Neutral sound, an utopia ?

home
back to Learn

.
post separation

What is ‘neutral’ (e.g. ‘flat’) sound reproduction ?
That one is rather easy to answer and can be short as well.
The reproduced sound should be an (exact as possible) replication of the applied electrical signal without any alterations.

Is ‘neutral’ sound reproduction a prerequisite to enjoy music ?
No, it is not essential that the entire reproduction chain is neutral. Certain ‘coloration’ may even be preferred by many people.
That said, I often found that the reproduction systems that I like/prefer do measure ‘neutral’ or measure close to ‘neutral’ and only have some slight bass boost or show just minor ‘flaws’ in tonality/sound quality.
Not all recordings sound pleasant and realistic on a neutral reproduction chain.

I really like the sound of my headphone as it is… am I missing something ?
Probably not. I you prefer the sound of a headphone as it is and don’t want anything changed about it then … just enjoy it.
That headphone may be neutral or it might not be. but YOU like it so…
Headpones that are not neutral may sound quite pleasant to many people.

What is a neutral frequency response ?
That one is already a bit trickier to answer.
One would assume a transducer (speaker or headphone) with a ‘flat’ frequency response is ‘neutral’.
A flat frequency response means all frequencies are presented in the same amplitude  over the entire audible range.

graph baselineThere needs to be additional info here as the amplitude will always vary over the entire frequency range and thus we have to specify the boundries where the differences should stay in.
For DAC’s and amplifiers it be stay within +/- 0.2dB (between 20Hz and 20kHz).
For headphones and speakers considerable wider margins are needed. There isn’t a single headphone around that measures ‘flat’ within +/- several dB.

This is quite easy to achieve with electronics like DAC’s and amplifiers. Although some people are of the opinion that differences between DAC’s and amplifiers are considerable and quite audible. I have not found this to be the case unless differences are quite measurable.

Unfortunately there is no loudspeaker (incl. the room it is in) on this planet, and certainly no headphone, that can reproduce ‘neutral’ sound with absolute perfection. Regardless of how much money you throw at it. Reason being… acoustics and electro- and mechanical limitations.
But …. it is possible to come close. Close enough to consider something ‘neutral’ in any case. Mostly because the hearing is easily fooled and adapts itself pretty easily.

There are many obstacles in the path of ‘neutral sound’ though … to name a few key bottlenecks:

  • Recording
  • Mastering (dynamic range/EQ)
  • Recording format
  • Reproduction chain (vinyl, DAC)
  • Required power levels (amplification)
  • Listening level
  • Dynamic range
  • Driver issues
    •  Frequency response
    • Phase response
    • Distortion
    • Decay
    • Impedance
  • Ear/head/headphone issues
  • Reference/brain/taste/music genre/training

The individual aspects above are described further down below… its a long read.
When the above peaked your interest, continue to read.

the Recording side

During the recording process we need to consider:
So many artists… so more many ‘tastes/preferences/opinions’ that come with them.
Different ‘music genres’ requiring different approaches in recording technigues.
There are a LOT of quite different recording ‘facilities’ such as studios, make shift studios, live stages etc. and all of them have varying acoustics and equipment.

There are many quite different types of microphones.
All of these microphones have distinct characteristics, are purpose built and optimised for specific tasks/instruments.
They differ in frequency range (tonal response/character) and in directivity patterns as well as maximum allowable SPL (Sound Pressure Level).
For example; a headphone designed to be used directly in front of a kick drum needs quite different requirements from the ones hanging over hi-hats, clipped on quitars, violins, brass instruments or used in front of (bass) guitar speakers, singers, pianos and other instruments. There are multi-purpose-microphones but these are not suited for all recording tasks.

Neutral (all round) microphones do exist but are rarely used for recordings.
That is .. perhaps with a few exceptions in the recording industry. Think of audiophile labels who mostly go to extreme lenghts to produce well-made recordings.
Music genres like classical, jazz, fusion and specific vocal, piano, or small ensembles are well suited to be recorded this way.

So microphones differ in tonal balance and may be far from ‘neutral’. Choosing the correct one for each specific task requires knowledge and experience.
The distance to the instrument, it’s position in the space around it, reflectivity of the environment around the instrument(s) etc. all have a substantial influence on the tonal balance as well as the room acoustics depending on the distance.

Then there is the recording console.
Microphone amplifiers, digital or analog ‘tone control’ and or specific equalization is often used to ‘tune’ the sound of each recorded instrument in a studio to how the recording engineer(s), artists, producers want those instruments to sound, on the used monitors that is.

Now we have landed at the next bottleneck.
The used studiomonitors, placement in the studio itself (size and acoustics), distance to mixing console, and above all those operating it all.
When ‘neutral’ monitors are used (one would expect them to be ‘neutral’) it is possible to create a ‘neutral’ recording. That is, to record each instrument faithfully to how it sounded while it was being recorded.

A recording of a piano, quitar or any other instrument made in an, as sonically as possible ‘dead’ recording room sounds equally ‘dead’ on reproduction as well. More often than not some reverb, echo, hall or other enhancements/effects are used to liven-up the recording.

It would be reasonable to think the final mix of a studio recording would be an exact ‘copy’ of how the instruments/voices were recorded. In about (my estimate) 95% of all studio recordings this will not be the case though.
Reasons: The used monitors (in a room) may not be neutral or the room it is in isn’t.
A too bassy speaker system will ‘produce’ recordings with too little bass, a monitor with too little treble will produce ‘sharp/sibilant’ sounding recordings for instance as the (almost always applied) EQ is applied by ear.
The listening LEVEL which is used (as well as time of day !) has an influence on final tonal balance of the mix, even on very neutral equipment.
When mixing is done at a very high SPL the tonal balance (tone control/EQ) will be different than when done at lower SPL.

The reason for that is the non-linearity of ears (Fletcher-Munson).

Naamloos
The Recording engineer(s), producer(s) and often musicians themselves determine HOW instruments should sound, or how they want them to sound. This may be far from ‘neutral’. Preference, training and experience is of influence here.
For instance a piano player may not only be skilled in playing the piano but is also quite familiar with ‘piano sound’. He may have preferences for certain brands or types of pianos, which in turn may differ from those of other piano players.
Consider how piansists hear the piano… from right in front of their keyboard.
The tonal balance of a piano, however, differs considerably between the pianists position and say from a few, to 10’s of meters away (concert hall).  In a reverberant room it will sound very different than played in the open field, or in a studio.
The pianist thus hears a totally different ‘piano sound’ than an audience.
WHO knows best how a piano should sound when reproduced from just a few meters away (think speakers) or directly into the ears (headphones) ?
The pianist (well trained, always the same position and/or piano) ?
The folks who listen to many live concerts (preferably not amplified) ?
The discerned listener at home, through speakers or headphones, and rarely hearing piano’s being played live ?
The recording guys in the studio(s) ?
The producers perhaps ?
For certain all of them will hear the same piano play the same music but with a different tonal balance as ‘reference’ to how it sounds.
The placement of microphone(s) to record a piano will thus also be of great influence to how it sounds.

For the recording side though the final mix, on studio monitors, will determine how instruments/voices on the final product (album) sound. Thus ultimately the tonal balance is determined by the recording engineers, producer and or artist(s). One can only hope they know what they are doing.
If this last part is up to snuff you can still expect a well made recording of a piano BUT it will almost certainly not quite sound how the pianist heard it during the recording itself nor how the actual microphone in the studio ‘captured’ it for sure.

