Raw, uncorrected frequency responses obtained from an artificial ear do not represent a headphone’s perceived frequency balance. To obtain a 'flat is right' frequency response that's indicative of perceived tonal balance, correction must be applied – but there is historical and ongoing disagreement regarding the appropriate correction to use. Free-field correction is the oldest and largely discredited today. It assumes that the ideal response is that obtained from a sound source located at head height, 30 degrees off the median plane towards the ear, ie the response of a perfect loudspeaker classically positioned for stereo reproduction. Diffuse-field correction, which gained acceptance in the 1980s and still has adherents, assumes that the ideal response is that obtained in a diffuse sound field where sound impinges on the head with equal intensity from all directions. Background on this correction is given here. In recent years the subject of headphone response preference has been extensively re-examined by researchers at Harman International who have progressively refined their suggested target responses (references {2} to [5] below).

Third-octave corrected responses are generated by post-processing the headphone's uncorrected frequency response (averaged for the two capsules). Five different corrections are applied for circumaural (over-ear) and supra-aural (on-ear) headphones: free-field, diffuse-field, Harman 2013, Harman 2015 and Harman 2017. For insert headphones three corrections are applied: free-field, diffuse-field and Harman 2016. The free-field and diffuse-field corrections used are those published by Dorte Hammershøi and Henrik Møller [1]. All the corrected responses are normalised to 0dB at 1kHz.

Test signal:                                                        pink-spectrum periodic noise

Sampling frequency:                                       96kHz

Bit depth:                                                           16-bit

FFT length:                                                        32,768pt

Frequency resolution (measurement):         2.93Hz

Frequency resolution (graph):                        ISO third-octave steps

Reseats on the artificial ear:                           10

Measurement averages:                                 10 per reseat

Measurement parameters (Bluetooth headphones)
Test signal:                                                        pink-spectrum periodic noise
Sampling frequency:                                       48kHz
Bit depth:                                                           16-bit
FFT length:                                                        16,384pt
Frequency resolution (measurement):         2.93Hz
Frequency resolution (graph):                        ISO third-octave steps
Reseats on the artificial ear:                           10
Measurement averages:                                 10 per reseat

[1] D Hammershøi and H Møller, "Determination of Noise Immission From Sound Sources Close to the Ears", Acta Acustica united with Acustica, v94 114-129, 2008 (download)

[2] S Olive and T Welti, "The Relationship between Perception and Measurement of Headphone Sound Quality", Audio Engineering Socity 133^rd Convention, October 2012 (available here)

[3] S Olive, T Welti and E McMullin, "Listener Preferences for Different Headphone Target Response Curves", Audio Engineering Society 134^th Convention, May 2013 (available here)

[4] S Olive, T Welti and E McMullin, "Listener Preferences for In-Room Loudspeaker and Headphone Target Responses", Audio Engineering Society 135^th Convention, October 2013 (available here)

[5] S Olive, T Welti and O Khonsaripour, "The Preferred Low Frequency Response of In-Ear Headphones", 2016 Audio Engineering Society International Conference on Headphone Technology, August 2016 (available here)