loud enough.
Jøkleba, Popcorn, at +6dB:
This last one is quite loud, probably louder than most will find comfortable for more than a short period of time.
Frequency range
In the How Much Power.. article it is explained how peak capacity does not necessarily reduce with increasing frequency for music signals. While spectral power distribution follows a downwards tilted sloping curve, peak amplitude can still be equal at high frequencies.
I recorded several RTA graphs while measuring loudness for this article. These graphs show the frequency distribution of the sound in the room, and provides information about capacity requirements in different frequency ranges. The graphs include the setting dialog, which obscure some of the data at high frequencies, I chose to leave it there because the numerical processing is important for how the graph looks like. The red line shows the peak level.
Jøkleba, Mayshia:
Flashbulb, Leaves:
September in Montreal:
We see that music has high-level content across a wide frequency range, some music even has content below 20Hz.
Why play loud
I was just sitting by the fireplace in Room2, the only warm place in the house. Perhaps The Mandalorian soundtrack can work as background music. And there it is – the bass. Just have to turn it up, and it gets better and more exciting and joyful the louder it gets. The physical tactile feel of the bass is addictive, no more background music, this is powerful and very present.
Instruments get a presence and realism that simply is not there at lower volume. And the physical tactile feel is not restricted to bass frequencies, the sound has this powerful feel up into the midrange, where instruments like snares and smaller drums hit your body with a sharp kick.
SPL requirements for full-scale sound
To enjoy this, you need to turn up the volume. And you need capacity.
Full-scale sound means exploring the full dynamic range of what we can hear and feel. One of the requirements to make this possible, is sufficient capacity. So that our loudspeakers never compresses transient peaks.
If it gets too loud, it is no longer possible to enjoy the music. But since it will be the peak transients that gives the tactile physical sensations, and sense of power and realism, overall loudness can be reasonable, even when drums have a nice powerful and punchy feel. Program material – the music – with good dynamic range and powerful transients, is another requirement.
On-paper, it should not be too difficult to find speakers capable of around 115dB peak – this equals (115-3-6)dB = 106dB rms at listening position from each speaker. Subtract 6db for distance, and we have 112dB/1m, which is within reach for many speakers – if we calculate from specified sensitivity and power handling. In reality, it is not so simple.
The real world SPL limit is when it still sounds reasonably pleasant, no distortion, no annoying resonances or other things that stick out when you turn up the volume. Exactly how loud this is, depends on the system – speakers, amplifiers, and room acoustics. When you reach for the volume control to turn it down because the trumpet screams in your ears, this limit has been exceeded.
There will be a usable SPL range. Too quiet, and you start to loose low level detail. Too loud, and it gets distorted or unpleasant. This usable range starts where it is loud enough to hear everything well, and stops when it starts to sound bad. Some systems have a wide range – sounds good at very low volume, as well as live concert levels. Other systems may start to sound bad before sufficient loudness to hear everything is reached – an unfortunate situation.
One important aspect here, is that louder volume must add something to the presentation – more excitement, more involving, an overall increase in emotional impact from the music.
Typical SPL requirements for different listening volumes
From the different graphs, we can see that roughly, most music has around 20dB louder peaks compared to rms SPL. We also see, that many recordings are mastered so they give around 90dB SPL rms at 0dB master volume, on a calibrated system. And 0dB can be used as a typical baseline for normal-loud volume.
This table represents rough estimates for SPL and listening volume:
| Loudness | Master volume | SPL | SPL peak |
| Full volume | 0dB | 90dB | 110dB |
| Normal “hifi”-loud | -10dB | 80dB | 100dB |
| Low volume – still hear everything well | -30dB | 60dB | 80dB |
| Very low volume – for late night relaxed listening | -40dB | 50dB | 70dB |
Even louder volumes, above 0dB master, becomes very dependent of production and the music arrangement. Some albums are best played at max 0dB, or they simply get too loud, while others – with larger dynamic variations and higher dynamic range – can be played at +6dB or even at +8dB. Vestbo Trio sounds just right at +8dB. Nublado should always be played at +6dB.
Classical music and smaller ensembles with acoustic instruments usually sound right at -10dB. Too loud, and it is no longer representative for how those instruments and the music actually sound in a live performance. Baroque music, solo violin. Voices of Music, Baltimore Consort, Nicholas Parent Trio.
For typical LOUD pop music productions, lower the master volume around -10dB. Many of those will sound equal in perceived loudness when played -10dB lower, so that Billie Eilish at -10dB master is perceived equally loud compared to Anne Bisson at 0db master.
Conclusions
- 120dB capacity across the whole frequency range 20Hz – 20KHz makes sense.
- Speakers with high capacity is required to do this properly.
- But there is no need to install a huge PA-system in your living room to be able to enjoy music at live concert levels.
- Capacity requirements are determined by peak SPL levels – music at 90 dB needs 110dB capacity.
- Even at reduced volume, small typical hifi-monitor-speakers will struggle.
- You decide your listening volume, there is no definite rule or number for correct level that applies for every person or music or situation.
You are welcome to use the forum to discuss and comment on this article.