So … a neutral recording probably won’t sound exactly how it was heard when you would have been present during the recording, sitting a few meters away.  That is if all the recording artists would have been present and playing live all at once anyway. In a lot of studio recordings some musicians singers may not even have met each other at all.

Neutral sound may very well be down the drain here already in cases where it concerns recordings other than the ‘audiophile’ ones.
This is the case for almost all ‘popular music’ where dynamic compression and many forms of manipulation of sound is applied in abundance because of ‘market’ demands.
Dynamic compression has nothing to do with digital (lossy/lossless) data compression of files.

Fortunately for the audiophile who likes to listen to well made recordings, with as little as possible ‘post production’ alterations, there are some labels that have their interest at heart. Those that prefer to listen to genres like classical, jazz, fusion, vocals, small ensembles etc and the occasional pop artists that actually cares AND has a vote in how the album actually sounds, will be able to experience a reasonably accurate rendition of instruments/voices on your hifi equipment when that equipment is tonally ‘neutral’ as well.

recording media and formats

The final recording also has to be on a ‘carrier’ (physical medium) which can be in single channel (old mono recordings), 2 channel (stereo) or multi-channel.
It can be an analog medium such as (magnetic)tape or vinyl or in digital format.
MANY different commercial formatsexist ranging from crappy 128kb/s MP3’s to very high resolution formats such as DSD512, DXD or PCM 384/32 bit.
This does have some influence on tonal balance. At least the smallest bitrates do.

Analog tape: recording speed determines the lowest and highest frequency extension as well as hiss audibility.
Azimuth of the recording and playback head determines high frequency reproduction.
Noise reduction techniques (when used) have an influence in dynamic behaviour of impulse response and frequency response.
Electronics matter; the magnetic field strength on the tape differs VERY much from the actual electrical waveform recorded and reproduced and a difference between them may have differences in tonal balance due to different ‘standards’ used and applied.
The accuracy of the tape speed between recording side and playback side could differ. Small variances in speed (think wow and flutter) created my motors, tape guidance, sticking of tape on the heads, dirt on the heads, wear on the heads etc all have influence.
Some on tonal balance, others on accuracy.
Tape recordings on a simple Nagra (professional taperecorder brand) like those of ‘the pawnshop’ show us that analog tape and a few microphones can capture tonal balance and atmosphere quite well so technical limitations need not be a bottleneck IF used wisely.

Vinyl: record speed (as well as actual speed as speed variances like wow and flutter) and the compression used during the recording as well as the essential RIAA correction all influence the tonal balance of what is present on the actual vinyl record.
Dynamic compression is essential to ‘fit’ a dynamic signal into a groove without the playback needle jumping out of grooves or having quite audible amounts of distortion. Frequency response is limited due to needle shape and cut-off filters during the production.
On playback even more can go wrong, aside from speed issues (speed accuracy and wow and flutter) like the spindle hole not being exactly in the middle of the record. Slip on the turntable mat.
Speed accuracy of even the best vinyl playback systems is hundreds of times worse than the cheapest digital audio players.
Cartridges and needle type used, as well as how they are connected, have a definite influence of the tonal balance.
Capacitance of the arm/pre-amplifier/transformer as well as the source (cartridge itself) and load impedance (input resistance of the pre-amp) can create peaks or roll-off in the treble area (and bass) in the order of many dB’s, which is definitely audible. A rise in FR is often seen and may give the illusion of more air and detail.
Also the (mild) compression used in order to fit music on a record while maintaining a reasonable signal to noise ratio will help in ‘enhanced’ detail perception compared to the actual master that left the studio.
Aside from the frequency response of the cartridge itself (and electronics it connects to) the accuracy and quality of the MM or MC input stage as well as the needed RIAA correction in order to ‘reconstruct’ the frequency response as it was BEFORE it was cut onto the vinyl master may differ resulting in more (or less) bass or bass extension (think rumble filter) or more or less treble and how it extends as well as the amount of distortion added in the process.
So.. don’t expect a vinyl record to be an accurate reproduction of the master tape that left the studio (when not already ‘optimized’ specifically for vinyl)… it simply isn’t.

That being said doesn’t mean vinyl reproduction can not sound good.
In fact, it may even ‘sound’ better to many ears despite the crackles/ticks, uneven frequency response, inaccurate speed, noise, dynamic compression and distortion to a lot of people compared to the sound of the actual recording. More apparant ‘detail, speed and dynamic range’ and the added noise level may agree more with many people than the more ‘sterile’ sounding studio master that lacks these ‘pleasant sounding alterations’ which vinyl does bring to the (turn)table.
Yes, vinyl CAN sound very good indeed (when you discard the noise and crackles) mostly thanks to how our brain handles the (changed) presentation of the vinyl recording.
It may sound ‘much better’ to vinyl addicts but no matter how they ‘twist’ it, more accurate or technically superior it is NOT.

All digital formats have the same ‘tonal balance’ when recorded, provided correct anti-aliasing filtering is applied during the analog to digital conversion of the signal.
Higher bitrates allow a wider frequency response.
128kbs MP3 is quite limited in frequency range (15kHz or less) and some other aspects.
320kbs MP3 is almost the same as CD quality in frequency response (18-19kHz) but can exceed dynamic range of CD which is limited to 16 bits. MP3 with the right encoding and decoding algorithm can actually reach 24 bit resolution.
CD is considered by most (certainly not all) people to be sufficient in dynamic range (difference between loudest and softest signal) and can even exceed the 16 bit barrier for low frequencies by use of ‘dithering’.
In tonal balance 1Hz to 20kHz is possible, you can even ‘record’ 0Hz (a constant DC level) on a CD which is not possible with analog recordings.
Bit depths above 16 bits allow a for an even higher amplitude ‘step’ accuracy which is essential during the recording process to have enough ‘headroom’.

As most people reproduce music at home at ‘normal’ levels the peak levels will be between 90dB and 100dB (average is thus much lower at 70 to 80dB) the actual noise levels and distortion levels of CD already are lower than what can be heard even in the quietest of living rooms.
Noise you can hear during reproduction is always a recording issue (noise levels of microphone pre-amps etc) and NOT a bit-depth/quantization noise issue.
MOST recordings people have at home, or stream, simply are 16/44.1kHz CD quality or lower.
A higher bitrate accomodates a frequency range that extends beyond 20kHz.
Young people can hear up to 20kHz (some even slightly higher), older people may be limited to 16kHz to 18kHz without even knowing it or even MUCH less.
Audiophiles usually claim an extended frequency range provides more ‘reality’ to the recordings rather than hearing the highest frequencies as ‘high pitched tones’.
It should be noted though that the masters used for a lot of ‘high definition’ recordings actually differs from most ‘CD masters’ of the same recording. This may well be responsible for the perceived higher quality of higher bit-rate/depth recordings instead of the ‘added’ frequency extension and sample size being responsible for the ‘higher quality’.
Tonal balance of all formats just below CD quality and highest bit/sample rates is essentially the same. One is not more ‘neutral’ in tonality than the other.
Arguably, highest bit/sample rates CAN provide a slightly more accurate reproduction of the original analog signal before it was converted to analog. The question of the audibility of this remains, except in the eyes of audiophiles perhaps.

