How to Read Headphone Frequency Response Graphs | Limits of IEC Couplers

When reading a frequency response graph for headphones or earphones, prioritize checking the broad tonal balance from the midrange up to around 2,700 Hz, and refrain from taking sharp peaks and dips above 8,000 Hz at face value. Once the remaining acoustic tube length to the eardrum reaches 10 mm, a quarter-wavelength resonance stands up near 8,500 Hz, meaning that even a slight difference in the artificial ear standard or insertion depth shifts the high-frequency peak position itself.

What to Prioritize First

This article follows Do You Really Need a DAC or Amplifier? and Output Impedance and Damping Factor Calculations as a practical guide on how to read audio specifications and measurements. After aligning the electrical conditions on the playback source side, the next question is where to draw the line when interpreting a frequency response graph of the headphone itself.

Two endpoint numbers printed in a catalog or specification table, such as “Frequency response: 10 Hz – 41,000 Hz,” tell you nothing at all about tonal balance. For example, if you look at the official product page for the Sennheiser HD 650 open-back headphones (300 Ω nominal impedance, 103 dB sound pressure level at 1 V, equipped with a 42 mm transducer and 38 mm diaphragm), the features section lists “Frequency response: 12 – 41,000 Hz (manufacturing tolerance ± 1 dB),” whereas the specification table at the bottom states “10 Hz – 41,000 Hz.” For the higher-tier HD 800 S (300 Ω nominal impedance, 102 dB sound pressure level at 1 V, equipped with a 56 mm ring radiator transducer, weight 330 g), the table lists “4 Hz – 51,000 Hz.”

The reason the lower limit for the same model splits into 12 Hz and 10 Hz comes down to a difference in definition regarding how many decibels of attenuation from the reference level are included within the passband. Furthermore, even if the attenuation tolerance were standardized across manufacturers, those two endpoint figures only indicate the extreme coordinates where sound output drops off; you cannot tell whether the bands in between are emphasized or recessed without looking at the plotted curve (the frequency response graph).

However, the moment you open a frequency response graph, a different physical constraint appears. Unlike loudspeakers, headphones and earphones generate sound by sealing a narrow acoustic cavity formed by the human pinna and ear canal, so they cannot be measured simply by placing a single free-field microphone in an open room. They must always be mounted on an “artificial ear (ear simulator or acoustic coupler)” standardized under ITU-T Recommendation P.57 (Artificial ears) or IEC specifications. The physical geometry of this artificial ear strictly determines which regions of a frequency response graph can be trusted and which regions cannot.

Boundaries of Interpretation Across Frequency Bands

The band divisions above represent structural boundaries derived from acoustic tube physics and official manufacturer specifications. Even within a single plotted curve, the reliability of the line drawn across the low-to-mid frequencies on the left is fundamentally different from that of the region above 8,000 Hz on the right.

The Ear Canal Acts as a Quarter-Wavelength Resonator at 2,700 Hz

Why can we not read the upper-midrange and treble of a headphone frequency response graph as a flat straight line? The physical reason is that the human ear canal is a “tube closed at one end.” The ear canal is terminated at its inner end by the eardrum (tympanic membrane) and open at its outer entrance. In acoustics, a cylindrical tube closed at one end and open at the other behaves as a “quarter-wavelength resonator,” generating a strong standing wave at the resonant frequency defined by the following equation:

F = v / 4L (where F is the resonant frequency in Hz, v is the speed of sound at 34,000 cm/sec or 340,000 mm/sec, and L is the effective tube length in mm).

For instance, when an acoustic tube has a length L of 75 mm, as in a standard hearing-aid tubing model, dividing 340,000 by (4 × 75) yields a first-mode resonant frequency F of 1,000 Hz. Because a quarter-wavelength resonator also resonates at odd multiples of its fundamental frequency, subsequent peaks appear at 3,000 Hz, 5,000 Hz, and 7,000 Hz following the 1,000 Hz fundamental. By contrast, in an average unobstructed human ear canal, the first-mode standing-wave resonance—known as the Real-Ear Unaided Response (REUR)—appears around 2,700 Hz.

In other words, when we listen to loudspeakers in a room or unamplified acoustic instruments in daily life, the sound pressure reaching our eardrum is not flat; it arrives with the region around 2,700 Hz substantially amplified by the natural resonance of our own ear canal. Unless a pair of headphones produces a comparable pressure rise at the eardrum, the human auditory system perceives the sound as unnatural and hollowed out in the upper midrange. Consequently, a raw measurement graph (uncompensated data) captured on an artificial ear normally exhibits a prominent rise between 2,700 Hz and 3,000 Hz, and judging a headphone to be “mid-centric” solely because that bump appears on a raw chart is a misreading of the measurement baseline.

The 8,500 Hz Standing Wave in a 10 mm Tube Shifts with Measurement Conditions

Even more problematic are the high-frequency standing waves generated inside a sealed coupler when an in-ear monitor (IEM) is inserted into the ear canal. As discussed in What Changes When You Swap Eartips, canal-type earphones seal the entrance of the ear canal and form a short closed acoustic chamber between the nozzle tip and the eardrum.

Suppose the remaining distance L from the eartip nozzle to the eardrum (or to the microphone diaphragm inside an artificial ear coupler) is 10 mm. Substituting the speed of sound v = 340,000 mm/sec and the tube length L = 10 mm into the quarter-wavelength equation F = v / 4L gives 340,000 ÷ (4 × 10) = 8,500 Hz. This means that in the band above 8,000 Hz (specifically around 8,500 Hz), a sharp resonant peak stands up not because of the earphone driver’s own mechanical voicing, but because the 10 mm column of trapped air is resonating inside the tube.

The practical implication of this formula is decisive. Because the tube length L sits in the denominator, changing the eartip size or altering the insertion depth into the artificial ear coupler by just a few millimeters shifts that 8,500 Hz resonance peak to a different frequency. Moreover, depending on whether the measurement rig uses a legacy 711 coupler (IEC 60318-4 series) or a newer-generation ear simulator engineered to approximate human acoustic impedance into the high treble, both the amplitude and the exact frequency of peaks above 8,000 Hz change substantially. The reason we should never treat a narrow treble spike on a single chart as absolute truth is that a single plot cannot separate intrinsic driver behavior from coupler tube resonance.

Never Overlay Graphs Captured Under Different Measurement Environments

Records to Hear Through This Perspective

Once you establish clear boundaries for reading frequency response graphs, hesitation during gear selection drops away. Only within a single database measured by the same operator using the same artificial ear and the same target compensation should you compare relative level differences across the bass and midrange up to around 2,700 Hz. Conversely, treat fine peaks and dips above 8,000 Hz as variables that move with insertion depth and earpad compression.

The recordings where this smoothness and weight from the low end through the midrange directly form the musical backbone are the works connected to The Lineage of Spiritual Jazz. In sessions where the woody body resonance of an acoustic double bass overlaps with the rich overtones of a tenor saxophone, a smooth transition across the low-to-mid frequencies and natural phrasing around the ear-canal resonance band (near 2,700 Hz) matter far more to long-term listening comfort than ultrasonic spec-sheet figures like 41,000 Hz or 51,000 Hz.

If you feel that the earphones in your hand are not delivering the bass weight shown on a published graph, the proper troubleshooting order is to check for acoustic seal leakage and output-impedance voltage division before suspecting your amplifier. A frequency response graph is ultimately a calculated cross-section captured when a transducer is sealed without leaks onto a standardized metal coupler; the moment it enters the final acoustic tube of your own ear, a second set of physical boundary conditions begins.