Confusion About DB Values Below 750 Hz

I’m a little confused about what should be the DB values below 750 Hz. From my understanding, Typically an audiologist looks for 0 DB at 250 and 500 Hz and normally does not check hearing below 250 Hz.

However the folks at seem to think that the ideal hearing at 750 Hz should be -1.5 DB, at 500 Hz should be -3.5 DB, at 250 Hz should be -11.7 DB, and at 125 Hz should be -25 DB. See chart below of their ideal curve.

What confuses things even further is what Behringer (maker of audio compressors) appears to think. I was checking my MDX1600 audio compressor yesterday to see when compression kicks in and with the threshold set at -10 DB, the compressor started compressing at 52 DB for all frequencies 750 Hz and above. For 500 Hz, it started compressing at 55 DB, at 250 Hz it started compressing at 58 DB, and at 125 Hz it started compressing at 67 DB. Therefore they start compressing audio 15 DBs earlier at 750 Hz and above than at 125 Hz. Also with the peak limiter set to + 3 DB, the audio at 750 Hz and above, hit the peak limiter 15 DBs earlier than did the audio at 125 Hz. The test was not based on hearing but was visual where the input to the compressor was fed different frequency tones increasing the output to the compressor by 3 DB until the compression and peak limiter lights would light and then recording the results.

Therefore it seems that Behringer may think they are using the correct hearing curve and therefore starting compression at different times depending on the frequency. Also it seems they think it is ok to exceed the maximum DB at 125 Hz by 15 DB more than at 750 Hz and above without doing hearing damage.

My current hearing curve at 125 Hz is 7 DB above the ideal hearing curve and 3 DB below the Behringer compressor response curve. If I adjust my equalizer so that curve below 500 Hz remains the same as it was originally and add up to 12 DB gain to frequencies above 750 Hz to try to create as flat of a curve as possible above 500 Hz, the nasal (muddy) sound through my Bose headphones disappears and clarity improves significantly but the pitch of voices are slightly too high.

If I reduce the gain of the 125 Hz frequency to become closer to the ideal curve from, the pitch of voices skyrockets and becomes extremely annoying. If I increase the gain of the 125 Hz frequency to move closely resemble the curve from Behringer, voices tend to sound lower pitched and more natural without any apparent loss of clarity. If I increase the gain at 125 Hz even further, voices tend to become even warmer and more natural.

Am I interpreting the Behringer curve correctly by using when compression starts? Is it also true that the ears can take more DB SPL at lower frequencies than at higher frequencies? Is there an accepted standard for hearing below 250 Hz?

I believe now I understand what happened to create confusion.

Why did voices have such a high annoying pitch when I reduced the 125 Hz frequency gain to follow the ideal hearing curve indicated by is a hearing research and consulting company that consults for colleges and companies around the world. I suspect they developed the ideal hearing curve from testing different people and determined that people with that curve had the best hearing determined by different tests. In my case, I didn’t follow their curve but instead only reduced the gain at the 125 Hz frequency. This created a very bad proportional relationship between the fundamental voice sound and the first and second level harmonics which were disproportionately too high creating extremely high pitched voices. If I would have reduced the gain of the 250 Hz frequency by about 9 DB and the 500 Hz frequency by about 3 DB to follow their ideal curve, I suspect the high annoying pitch of voices would have disappeared.

Why does the compressor start to compress at different levels for different frequencies?

It doesn’t. After a long day of setting up the compressor and equalizer, I accidentally left the equalizer enabled when I was running the tests to determine where compression starts at different frequencies. Naturally with the equalizer producing additional gain for the higher frequencies, the tone generator incrementing by 3 DB along with my eyes watching for compression to start, that would indicate too low of a level for compression to start for the higher frequencies. Upon retesting with the equalizer disabled, frequencies from 30 Hz through 125 Hz indicated 3 DB less than frequencies between 187 Hz and 16 KHz. I contribute that slight difference to the characteristics of my laptop not being able to respond as well at the lower frequencies as compared to the higher frequencies.

I’m wondering if the issue here is a confusion of the dB scale.

There are a wide range of dB scales (dB SPL, dB A, dB C, dB HL etc), each with a specific purpose in mind. So if you start trying to compare HL vs SPL you could certainly run into confusion.

When hearing is tested, the scale is such that every single frequency should be along the 0dB HL line at every frequency. However minus 10dBHL to plus 20dBHL is still considered in the normal range.

This was my thought as well.
Their values of -12dB have no meaning when you don’t know what’s in relation to. What’s the dB scale they are using?