So don’t worry about the tonal balance of different digital formats unless you are using low bitrate and/or digitally compressed formats below CD quality. Tonal balance issues between CD quality and higher quality formats are always caused in the reproduction path, the actual DAC used.
Not particularly the actual used DAC chip/method but rather how the ‘filtering’ (or lack of) has been done in both the digital plane as the analog part.
This automatically excludes most ‘slow roll-off’, and generally only post ringing or hardly ringing (when viewing square-waves that don’t exist in real-life recordings), non-post-reconstruction-filtered DAC’s  from being ‘neutral’ as high frequency roll-off in the audible band is quite audible.
Some people actually prefer the rolled-off frequency response and argue it is ‘better’.
Like vinyl and tape though it is NOT better as in a more accurate ‘copy’ of the original sound being recorded, but is more ‘pleasant’ for them to listen to.
Nothing wrong with that, but they shouldn’t say it is ‘better’ because square-waves, that do NOT exist in music, are reproduced with nicer and sharper edges and thus also the ‘transients’ in music must be rendered better. In fact they are rendered worse and have actual ‘steps’ in the analog signal.

Amplifier power / headroom

In order for a headphone to sound any good you need enough power to drive it.
This isn’t as easy as one might think. Yes, some headphones sound fine just out of a mobile phone, laptop, tablet or portable player. These headphones are purpose built for this task and generally have a low impedance and a normal to high efficiency.
Not all headphones for home usage have low impedances or a high efficiency. Some of them require either a much higher voltage or require more current than portable devices are able to deliver.
In this case an amplifier is needed unless the ‘home system’ components already have a good amplifier on board which is able to drive most headphones ‘properly’.

There is much confusion about how much power headphones demand/need.
Most headphones do not require much power to reach decent levels at all.
Think a few mW (1mW = 0.001W) Some need a higher voltage, others a higher current. There are only a handfull of (special) headphones that require much more power. The talk of ‘headphones that are difficult to drive’ is actually nonsense, but heard often.
Headphones aren’t hard to drive but may be driven with either too little (limited by the electronics) voltage or current or simply don’t have a decent tonal balance or need more specialised amplifiers.

To properly drive high impedance headphones (300Ω to 600Ω) a higher output voltage is needed but very little current is drawn.
To properly drive low impedance headphones (8Ω to 50Ω) the output voltage can be much lower but more current is drawn.
Note: Voltage is applied by an amplifier, the drawn current is determined by the headphone’s impedance. Current is not ‘demanded’ as is often said, but the result of an applied voltage and the impedance. Not enough Voltage or current always results in clipping.
Below a scope plot of a clipped signal and the signal when it is not clipped.
The ‘sine wave’ is the unclipped signal, the signal with the flattened top is the clipped signal.
Differences between the signals are a bit hard to make out because they are overlayed.

hardclipping

An amplifier must thus be able to provide enough current and must be able to supply enough voltage in order to drive the entire variety of headphones.

If you like to know how much power, voltage, current, impedance and/or efficiency your headphone needs/has then download THIS pdf file which has data on most of the popular and well known on-ear and over-ear headphones.
There is no info about Bluetooth, Noise Cancelling, earbuds, and IEM’s is in there though.

When you want to find out more about how much power/voltage/current one actually needs, which differs considerable per headphone, you can find info HERE , HERE and HERE.

For a neutral sound sufficient power should be available but more important for tonal balance is the frequency range. The amount of distortion isn’t contributing to the tonal balance, at least not when it remains below 0.1% (some even claim below 0.01%).
In order to have a neutral sound the frequency range should be (at least) between 20Hz to 20KHz within 0dB to – 0.5dB.
It is quite possible to go much lower and higher though. 0 (zero) Hz (=DC), or just a few Hz on the lower part of the frequency range to well over 50kHz, or even above 500kHz, is not so difficult to achieve for headphone amplifiers.
When looking at frequency range specifications sometimes the ‘cut-off points’ aren’t even given. In that case a specification of ’20Hz – 50kHz’ for instance is meaningless as that may well be at +/- 3dB or even more… we just don’t know. This means at 20Hz or at 50kHz the signal may be down -6dB ! This also means that there will most certainly be an audible rool-off already for the (sub)lows and treble frequencies.

In order to reach 20Hz to 20kHz (-0.5dB) you need at least a frequency range of 5Hz to 100kHz (-3dB). This is something one needs to consider.
Some manufacturers even publish frequency ranges within 0.1dB, others specify at -0.5 or -1dB where yet others specify at -3dB (which is kind of a standard).
3dB differences are quite audible already.
Those that specify the frequency range within +/- 3dB or do not specify the cutoff points at all may not comply to the minimal requirements for ‘neutral’ sound reproduction.

For a neutral sound a frequency range should at least be from 20Hz to 20kHz within 0.5dB and distortion levels should stay below 0.1%.
Higher distortion levels may not be problematic though, but this depends on the harmonic spread of the distortion products. Distortion values well above 1% don’t have to be detrimental to the sound quality at all and may in fact add some ‘pleasant’ coloration to the tonal balance. Neutral, this is NOT.

Listening level

This is an often overlooked aspect in the assesment of ‘neutral sound’ but should certainly not be dismissed or overlooked.
In short: Listening levels are very important for judging neutrality of sound.

Many researchers have looked into this aspect, extensively.
At low average listening levels (you know, quietly listening all evening to music) bass levels are perceived softer than when listening to the same music but with the volume turned up. At louder levels (subbass) becomes more apparant.
The most well known research is done by Fletcher-Munson but others have found similar but slightly different ‘curves’.
This is because it is not something that is easily and objectively measured but highly subjective as well, as it involves both actual Sound Pressure Levels and Perceived levels. Training and personal preference play a role here too and so your personal ‘Phon curves’ may differ somewhat from the averaged ones you see published.

Naamloos
Fletcher-Munson curves have been updated in 2003 because of new insights, so even ‘standards’ change over time due to progress in research.
What these curves basically show is how much sound pressure (and thus also electrical amplitude) is needed for different frequencies to be ‘experienced’ as equally loud compared to 1kHz.
Each line is drawn based at a specific SPL level at 1kHz and the resulting lines/curve shows how much more (or less) actual sound pressure is needed to perceive a 100Hz tone equally loud as the 1kHz tone. These lines represent Phon levels.
so a tone at 1kHz with a level of 80 Phon is perceived equally loud as a 10kHz tone of 80 Phon and a 60Hz tone of 80Phon.
The ACTUAL soundpressure level, at 1kHz, of 80Phon = 80dB SPL = 80dBA  but at 60Hz you need about 100dB SPL = 70dBA to hear it as 80Phon.

