back to Learn

published: Aug-6-2016, updated: Apr-21-2019

post separation

One of the reasons some headphones get other reviews may (partly) be caused by the fact that our hearing is not ‘linear’ at different SPL’s (loudness/volume).

This is a perception effect and not the headphones not being able to reproduce bass at lower listening levels. Shown below the K701 measured at 90dB SPL, 80dB SPL and 70dB SPL. The individual traces are overlayed in the plot so tonal balance changes can be spotted easily (don’t mind the 100Hz and 200Hz wiggles … = hum)90dB, 80dB, 70dB K701 fr changesThe short story is… the frequency response doesn’t (shouldn’t) change at different SPL

Most people have heard of Phon curves  which graphically show how much SPL is needed at different frequencies to get the same perceived loudness.
This can easily be heard when listening to music at different SPL levels.
Equal loud

Above a picture of Phon curves. (taken from the web and edited)
These Equal-loudness contour curves differ from Fletcher-Munson curves and are more relevant to headphone listening.

For instance when listening to music at a very low level (volume) bass seems to have gone much more quiet than when playing the same song at louder levels.
With some music/recordings the bass reaches ‘normal’ levels at lower SPL levels than in other music/recordings.
As a side note: For this reason many audio equipment from yesteryear had the ‘loudness’ button which, when engaged, increased the bass at lower volume control settings. For this trick to work well the sensitivity of the speakers and output level of the music source needed to be ‘correct’ which they rarely were.

  1. Varying bass levels in music/recordings can be caused by a few things during the (final) mixing stage of the recording.Monitors were used with either too much bass or too little bass or a poorly designed mixing room. When too bassy monitors are used this results in recordings with less bass. When monitors were used with too little bass the end result will be a very bassy recording.
  2.  The artists/producers purposely wanted more or less bass.
  3. The final mix was done at low SPL levels (resulting in bassy recordings) or done at too high SPL levels (resulting in less bassy recordings)

That last part is due to the non-linearity of our hearing when it concerns how loud we experience different frequencies at different loudness levels.
When we assume the recording chain was well done and the final mix was done at ‘average’ listening levels of around 80dB SPL (peaks can easily reach 90-95dB here) using ‘flat’ speakers at a few meter distance in a moderately reverbing room than the recording will probably sound pretty realistic and tonally ‘balanced’ when played back at the same listening levels.
This basically means that in the electrical signal bass frequencies are present in much larger amplitudes than mids and treble frequencies.
This is clearly visible in the frequency spectrum of the song below (Keb Mo, the worst is yet to come). The plot is made with audacity, Hanning 2048 and shows average levels not peak levels during the entire song.


What the plot above shows is that bass frequencies (around 60Hz) are recorded about 20dB louder than frequencies around 1kHz.
This is not purely coincidal but because of the hearing properties as shown by the Fletcher-Munson plots.
When you look at the plots below where the Phon curves are overlayed with 1kHz at 0dB then it is obvious that frequencies of about 60Hz need to be 20dB louder (at 80 Phon) compared to 1kHz.

F-M overlayed 1

This means that one does not need to worry about the frequency response curves as shown in the phon plots at all. With music recordings this ‘effect’ is already incorporated INTO the music recording.

That is …when the recording is mixed/produced on ‘flat’ speakers in a studio, with not too much reflections, at around 80-85 dB average SPL by knowledgeable people.
When the mix is done with non-flat speakers and/or at considerably louder SPL than recordings may sound bassier or have less bass than what was heard during the final mix when played back at 80dB average SPL. 80dB SPL is a comfortably loud level b.t.w.
A while back I wanted to know how loud one actually listens and determine how much power an amplifier needs to have. you can download my findings HERE.

Based on these findings the ‘average listening level range’ roughly is ranging between 60dB SPL (very soft) and 100dB SPL (uncomfortably loud). This is NOT the same as the dynamic range or the range we can hear.

Assuming music has been mixed/produced/recorded at 80dB average levels on well made recordings we can assume the sounds we heard during the recording/producing in the studio sound realistic. This means that to us (humans) instruments and voices sound ‘flat’. To illustrate this below a picture where the 40, 60, 80 and 100 Phon levels are overlayed at a 0dB line (at 1kHz).

F-M overlayed 1

This picture shows us that the lower the average listening is the higher the lower frequencies need to be to be perceived equally loud (= tonally balanced).

When we take the picture above and consider the 80 Phon line as ‘audible neutral/flat’ then we get the picture below.
On the left the overlayed phon curves, on the right the phon curves normalised to ‘flat’ at 80dB SPL. This is the level where most music is mixed at in studios (80 to 85dB average)

F-M overlayed

Because we are only interested in the range from 60 to 100 Phon and intermediate steps of 10 Phon are easier to show the effect in 70 Phon and 90 Phon traces are added (interpolated).

This shows us how to compensate a recording that sound realistic at 80dB should be EQ’ed when we want to experience the same ‘tonal balance/signature’ when playing very softly, to very comfortable long haul levels to normal levels and when listening at loud levels and at very loud levels.

As can be seen the sub-bass needs to be boosted by +10dB at very soft listening levels to around +5dB at comfortable levels.
As most people tend to listen at around these average levels a compensation is needed.

At ‘realistic’ levels no compensation should be needed as the recording studio mixed this at these levels and providing the headphone is ‘flat’ from at least 20Hz to 1kHz.
When listening at loud levels bass may be a little too loud BUT as we are miss feeling the bass at these levels some extra bass doesn’t hurt….
BUT one should realise that most ‘bass’ signals usually don’t go below 40Hz.
As can be seen the bass levels only need to be boosted by +6dB at very soft listening levels to around +3dB at comfortable levels.

What the consequences for us, headphone people, means is that if we almost exclusively listen to music at soft listening levels a headphone with about +5dB boost in the bass to +10dB in the subbass area gives a fairly accurate/rich tonal balance.
When we mostly use headphones at comfortable to ‘normal‘ average levels about +2dB boost in the bass to +4dB in the subbass area gives a fairly accurate tonal balance.
At higher levels you don’t really need to compensate lower frequencies.

post separation

Compensation of frequency plots on this website

The headphone plots on this website already have this compensation applied.
A horizontal line in the frequency plot like the one below thus ‘secretly’ has about +5dB subbass boost in reality and +3dB bass boost in reality so the headphone will sound audibly tonally balanced when played at very comfortable (softer) levels (70dB average) which can be listened to for hours on end to just below ‘realistic’ levels.graph baselineThis has nothing to do with ‘room’ compensation but for lower frequencies (everything below 1kHz) does have a similar effect.
Should the plots on this website not have had any compensation in the lower frequency range a headphone that meansures like the horizontal line in the plot above should measure like the trace in the plot below.correction

As the Kameleon amplifier filter modules are based on measurements the music played on this amplifier will sound very realistic at comfortable to normal average SPL levels with a great LF extension down to the subbass area.

post separation

back to Learn