Also, a side note. When an audiologist tests hearing we aren’t looking for 0dB. 0dB is not an absence of sound since the dBHL scale is a logarithmic calculation to be able to translate dBSPL into a graph where normal hearing would be displayed in a fairly flat line. So there is no such thing as perfect hearing. Only normal or not normal.

Well okay to expand on that, the dBHL scale was derived from finding an average threshold level of ‘normal’ healthy human adults, and these hearing levels were defined as 0dBHL on the scale we use.

So by definition, normal hearing is 0dBHL at every frequency. However, there is a normal range of hearing, which is essentially -10dBHL to +20dBHL at all frequencies. Once you start slipping out of this range you clearly have a need for some help.

And as DocAudio said, the purpose of the dBHL scale is to more easily see deviation from the definition of ‘normal’ hearing, since it is pretty easy to see if hearing test results are not a straight line around 0dBHL.

If we tested hearing thresholds on the dBSPL scale (for example), normal hearing would be a curve, peaking just past 2KHz. Comparing a ‘normal’ curve to an abnormal curve would be confusing and less visually obvious.

After a lot more research, I discovered why I had a major problem. First I knew that I had up to a 0-15 DB variation between my left and right ears but I was ignoring that fact because I thought that if I equalized the differences between my ears, I would have a problem with the compressor settings. Generally my left ear was overall better than my right ear.

However after retesting my ears independently again, I realized that my left ear below 250 Hz was fairly flat and well exceeded the ideal hearing curve and my right ear (worst ear) was much closer to the ideal hearing curve below 250 Hz but was extremely erratic (did not have a smooth flow downward from 500 Hz to 30 Hz).

Watching tv with my mixer set to mono to produce the same sound to both ears supported my findings below 250 Hz by listening with one ear at a time. The left ear was louder but much more muddy/nasal sounding than the right ear. The loudness was because the left ear was about 10-15 DB greater than the right ear at 90 Hz and 125 Hz and the muddy/nasal sound of the left ear is caused by the very high flat curve for that ear. When watching tv with both ears, the loudness of the muddy/nasal left ear drowns out most of the sounds of the less loud clearer right ear.

When I previously pulled down both ears by about 9 DB at 125 Hz, I created an extremely bad curve for my right ear causing severe harmonic distortion giving extremely high pitched sounds and then hearing fatigue after 2 hours. Ringing occurred in my right ear only.

So I decided that I need to balance my ears. My first attempt was to pull up my right ear to equal the hearing in my left ear but that was a disaster which created even a more muddy/nasal sound. My ears were nearly perfectly balanced but now both ears were getting the muddy/nasal sound intensifying the muddiness.

I then decided that I would try to follow the ideal curve as closely as possible to see what type of results I would get using my 31 band equalizer (1/3 increasing octave equalizer) with independent sliders for the right and left ears. The following are the curves for my ears after equalizing in relation to 750 Hz (0 DB).

Left Ear

45 Hz = -36 DB
60 Hz = -27 DB
90 Hz = - 24 DB
125 Hz = -21 DB
187 Hz = -15 DB
250 Hz = -12 DB
375 Hz = -6 DB
500 Hz = -3 DB
750 Hz = 0 DB
1.0 KHz = 0 DB
1.5 KHz = -9 DB
2.0 KHz = -12 DB
3.0 KHz - -21 DB

Right Ear

45 Hz = -33 DB
60 Hz = -30 DB
90 Hz = - 27 DB
125 Hz = -27 DB
187 Hz = -18 DB
250 Hz = -12 DB
375 Hz = -6 DB
500 Hz = -3 DB
750 Hz = 0 DB
1.0 KHz = 0 DB
1.5 KHz = -9 DB
2.0 KHz = -15 DB
3.0 KHz - -21 DB

I would have equalized better but the equalizer ran out of capabilities. The left ear is correct at 500 Hz (-3 DB) and 250 Hz (-12 DB) but is too high at 125 Hz at -21 DB (should be -24 DB) and at 90 Hz at - 24 DB (should be -39 DB). The right ear is still more erratic but is ok at 500 Hz (-3 DB) and 250 Hz (-12 DB) but is too low at 125 Hz at -27 DB (should be -24 DB) and too high high at 90 Hz and below. There are also some differences above 1 KHz between the right and left ear.

Since I didn’t have another independent left/right control equalizer available to finish the job, I decided to check out the sound and the clarity of the current settings. The sound was fairly natural but still not 100% right. The clarity of movie channels was clearer than with my ITE hearing aids with my Bose headphones. I haven’t done extensive testing to check out the equalizer since I am still gun shy about hearing fatigue and until I get the curve 100% correct and the sound 100% natural, I’m not going to do any extensive tests.