A tone at 1kHz with a level of 60 Phon is perceived equally loud as a 10kHz tone of 60 Phon and a 60Hz tone of 60Phon.
The ACTUAL soundpressure level, at 1kHz, of 60Phon = 60dB SPL = 60dBA  but at 60Hz you need about 90dB SPL = 60dBA to hear it as 60Phon

The observant reader will have noticed that the actual SPL at lower frequencies is MUCH higher at different Phon levels.
Another interesting thing that can be seen is that the dBA values correspond pretty close to the Phon values at around 60Phon but deviate at other Phon levels.

dBA weighted values are based on the (old) Fletcher-Munson curves at 40Phon.
The dBA standard is fixed though (ANSI S1.4-1981) but the Fletcher-Munson curves did change so when put side-by-side now the dBA curve is almost exactly the opposite of Fletcher-Munson curves at 60Phon.

graph-dBA

When average SPL levels are measured to determine (safe) levels they are given in dBA. This differs considerably in value than when measured in dB(C) or dB(SPL) when it comes to music signals due to the amount of low frequencies (bass) which are ‘weighted’ less.

The implication of this is that we do not ‘hear’ music in dB (SPL) but in Phon.
Meters on your player/PC/tape recorder etc., however, show their values in dB where 0dB is either a certain voltage level, saturation level (tape) or the maximum value of a digital signal.
Quite confusing … all these dB’s. HERE is some more info on this subject.

VU meter kl
Basically when we are looking at the dB meter on the computer or DAC actually you see the peak levels of the electrical waveform of the analog output signal. It may be obvious when looking at the meters while listening that the largest ‘swings’ of these meters coincide with the bass or bass drums.

When looking at the amplitude in the frequency domain of a recording (see the plot below) one can easily see the lowest frequencies (bass) are MUCH higher in amplitude than those around 1kHz (mids) and the highest frequencies (treble) are again much lower as higher frequencies mostly consists of harmonics which generally are lower in amplitude.
spectrum

At 60Hz for instance the amplitude is about 20dB higher than that around 1kHz.
This is not purely coincidental but due to the Phon curves. In music we perceive the bass about equally loud as other instruments/singers so the electrical representation of the music signal itself also has larger amplitudes in order to make the instruments sound balanced. This is, of course, around 80Phon at which monitoring levels the better studios mix their albums.

When we have a look again at the Phon curves we can see that the Phon curves show us how we perceive the electrical signals of those levels as ‘equally loud = flat = neutral’.
But the slope/angle of the Phon curves differ at the various Phon levels.

When we were to overlay the phon plots (see left picture below) from different Phon levels and normalize them at 1kHz it becomes even clearer that at different SPL levels the amplitude of the needed amount of lower frequencies also differs.
The louder the music the better we hear the lowest frequencies. When the average levels become softer relatively higher amounts of lows is needed.

As recordings are (should be) mixed at around 80Phon and assume the studio has reference grade speakers that measure ‘flat/neutral’ (always at all frequencies and SPL levels) and the producer has a good set of ears and scruples we can consider the 80Phon curve as ‘flat’ in a thought experiment the plot on the right below appears.
F-M overlayed

What the plot above shows is that when we perceive the tonal balance of a recording as ‘real/neutral/flat’ at levels of around 80Phon average then we won’t hear the same tonal balance (realism/neutrality) of that recording/music when we evaluate the speakers (or headphones) at 70Phon, 60Phon nor as well as at higher average levels of 90 Phon or at disco levels of 100Phon average.
One could conclude that it isn’t the best thing to evaluate the tonal balance for neutrality later in the evening when playing much softer for instance… unless the headphone has an almost unhealthy ‘warm/bassy’ sound at louder levels of course.

There are technical implications for the needed ‘headroom power’ of amplifiers as well because of this. This is why you need more ‘power’ available than most people ‘calculate’. Further down below there is some info about power levels.

This means that we perceive ‘neutral’ sound differently at different average SPL’s.
What would sound ‘real’ at 90 Phon will sound ‘thin’ at 70Phon and what sounds ‘real’ at 70 Phon will sound way too bassy/fat at 90Phon.

Dynamic Range

When talking about average Phon levels one should realize that music consists of peaks as well as very soft levels around above and below those average levels.
Below an oscilloscope shot of peak actual peak levels over a period of 20 seconds.
The softer signals are not visible, they are present as small modulations on the larger swings and so many dB lower that you cannot ‘see’ them in linear scale plots like the one below.

uncomfortably loud

What is obvious here is that a music signal is Dynamic (has large differences between peaks , average levels and soft levels. It thus consists of a large voltage Range.

The DR rating says something about the difference between measured average SPL and peak levels present in that recording but says nothing about how soft the lowest signal levels are.
A recording/song with a DR of 15 will have an average SPL that is 15dB lower than the peaks. Some popular and rock music has a DR of just 5 or even much smaller and sounds highly compressed (sound is almost constantly present at the same loudness).
Highly compressed music sounds ‘better’ to the general public (not being music lovers and audiophiles) at lower listening levels (speakers of small radios and mobile phones) and in ‘noisy’ environments such as cars, buses, bars and at parties where a lot of people are talking. This is where most music is played so that’s the ‘target’ of the popular music industry.  This is a different ‘audience’ as the audiophile music lovers.
Highly compressed music often has the bass levels compressed or even clipped in order to ‘fit’ all those different amplitudes in a smaller dynamic range.

This means one should not evaluate speakers and headphones using recordings that have a low DR. Simple because the tonal balance of the recording is ‘f-d up’ and thus far from realistic.

Missing bass

For headphone listening (what this site is mostly about) there is another aspect to this.
When listening at ‘realistic’ levels and louder levels part of the ‘bass sensation’ we get when listening to speakers is due to ‘feeling’ the bass with the body (torso mainly).
We can feel the bass and this ‘tactile input’ to our brain adds to what is heard with the ears, creating the total listening sensation.
In most rooms, and with most speakers, the lowest frequencies heard and felt are around 30-40Hz. Below that the SPL drops off quickly.  In that part of the frequency range certain frequencies can be substantially boosted while frequencies slightly higher or lower may be attenuated substantially.
Also the reflection of sounds from walls (ceilings and floors included) also add to the perceived amount of bass.

Speakers that measure flat in anechoic conditions will reproduce the most realistic sound. These speakers, however, are never used in anechoic circumstances but always in living or other listening rooms. Those rooms are all echoic.
What happens is that all sounds emitted from the speaker arrive at our ears directly AND bounced off from walls, ceiling, floor and stuff in the room.
The higher the frequency the more directional the sound becomes. Also our hearing is quite capable to separate the direct sound from echo’s. This works well from a few 100Hz and up to about a few kHz.
Below a few hundred Hz our brains cannot separate direct from reflected sound. The long wavelength combined with the speed of sound and distances involved is the root cause.
The lower frequencies fold around the speaker and becomes a point source (omni-directional). Therefore the lowest frequencies bounce of walls, floor and ceiling and all ADD where mids and higher frequencies do not add because the brain can separate direct from reflections. Because of this a speaker that MEASURES flat in an anechoic room will have a raised bottom end in a ‘normal’ room.
This too will be the case in a studio when doing the final mix.
When one considered that ideal speaker in a normal room as ‘neutral’ that means that the bass will be raised.

Headphones do NOT exhibit the same behavior. All sounds are coming directly to the ear so bass is not added. A headphone that measures flat in the bass thus sounds bass shy compared to a flat speaker in a room.
To compensate for the absence of the room reflections for the lower frequencies this part of the frequency response has to be elevated in order for the headphone to sound as good as a good speaker in a normal room.

This phenomenon has also been studied by Sean Olive and Todd Welti. They came up with a +4dB boost below 60Hz (for headphones) which is quite similar to my findings and those of Golden Ears. Personally I find the Harman curve a bit too much so the plots on this website have slightly less compensation and slightly different in curve.

