AES British Section lectures - October 2007

Headphone measurement

Keith Howard

www.audiosignal.co.uk October lecturer Keith Howard

On October 16th the UK Section of the AES hosted "Headphone Measurement" by Keith Howard. Keith is consultant technical editor on Hi-Fi News magazine and a contributing editor on Stereophile.

Keith started by pointing out that there are several reasons (other than just for the fun of it) that we should measure headphones. The European headphone market is growing at 15-20% per annum with over a fifth of the headphones sold costing more than £60. Today many people's principal exposure to high quality sound is via headphones.

Keith then went into the techniques he is using to measure headphones. He employs a GRAS artificial ear with realistic pinna and IEC 60711 eardrum simulator, to allow measurement of all types of headphones from circumaural to insert. He noted that loudspeaker measurement software is equally useful for measuring headphones. Because the output impedance of the headphone amplifier can have a significant effect on the frequency response at the headphone terminals, it is important that this terminal response be measured and subtracted from the measured acoustic response.

To investigate diaphragm resonances, the headphone can be hung in free-space and the measurement microphone placed in the near field to generate a cumulative spectral decay 'waterfall'. Another significant parameter in some applications is a headphone's isolation of external sounds. Ideally a reverberation chamber is required to measure this but a diffuse sound field can be approximated within a domestic room by maximising the distance between the sound source and headphone and by using an omnidirectional sound radiator.

To determine how good a headphone is from its measurements is tricky as headphones, unlike loudspeakers, are not meant to have a flat frequency response. This means that post-processing of the measurements is required to apply appropriate correction. Keith explained why diffuse-field correction is believed to be more appropriate than free-field correction, quoting research conducted at IRT in Germany. Appropriate fre-field and diffuse-field corrections for artificial ear measurements have just become available as a result of research conducted by Hammershoi and Moller.

Keith mentioned a few potential pitfalls in headphone measurement, including room reflections with open-back headphones, the aforementioned effect of amplifier output impedance and repeatability issues due to external interference at LF and small changes in headphone placement at HF.

Keith finished by showing some early results and commented that the measured data didn't quite line up with the results of the accompanying listening test. There followed a lively question and answer session.

A recording of the lecture is available from the UK section website at www.aes.org/sections/uk/meetings/a0710.html.

Report by Tim Harris