On Tuesday I am supposed to get another independent left/right control equalizer via UPS. This will be a 15 band equalizer and I will setup that like the 31 band equalizer as the first stage of equalization. The 31 band equalizer will then be the second stage of equalization and will perfect the curve, balance the ears 100%, and increase the high frequencies (1.5 KHz and above) by up to 12 DB. If everything sounds natural, I will then do extensive testing.

Headphone manufacturers understand the importance of the curve below 250 Hz. Bose tries to be as natural as possible but gets criticized by many for lacking bass (look at their reviews and you will see they only get about 3 1/2 stars where most other high end headphone manufacturers get about 4 1/2 stars or greater). Sennheiser typically adds about 10 DB response to the 125 Hz band to get more bass. Some manufacturers add a 25 DB or more response to the 125 Hz band to get a very deep bass sound (pretty useless for watching tv but can be nice for listening to music). I can produce any of those sounds with my equalizers just by messing with the frequencies below 500 Hz. Although that is also harmonic distortion, that will normally not cause hearing fatigue since ratio between the harmonics is decreased in relation to the fundamental sound and that is what happens when someone developes high frequency hearing loss as they get older.

If there were 4 audiophiles all with good hearing checking out headphones, one may say it sounds too muddy, another may say it sounds great, another may say it doesn’t have enough bass, and another may say it has too much bass. The difference is likely that their hearing is different below 250 Hz.

The ideal curve is similar to the ISO curve. The ISO curve is mainly used by manufacturers to design sound equipment.

To give you a little better idea of what I am up against, the following two jpeg files are plots of my left and right ears. In the plots is a line which is the ideal hearing curve according to and I am trying to get as close as possible to that line by using equalizers.

Over the next couple of days, I will be setting up my 15 band equalizer and my 31 band equalizer to try to create the ideal curve. The curve is in relation to -66 DB (the best hearing of my left ear at 500 Hz).

You will notice that my worst ear (right ear) most closely follows the ideal hearing curve below 750 Hz. Getting my ears balanced below 750 Hz as well as trying to obtain the ideal curve below 750 Hz at the same time is quite complex and time consuming.

You can see that my left ear curve is fairly flat below 750 Hz giving me loud muddy sounds for the speaking voice (85 Hz-250 Hz) where as my right ear follows the curve more closely giving me quiet but fairly normal sounds for the speaking voice.

Just finished setting up my equalizer and the results for the left ear are in the attachments at the bottom. The first plot is for the initial left ear, the second is the plot of the left ear after the 15 band equalization, the third is a plot of the left ear with the 15 band equalizer feeding the 31 band equalizer, and the forth is a plot of the 15 and 31 band equalizers enabled and a 9 band equalizer is enabled to add bass (62.5 Hz, 125 Hz, and 250 Hz bands are pulled up with all other bands at 0 DB).

Since only a maximum of 5 attachments is allowed, the right ear plots will be in the following post.

The 9 band equalizer was implemented to improve the sounds of music (added depth), mellow the sounds of tv programs, and to increase the loudness of the speaking voice (fundamental frequency between 85 Hz and 250 Hz). All the equalizers have a button to enable or disable that equalizer so therefore I have a choice of which equalizer(s) I want to use but should have at least the 15 band equalizer enabled since that does the majority of the work to balance my ears as well as develop the correct curve.

With a reduction of 21 DB for my left and a 24 DB gain at the higher frequencies, there is a maximum equalization between the speaking voice and it’s harmonics of about 45 DB with the 15 and 31 band equalizers enabled.

With limited testing with only the 15 and 31 band equalizers enabled, the clarity of movies has dramatically increased. Prior to the equalizers, my comprehension of dialog in movies was extremely low when using headphones/earphones. My comprehension level was so low that normally I would have to turn on “closed caption” to follow the plot of a movie. Having my ITE hearing aids in my ears as well as the headphone on helps but my comprehension level is still a struggle and I can normally follow the plot but I miss a lot of the dialog. With the equalizers enabled, my comprehension level increases to about 95%-99%. I watched segments of several movies that I previously had extreme difficulty hearing the dialog even with my hearing aids and headphones combined. One was the “The Social Network” where I watched the most difficult scene where they were at a nightclub with loud background music playing and they were speaking softly. Before I didn’t understand a word that was said but now I heard the vast majority of the words. Another small segment that I watched was one of the Hannibal Lecter movies which always gave me extreme difficulty in understanding the dialog but with the equalizers, I have very few problems.