Below the plot of the applied compensation for measurements on this website.
A headphone that would measure ‘flat’ as in equal sound-pressure for each frequency will look exactly as shown below. A little bit too bass-shy as it were.

corr plot

Headphones don’t exhibit that behavior and thus can sound quite different from speakers. Something to keep in mind when comparing speakers to headphones when it comes to neutrality and tonal balance specifically in the bass region.
Headphones can reproduce MUCH lower frequencies (subsonic) than (most) speakers can.
Frequency responses down to 5Hz (0dB) are possible when a good seal is present.
The ‘bass input’ for the brain, however, is missing that tactile feel (which adds to what we hear as a total ‘sensation’) when listening to headphones.
For this reason some boost (below 100Hz) is needed to get a similar ‘sensation’ as speakers at 80 Phon but now solely by input from the ears (and hairs on the ear and skin close-by).
That (sub)bass-boost thus sort-of compensates for the lack of tactile sensations coming from the body. Some people do not feel that this is needed, others may feel some extra bass boost is needed in order to sound ‘more realistic’. Its a personal preference thing.
Websites like Golden Ears and researchers like Sean Olive and Todd Welti (Harman) each have their own idea of how much (sub)bass boost is needed. Below 2 ‘target plots’ where Golden Ears is of the opinion that only the subbass needs to be boosted by approx. 5 dB and bass frequencies (40-80Hz) just need a mild boost (approx +3dB).
Research Olive-Welti did, based on listeners preferences (averaged over many people), concluded that more bass boost (40-80Hz) is needed (approx +5dB) and some treble roll-off as well.

OW-GE2

Personally I find the O-W target to have too much bass boost and treble roll-off when listening to well recorded music. With highly compressed popular music the O-W target makes a lot more sense.
The Golden Ears target sounds much better to me with well recorded music but makes a lot of popular music sound a bit wanting in the bass and slightly too bright.
Below my target curve that will produce an audibly ‘flat’ response in the plots on this website.

graph-bass correction

A headphone without that extra bass boost thus may be ‘neutral’ to person A  but may be ‘bass-shy’ to person B.
A headphone with 4 to 5dB bass boost may be ‘bassy/fat’ to person A but will be ‘neutral/full bodied’ to person B.
Then there are people that perceive the tonal balance somewhere in the middle of this or crave to hear even more bass (Bass-heads). For them 10dB (sub)bass boost is the minimum required in order to find the headphone listen-able.
So… when someone feels a headphone is bass-shy, bassy or neutral to them …. don’t take their word for it and listen for yourself or let measurements be the judge (if you dare).

This means that in order to perceive a headphone as ‘neutral’, at an SPL of 80Phon, we may need some extra (sub)bass in order to compensate for the ‘missing’ tactile feel.
Meaning that when an equal amplitude of frequencies is applied to the headphone the acoustical output (SPL) for the lowest frequencies should be approximately 3-4dB higher when we average the ‘tonal balance preference’ of the majority of people.

Taste and music genres

Let’s face it… those that listen to classic, jazz, and blues recordings will probably know that the bass levels on these recordings seem to be higher than those of rock and certain ‘popular’ music recordings.

Rock music is (generally) recorded for those that like (live ?) rock music which is generally listened to out louder levels. The target audience in, general, is less concerned with how long their ears will last and crave ‘live levels’ on their headphones.
Rock concerts are always LOUD in reality, ear damaging levels are ‘normal’.
For this to sound right a either a ‘neutral-ish’ headphones (or one with a dip around 3kHz to tone down screechy sounds at higher volumes) a higher volume level is needed. At lower volume a raised bass will make this sound better. At soft volumes it needs even more raised bass, bass-head types of headphones are needed.
A DR between 2 and 7 is quite common.
The tonal balance usually is ‘clarity’ and fast/tight but not too loud bass.

Compressed pop music has the bass levels compressed a bit (Fletcher-Munson) in order for it to sound ‘loud’ on radios (compared to the annoying adds with which the loudness wars took off) and played in cars and while traveling but NOT at loud volumes but rather at moderate levels.
It needs boosted bass to compensate for the dynamic compression at moderate volumes to sound ‘more realistic/fuller’. At louder levels a dip around 3kHz, some roll-off at higher frequencies and some bass boost will sound good. At softer levels ‘bass-head’ type of headphones are needed.
Pop music, just like rock music in live stages is usually played loud so compressed music will sound ‘bassy’ with tight bass in that case. Feel the bass on the chest type of loudness. a DR of 4 to 10 is normal here.
The tonal balance usually (not always) is that of punchy but not raised bass (because of compression) but still ‘feels’ like it has bass.

Blues bands in bars usually play louder than the rumor the crowd makes in the venue they are playing in but less loud than rock musicians.
Bass is slightly elevated and the music sounds high quality and dynamic and with a louder volume that is loud but can be listened to for a while.
The target audience thus listens at ‘realistic’ levels to ‘normal levels’. A DR between 7 and 15 is normal.
Tonal balance is usually ‘flat’ to ‘somewhat raised in the bass’.

Reggae music is not played very loud as well and is bass oriented. It has a very bassy/warm tonal balance by itself and the target audience likes to play it on boomboxes (which don’t so low bass that well) and is not listened to at higher levels in general but more as a relaxed background music. a DR of 5 to 12 is normal here. Tonal balance is quite bassy and with softer treble.

Jazz music generally is a bit softer in volume levels compared to the other genres and may not even be amplified (that much).
The target audience thus listens to that music at home at normal to soft volumes.
So the recording will have higher bass levels in order to sound real. a DR between 7 and 15 is normal. The tonal balance is ‘flat’

Classical music usually is not amplified that much if at all. It is being played dynamically between soft and loud. How loud the volume is also depends on where you sit in the concert hall.
The target audience at home likes to play at normal to soft levels, rarely at real concert levels which would be very loud and very dynamic (because of high DR)
The tonal balance usually is between flat and sometimes needing some treble lift.

Of course there are lots of other types of genres but most of them will fall somewhere in the examples above what concerns levels and used forms of compression/DR.

So different ‘genres’ have different target audiences that generally play at different volumes at home and also have ‘preferences’ in bass levels because of this.
This requires different listening levels to have a ‘live-like’ experience and also different tonal demands at other volume levels.

This means the headphone tonal balance and volume play a different role for different genres.

A more bassy recording on a bassy headphone (Vmoda M100 for instance) at a louder volume level sounds ‘bloated/fat’.
A more bassy recording on a neutral headphone at a louder volume sounds just ‘bassy’
A more bassy recording on a bass-shy headphone (K701, DT880 as an example) at a louder volume sounds ‘just right’

A neutral recording on a bassy headphone at a moderate volume sounds ‘bassy/full’
A neutral recording on a neutral headphone at a moderate volume sounds ‘just right’
A neutral recording on a bass-shy headphone at a moderate volume sounds ‘snappy/tight but wanting in the bass levels’

Loudly recorded (compressed) music on a bassy headphone at a softer volume sounds ‘about just right’.
Loudly recorded (compressed) music on a neutral headphone at a softer volume sounds ‘a bit thin/bright’.
Loudly recorded (compressed) music on a bass-shy headphone at a softer volume sounds ‘deprived of bass/cold/ like 5#it’

One could also write it as:

A more bassy recording on a bassy headphone at a soft volume sounds ‘good/full bodied’.
A neutral recording on a bassy headphone at a moderate volume sounds ‘good/full bodied’.
Loudly recorded (or compressed) music on a bassy headphone at a ‘normal’ volume sounds ‘good/full bodied’.