The biggest problem with using equalizers is maintaining the perception of loudness. Because of that issue, I have a sound level meter to check the loudness level before I use the system. Although a sound level meter may not give an accurate level when testing headphones, I have never lost the perception of loudness and therefore have the ability to determine what is a comfortable safe sound when using the headphones without the equalizer being enabled. Once I determine that level, I use the sound level meter to determine what is that level. For example, I can play a bunch of songs from Spotify and set the volume to be comfortable. If the compressor is enabled, the songs will likely be compressing (a red light on the compressor will be lite), I can then check the sound level meter to see what it indicates. As long as I don’t increase the volume level beyond that level when the equalizers are enabled and compressing, I’m safe.

Perception of Loudness Issues

Over Driving the Equalizers

Equalizers have to work within a certain range of amplitude and when 12 DB is added to any frequency, the equalizer will tend to intermittently be over driven producing a sound artifact that sounds like “lip smacking” when someone is speaking. Therefore the input level of the equalizer must be reduced by about 5 DB (a level control is on all equalizers). With two equalizers, this causes a 10 DB loss in gain. Just turning up the gain on the headset or a headphone amplifier by 10 DB does not solve the problem since the compressor is set to the threshold level that was previously set. In my case, the threshold level was set to -10 DB which basically restricts the maximum stream level to about 75 dBu, By turning up the volume on the headset by 10 DB SPL with the compressor still set to -10 DB, a loud sound could drive the headset to 85 DB SPL.

To solve the problem, the compressor must be turned down by another 10 DB to -20 DB restricting the stream to about 65 dBu. Then by increasing the volume on the headset or headphone amplifier by 10 DB will not cause any problems since the compressor will kick in early restricting the headset to about 75 DB maximum.

Equalization and Loudness

In order to get my hearing near the ideal curve, I reduced the gain of my left dominant ear by 21 DB at the 125 Hz frequency range (fundamental range of the speaking voice). This reduces loudness but dramatically increases the harmonic ratio between the harmonics and fundamental voice sound. To get the perception of loudness back, I can increase the curve below 500 Hz but for every 1 DB that it is increased, I would lose 1 DB of the harmonic ratio in relation to the fundamental voice sound and would therefore lose clarity.

In my case since my left dominant ear was 21 DB above the ideal curve at 125 Hz and probably about 10-15 DB above the normal hearing curve, I alway kept the loudness of the tv at a relatively low level (in the 40 DB to low 50 DB range). Someone with the ideal hearing curve would likely have the volume set to produce 60-70 DB. Therefore I am just going to have to initially use the sound level meter until I have adjusted my perception of loudness.

Having significantly better hearing than the ideal curve in both ears below 75 Hz has been a curse. Those are normally the sounds of hum in stereo systems, vibrations and floor rumble caused by speakers, hum caused by high tension lines, road noise, and certain noises caused by motors. For the last five years I have lived in the top floor of a condo and all the electrical equipment is placed on the roof (air compressors, heat pumps, etc.) and every once in a while, one of those with bad bearings or loose mountings starts up just as I am falling asleep and I am jarred awake by the hum. Rolling over on my other ear doesn’t help since both ears have superb hearing ability in that frequency range.

Frequencies below 75 Hz is such a problem for sound stage or studio recordings that much of the sound stage equipment has a 75 Hz cutoff switch (equalizers, crossovers, etc.) to reduce loudness below 75 Hz.

Compressors/Equalizers and the Perception of Loudness

A compressor is needed to protect your ears primarily from sounds that you don’t hear (harmonics) or you hear a sound at a low level (musical instrument at a high frequency) while not realizing that a large DB SPL may currently be applied to your ears. By increasing the gain of equalizers at the higher frequencies, you risk the fact that a musical instrument could be playing at that frequency at 70 DB and an equalizer gain of 24 DB would produce a 94 DB SPL sound to your ears without the protection of a compressor. The same can occur with harmonics.

This is a major problem since the compressor compresses all frequencies and not just the sound that is exceeding the threshold. Therefore if a sound of 94 DB is produced (including the gain of the equalizer) and the compressor has a threshold of 75 DB and a compression ratio of 8:1, all frequencies (including the fundamental voice sounds) will have their gain reduced by about 17 DB. As you add more equalization to increase the gain of the high frequencies, the problem becomes worse.

To try to solve this problem, hearing aids have multiple crossovers and compressors. The newest most expensive hearing aids have as many as 16 crossovers and 16 compressors. Crossovers just divides the full frequency range into bands and feeds those bands to different compressors. Now when compression occurs, the compressor only compresses frequencies in it’s band. After compression is completed, all the compressors feed a mixer to put the bands back together into one big band that covers the full frequency range. I suspect that the crossovers in hearing aids produce a fixed size band. So therefore it is likely that if a hearing aid was supporting frequencies between 0 Hz and 8 KHz and had 16 crossovers and compressors, each band would be 500 Hz (8,000/16) and each of those bands could move up and down in gain independently of the other bands.