A more bassy recording on a neutral headphone at a soft volume sounds ‘good’.
A neutral recording on a neutral headphone at a moderate volume sounds ‘good’.
Loudly recorded (or compressed) music on a neutral headphone at a higher volume sounds ‘good’.

A more bassy recording on a bass-shy headphone at a soft volume sounds ‘lacking in body and warmth’.
A neutral recording on a bass-shy headphone at a moderate volume sounds ‘decent but lacking foundation’.
Loudly recorded (or compressed) music on a bass-shy headphone at a higher volume sounds ‘thin/lacking fullness’.

The above examples show that recording ‘quality’ and audience target will yield different ‘tonal balance’ effects with differently tuned headphones at different volume levels (SPL).

So it would appear that different genres may or do require different tonal balanced headphones when listened to at other than intended volume levels.
You can thus use different tonally balanced headphones to suit genres/personal taste at different average volume levels.
But also different genres can sound good on a single headphone (with a specific tonal balance) but will sound ‘right’ at different volume levels.

Genre, taste and volume levels are thus essential ingredients to what one finds ‘realistic’ at their preferred listening level (volume).

Measuring headphones

When me ‘measure’ headphones we can simply show what SPL comes from the driver. This is what most D.I.Y. measurement rigs exactly do (assuming the used microphones are linear which often they aren’t exactly) and display that amplitude content for increasing frequencies. On these measurement rigs a raised bass in the plots will be perceived as ‘neutral as through speakers’. That may be ‘more correct’ when we want to measure the actual SPL but doesn’t jive with our perception of the tonal balance.
Another way to make the perceived tonal balance ‘visible’ is by compensating for this perceived loss of (sub)lows in the plots.
This way a perceived neutral sound (as if speakers in a room) will show a ‘horizontal’ line for all frequencies while in reality is having a mild (sub)bass boost when looking at the actual measured SPL.

Measurements made with HATS (Head And Torso Simulator), and the measurements on this website (Using a D.I.Y. flat coupler compensated for lower frequencies), have this ‘sub-bass compensation’ already included in the plots.

Headphone issues

This leads us to an even bigger bottleneck. The headphone itself.
A headphone basically consists of a few ‘components’ which all influence the tonal balance.

  • The driver itself
    • Type (planar, dynamic, electrostatic, ribbon)
    • The diameter
    • The materials the membrane/cone is made off
    • The magnetic (or electrostatic) motor driving the membrane
    • The suspension or tension of the membrane
    • Mechanical damping
    • Impedance/capacitance
  • The pads
    • size/diameter/shape
    • depth
    • materials used
    • seal
    • damping/bounce
    • compliance of the foam
  • The space around the ears (in case of circumaural)
  • The seal one gets (leakage caused (sub)bass roll-off which can be rather extreme.
  • Position (and size) of the driver opposite the ear-entrance.
  • Angle of the driver (slight influence, not as big as made out to be)
  • The ear-driver distance.
  • The way the driver is ‘loaded’ on the rear (open, (partially) closed, materials behind it etc.)
  • The size, angle and position of your pinnae inside the cups.

All of these aspects have an impact on the tonal balance, not just the drivers or the pads.
This leads to many questions about the usefulness of acoustical headphone measurements in particular.
Do the measurements correlate with what is heard ?
Well… when looking at graphs of similar headphones on different sites you will certainly find some correlations between various measurements but also (sometimes even substantial) differences.
Assuming product variation isn’t the root cause of these measured differences, then it stands to reason to question the validity of measurements when trying to ‘evaluate’ a headphone based solely on measurements. Differences between different measurement methods can amount to many dB’s which would be more than audible if these differences were really there.

Some may say… product variances could be the root cause especially when one considers the many reviews of the same headphone which can be all over the place.
However, I suspect the reported differences in sonic ‘descriptions’ vary because of many reasons.

  • Inexperienced versus more experienced ‘reviewers’.
  • Differences in ‘preferred tonal’ signatures (taste)
  • seal
  • positioning
  • clamping force
  • pad compression
  • age of the headphone(pads)
  • choice of music,
  • different average Phon levels
  • different amplifier output resistances
  • different music sources etc.

These factors will all result in many different views about the performance off the same headphone. Who is right and who is wrong ?

This means that due to many different circumstances a single headphone can be found to be ‘neutral’ but by others may find it ‘bassy’ or ‘thin’ or ‘scooped’ or ‘midrangy’ or whatever tasty description can be given to it.
So it would appear that personal opinions are not the best base on which to judge whether or not a headphone is ‘neutral’ or ‘colored’.

One could say that the best way to asses if a headphone is neutral is to compare it to (known good) speakers with the same music playing equally loud at the same time.
So when we put on the headphones while playing speakers the tonal balance doesn’t change (obviously the stereo ‘imaging’ changes dramatically).
When they sound closely the same in tonal balance, when putting the headphones on and off, that headphone sounds neutral to you, when referenced to your speakers in your room with your music on your preferred SPL. But what if the reference speakers (in the room) are not as ‘reference-neutral’ as one thinks it is, based on supplied graphs of the speaker, in MUCH different circumstances. In that case your ‘known reference’ may not be exactly that… a reference.
This comparison trick may work better with (active) near field monitors as the influence of the room is much less… (unless you are looking for your speakers in room sound)

Since all listeners, rooms, average SPL’s, speakers (positioning, distance, angles and dispersion pattern) differ this means that not all listeners will end up with the same set of headphones he/she is finding ‘the most neutral’.

This is where Olive-Welti have focused their research on.
It would appear that, in general, when all results are averaged a preferred curve would appear close to the O-W target from Harman for loudspeakers.

OW-GE2
This means boosted bass, in the case of O-W, is a bit too much for my ears when using headphones, and the slightly downwards sloping higher frequencies feels a bit too ‘subdued’ for my middle aged ears.

compensation and smoothing

When a headphone is measured the ‘raw’ signal will need some compensation. HATS and Ear Simulators need corrections for the used artificial ear-canal and Pinna. These are made according to standards.
For many reasons, above 1kHz, there are (sometimes substantial) differences between different measurement setups and different brands of HATS even though they all apply to certain standards.
Compensations may differ between brands and/or slightly different standards being used.
So above 1kHz there may well be differences in measured amplitudes at various frequencies. The overall (when heavily smoothed) measured signal should not differ that much though if the compensation and microphones are well matched.

Not smoothed plots show aspects about narrow peaks and dips which could give some clues (not definitive answers) to how the treble response of a headphone is but makes it more difficult to asses the tonal balance.
Some of these peaks and dips may very well not be there or are not perceived that way when music, or even test-tones, are being played and one is trying to evaluate the frequency response by ear. Ears aren’t the best absolute value meters out there even though many people feel they are more capable than any measurement system out there.
Overly smoothed plots may give a better ‘feel’ of how experienced listeners find the overall tonal balance of the headphone but ‘remove’ the information about local peaks and dips that may (or may not be) there.