Therefore to solve my problem, I need multiple crossovers and compressors and a mixer. I currently have 4 compressors (2 per channel) and am currently using two (one for each channel). I also have a 2 way crossover (4 bands total or two bands per channel), and a mixer but am waiting on the cables to test out the crossover to make sure that sound artifacts aren’t produced at the crossover point. If i get a three way crossover and another 4 channel compressor, this will allow me to create 4 bands per channel. That may be enough compressors and crossovers since they are primarily needed on the declining hearing slope and therefore I would likely create a band between 0 Hz and 1.5 KHz, another between 1.5 KHz and 2.0 KHz, another between 2.0 KHz and 2.4 KHz, and the final band to include 2.4 KHz and above. If that is not enough, another 3 way crossover, another 4 channel compressor, and another mixer would be needed to support 6 bands.

Part of what I am doing is to help me hear better but the engineering part of me is doing it to see how far I can push equalization without any major problems. Although one two way crossover would likely solve my current problem, having the system work with more compressors and crossovers may help solve my hearing problem as my hearing gets worse in the future.

The following are the plots for my right ear that I couldn’t put in the previous post due to the limitation on the number of attachments.

I just re-tested and plotted the curve for my hearing aids. The following attachments are the plots for the factory setting and reduced loudness settings for each ear. The gray line is the ideal curve and the red line is my initial hearing test for that ear.

It appears that the manufacturer doesn’t overall add gain to the bands below 250 Hz since the gain below 250 Hz is similar to the initial hearing test. There is a minor discrepancy in the curve below 90 Hz on the left ear. Either the factory may have mis-adjusted those bands or that I made a mistake trying to determine equal loudness for the left ear. Normally I don’t spend much time trying to determine equal loudness below 90 Hz since the tones are more like vibrations than tones and it is difficult to try to relate the loudness of vibrations to tones. Either way, the gain of those bands are relatively unimportant since they produce very little useful sounds.

The biggest issue is that it appears that the factory setting pulled up the 750 Hz and 1 KHz bands above the 250 Hz - 500 Hz bands by 6 DB for the left ear and 12 DB for the right ear. The reduced loudness setting seems to be a much better setting since it has a fairly flat curve from 250 Hz - 1 KHz.

I expected to see the lower frequencies show added gain when using the factory setting since I hear road noises from a freeway that is about a mile away when the windows of my condo are open but that is not the case. I didn’t think that the fundamental sound of road noises were in the 750 Hz - 1 KHz range but that seems to be the only conclusion that I can draw from the plots.

I think I understand the association between the ISO curve and the ideal curve. In the 1930s, a couple of researchers (Fletcher–Munson) developed the first equal loudness contour hearing curve. They did this by putting people in a quiet room without headphones and played tones of different loudness at different frequencies and asked people to tell them when the tones had the same equal loudness. Then manufacturers used those curves to develop audio equipment to produce the ideal curve (flat curve from 750 Hz upward) so that if a person with average good hearing used that audio equipment (eg. headphones), they would have ideal hearing. Therefore manufacturers attempt to produce headphones that under or over respond at different frequencies to produce a flat curve above 750 Hz for a person with average good hearing.

In 1956 after more research and some changes to the curve, it became an ISO standard and then again later changed to a new ISO standard when research determined that the curve was incorrect.

There is a minor flaw with that approach. Although there is a defacto standard that headphones should have a flat response for a person with good average hearing above 750 Hz, there isn’t an ISO standard for the complete ideal curve so it is up to each manufacturer to decide what is the complete ideal curve. A company like has one such ideal curve and I suspect many manufacturers try to follow that curve or a somewhat similar curve.

Therefore when a person with average good hearing takes a hearing test with good quality headphones, the test should indicate that the person has ideal hearing.

Since it would be nearly impossible for the hearing aids speakers at 1/10th to 1/5th of an inch in diameter to respond as well a good quality headphones with speakers of greater than 1" in diameter, the manufacturer of the hearing aids may possibly try to compensate for response flaws in the hearing aid speakers by increasing or decreasing gain at different frequencies in an attempt to produce the ideal curve.

Since the ISO curve is similar but not exactly the same as the ideal curve below 250 Hz, I suspect manufacturers of inexpensive earphones do not attempt to over or under respond to different frequencies in that area to create the ideal curve. Also headphone/earphone manufacturers may want more bass in their headphones/earphones and may over respond at 125 Hz to add more bass.