Below a frequency plot of the right channel from a Beyerdynamic DT150 but with different types of smoothing applied. No smoothing, 1/3 octave smoothing, 1 octave smoothing and acoustical smoothing (acc. to REW) applied.
The heavily smoothed plot does show the tonal balance but fails to show the soar points. This smoothing is rarely used though. No smooting may show wiggles that almost certainly will not be perceived that way. The 1/3 octave smoothing differs from the acoustical smoothing but both show there are some ‘problems’ in the treble area but suggest a different tonal balance.

smoothing DT150

Even though the smoothed and not-smoothed plots do show differences it still tells us that this headphone has a few dB bass boost (which may not be a bad thing) but the wiggles in the treble range also tells us that even though the tonal balance is not that bad on average the treble will sound uneven/coarse as some harmonics may be accentuated or attenuated by more than 5dB which is audible for sure.

Below the king of ‘tonal  accuracy’ (above 500Hz) the HD650. Here we can see that the wiggles in the treble range are much smaller in amplitude. It is also very easy to see how the tonal balance of this headphone is. Slight upper bass/lower mids boost giving it a slightly warmish sound. Bass starts to roll-off below 40Hz. The sub-bass extension (the ‘thump and rumble effects’ are somewhat (and audible) below par. The highs are accurate and the top end extension just rolls-off ever so slightly. Smoothed and not-smoothed plots don’t differ that much with a good headphone.

smoothing HD650

To determine the overall tonal balance smoothed plots may tell you more but less smoothed plots will tell you more about nasty peaks, resonances etc. which also affect how ‘real’ and ‘pleasant’ a headphone sounds.
That is IF the measurements are done on a well made measurement rig by someone who knows its limitations and how to use the measurement device.

Speaking of which. Tyll (Inner|fidelity) has an incredible amount of high quality measurements which are known to not correlate that well to how the headphones in question are perceived.
With this I mean everyone interested in graphs and headphones can clearly see that even the brightest sounding headphones appear to be rolled-off in the treble which of course they aren’t.
While this doesn’t matter that much when comparing headphones on that website it is very hard to ‘see’ how the perceived tonal balance is. Even when you ‘know’ how other measured headphones you know quite well sound in comparison. Measurements of Golden Ears and Rtings are much closer to ‘reality’.
In case one is wondering how to make Tyll’s plots correlate to reality (he used the wrong compensation) then below a compensation suggestion on how to ‘read’ Tyll’s frequency plots compared to a ‘neutral reference line’.
comp curve IF

To give two examples on what this correction of Tyll’s compensation does below the plots of known ‘neutral’ headphones.
on the top the original plot (with the green reference line is drawn in) and below it the plot when that the signal from 1kHz upwards is de-tilted so the ‘neutral reference’ now is a straight line.

4 x Tyll compensated

Of course there is more to to realistic sound than just the frequency part of it.
The phase between channels and between the lower and higher frequencies also should not vary too much. And if it varies then this variance should be gradual.
This is because for sound localization phase differences between left and right (> 500Hz) are even more important than amplitude variations.
Then there is also the ‘speed‘ of the membrane. How fast it ‘stops’ when the signal is no longer present is also quite important for a realistic sound.
Distortion also is a factor.  As long as it is reasonably low (below 0.1%) it doesn’t have that much influence on tonal aspects.
Even though the phase, speed and distortion contribute to the overall sound quality they do not participate in the perceived tonal balance.

ears and brain

Your personal reference also plays a huge role in what one perceives as neutral.
When you visit live performances of your band and try to recreate the same ‘feel’ and sound from the live recordings than most likely you will prefer headphones and speakers with raised bass when playing at ‘normal levels’ unless you play music very loud.
When you have always listened to music with the bass turned up high you will probably never like ‘neutral’ sound but feel a headphone with insane levels of bass sounds ‘good’.
It will sure be different from someone who likes jazz and plays music at home on a high fidelity system.
Simply because the ‘reference’ your brain is set to by the listening habits. One could call this taste but part of taste is determined by the reference you have.
Different references and tastes thus determine what one will see as ‘neutral’ based on listening only.

In the assessment of ‘neutral’ sound there is the ear and above all the brain.
Not all ears are created equal. Pinna sizes, shapes and ear canal sizes and geometries differ. Some people may have ‘gaps’ in their hearing (hearing damage) or are afflicted with other types of hearing loss without even knowing it. Surely this alone has an influence on what we perceive as ‘neutral’ in some way.

Hearing is not like eyesight in that one realizes it when they cannot read or see as well as the used to or as might be expected. When hearing slowly becomes less and less over the years it is much less obvious for us than it is for declining eyesight. A gradual decline in high frequencies isn’t something we realize unless one really starts to hear badly or tests their hearing.
As the brain constantly ‘calibrates’ the hearing, because we are constantly hearing real sounds, even with rolled-off hearing (due to age ?) our brains still hear ‘natural sounds’ as ‘natural sounds’ and thus also ‘accurately reproduced sounds’ as equally ‘natural’.
Diminished hearing thus isn’t as bad for our listening experience when it comes gradually over many many years. A sudden loss of hearing abilities of course is something that the brain won’t/can’t compensate for.

Another problem with deciding what is ‘neutral’, when listening to a reproduction of music in general, is sometimes caused by lack of experience. Experience with knowing how real life instruments and voices actually sound.  Lack of a real ‘reference’.
When one has always listened to old English speakers (which are not known for their crisp sound) and consider that sound ‘neutral’ because that is the only ‘reference’ the brain has then, a truly neutral sound, would probably sound ‘bright’  and ‘cold’ to them.
Add to that things like:

  • the mood you are in
  • the time of day (or should I say evening/night ?)
  • being tired or not
  • lighting conditions(bright Fluorescent light or candle light)
  • being alone or in company
  • and perhaps some other physical aspects

These aspects do all change how we perceive everything around us (not just hearing by the way) and certainly the sound quality and tonal balance. This mix of aspects really makes it difficult to determine, by ear alone, what ‘neutral’ and reference quality when you do not have the proper ‘reference’ around.

modifications

Modifying headphones can be a way to improve certain aspects that need to be addressed or one likes to change. After all NO headphone/speaker is perfect.
You can modify headphones with various materials and different pads.
A disadvantage is that when you change one thing something else that doesn’t need changing, or you don’t want changed, might change as well. And not always for the better.
Also you may need some experience to do this properly and do not have to be afraid to break or damage something.
When dealing with expensive headphones this may not be the best thing. Also when changes are made the headphone may sound better to you but may be further away from ‘ideal’ headphones. mission accomplished anyway I would say when you prefer that newer sound signature.
From my point of view modifying headphones only should be undertaken when it is not destructive and has more benefits than downsides. It is best to use a proven concept (mods done my many others before who all feel it is an improvement) or has been verified by measurements. Again … ears are quite poor spectrum analyzers as references can (and do) vary.

Equalization

When you don’t want to modify your headphone (which is probably so for the majority of people) or you don’t have the means/willingness to buy the more expensive flagship headphones but would like to ‘improve’ the sound of the headphones you already own equalization CAN be beneficial.
Equalization can only make changes in tonal balance and can remove peaks. Some dips can be filled in (or partially filled in) as well.
Dips usually aren’t as sound degrading as peaks are so you may want to leave them alone or just compensate them partly.

Not all headphones react positively to equalization as well. You can’t simply equalize a cheap headphone ‘flat’ and it would magically turn it in a great one. There are other aspects (like resonances, break-up, partial vibrations etc. also at play).
Lots of dB’s correction, and/or elaborate EQ using many notch filters etc. will almost never lead to satisfactory results and improved sound quality even though the tonal balance might be much better.
You should only equalize headphones that just need a tiny amount of correction(s) and the filter slopes used should be gentle (as opposed to sharp notch filters).
It’s these headphones that are worth it to equalize and will improve in sound quality.

Below a few frequency plots of some headphones with and without EQ.


For instance the HD800 (plot top left corner). Some find it perfect as it is. Others hate the elevated treble. Others find it lacking in the sub/bass/fullness. Yet others (like me) find it is lacking in fullness AND find the elevated treble a problem.
If you do not intend to modify it (which still can not make it perfect, just make it perform somewhat better in certain areas) then EQ may be your friend in order to address the tonal deficiencies only. In this case the overall tonal balance and elevated treble/treble peak.

There are several ways to apply a tonal correction in order to make the headphone be closer to neutrality or your personal preference.

  • Digital EQ like a plug-in in a (computer/Phone/tablet based) playback system. You do need almost expert knowledge to apply the correct EQ and that should be based on measurements. The ear is a poor measurement tool.
  • Use the tone controls in the playback system you already have. These tone controls are often quite limited and possibly don’t work in the frequency band(s) you need to address or don’t have the correct bandwidth/slope.
  • Use dedicated software like Accudio , Sonarworks or Dirac that have ‘generic’ compensations for several known headphones. This is easy, you don’t need expert knowledge. You only need to know how to install plug-ins like this. Sonarworks is not free software. Accudio is a much cheaper option.
    Not all players support these solutions.

In all 3 cases above digital manipulation is the key word which means:
The overall volume is lower because in order to boost frequencies you need to attenuate the  entire signal first, otherwise the digital signal will ‘clip’.
The digital signal from the source is no longer ‘bit perfect’ (which some feel is important)
Unwanted side effects due to complex digital filtering may possibly reduce the sound quality when the used algorithms aren’t as good as one hopes/expects they are.

Another way to go about this is to use analog tone control or EQ. Here too you need some expert knowledge to apply the correct EQ and the available devices (multiband analog equalizers or parametric equalizers) are expensive and large. Also these devices often add noise which may even become audible.

For this reason I designed an analog equalizer that uses ‘plug-in modules’ for different headphones. The supported headphones are rather limited because either I haven’t measured them or they don’t react well to equalization or it is pointless to EQ certain DSCN0159headphones. This solution is more expensive than software though.
The Kameleon either doubles as a headphone amplifier (portable or desktop) or needs to go between a source (line level) and an existing headphone amplifier.
The advantage is you don’t need to modify the headphone (remains stock, retains its value better). The tonal balance gets closer to ‘neutral’ (perfect is not possible). The digital signal remains untouched (bit perfect) if that’s important to you.
Of course, the modules change the tonal balance closer to ‘neutral’ as I see it, which works very well for me, but may not be the tonal preference of others.
This analog solution is called ‘Kameleon’ and is not available in any shops (yet). So far it is a DIY project that requires mechanical work and soldering as well as sourcing the parts yourself. A commercial built version is being worked on.

conclusion ?

Neutral electronics gear (DAC’s, CDP, DVD, Vinyl, amplifiers) are easy to obtain, some devices have certain colorations so one has to pick them with care.
Neutral speakers are also available BUT may not be that neutral in (living) room conditions than they are in an an-echoic room measured at 1 meter distance under an exact 90 degree angle. Nor will they be when the manufacturer measured them free-field in a simulated listening room.
Nearfield monitors (that are actually used nearfield as well) are a logical choice when looking for a ‘reference’.
Neutral headphones are not available in abundance and the better ones are usually expensive. Neutral-ish headphones can be found in a lower price bracket.
It may not be the wisest thing to trust other peoples opinions on what sounds ‘neutral’ to them. They may be right but may also be very wrong about this. Your preference may not be the same as theirs.
Neutrality in headphones can be measured, and shown in frequency plots, providing the measurements and compensation (if needed) is done properly.
Don’t let anyone tell you, you NEED a neutral (measuring) headphone… you don’t.
You may not even like it at all.
Chances are though that some people do like ‘neutral’ and even prefer it (as I do).
For a headphone to be enjoyable they certainly don’t need to be neutral. You may prefer a less neutral sound better for many reasons (taste, music genres, low listening levels etc.)
Headphones that just need some minor corrections to come closer to a neutral sound can be used when modified or EQ is applied (digital or analog).

In order to be able to tell, with a reasonable certainty, that a headphone is ‘neutral’ it is recommended to have the conditions below covered.  Having a short listening session with many ‘unknowns’ (say in a hifi-store) while not meeting all the variables below may not be the wisest thing.

  • Use a known well made recording (preferably one you know well).
  • It is quite beneficial to know what real instruments sound like.
  • Play the music through a good DAC.
  • Use an amplifier that can properly drive the headphone.
  • Position the headphone properly/carefully as some headphones are very sensitive to positioning.
  • Ensure to have a good seal, press the headphone against the ear to check. When the tonal balance changes considerably you have sealing problems.
  • Evaluate while being relaxed and well rested, preferably in the evening.
  • Play at realistic, yet comfortable, levels or at the level you mostly listen at.

When the conditions above are met and you really like the sound of that headphone chances are that headphone is neutral or at least that tonal signature is close to it and right for you in any case.

You are the only one that will probably listen to that headphone so who cares if it actually is neutral or not anyway. Just enjoy it..

listing of (fairly) neutral headphones

€10.- to €50.-
Koss KSC35/PortPro (not really neutral but isn’t too far off)
Sennheiser PX100-II (not really neutral but isn’t too far off)

€50.- to €200.-
Audio technica ATH-M50X (bass somewhat boosted)
Beyerdynamic DT250-250, DT150 (with DT100 pads)
AKG K181
Roland RH-300, RH-A30

€200.- to €500.-
AKG K712 (slightly lacks sub-bass), K267
Audio Technica ATH R70X (slightly lacks sub-bass)
B&W P7
HiFiMan HE400S (lacks sub-bass unless Focus A pads), HE500 (slight lack of clarity)
Sennheiser HD600/HD650 (lacks sub-bass)
Shure SRH1540 (slight bass boost)
V-Moda M80

€500.- to €2000.-
AKG K812 (slightly on the brighter side)
Sennheiser HD630VB (bass in middle setting)
(Newer) Audeze LCD-2 rev 2.1
HiFiman HE-Edition X, HE-5*, HE-6* (slight lack of clarity)
mr. Speakers ETHER flow (open version)

above € 2000.-
Abyss AB-1266
Audeze LCD-X, LCD-4
Focal Utopia
Grado HP1000 / HP2 (only 2nd hand)
Hifiman HE1000
Sennheiser Orpheus HE60 / HE90 (only 2nd hand) / HE1060 (Orpheus-2)
Stax SR007 /   Omega 2 / SR009 + amp / SR-L700 + SRM-353X

.
post separation

home
back to Learn

Advertisements