What’s wrong with headphones?

Many headphones suffer from a limited frequency response, but what’s worse is that the frequency response is basically different from person to person. This makes mixing for headphones a moving target.

The target #

Headphones aim to reproduce an acoustic wave accurately at a specific place. This place is the target.

In-ear headphones target the ear canal opening, because that’s where they are placed. But blocking the ear canal changes its resonant frequency, so in-ear headphones have unique difficulty in achieving this goal. When the ear is plugged, the original resonance is muted, while a higher resonance is produced by the now shortened ear canal.

Is the ear canal opening a suitable target for over-the ear headphones? Frequency response would be highly sensitive to placement over the ears, and the shape of the listener’s pinna.

Calibration #

The acoustic effects of both the pinna and ear canal are well-known to psychoacoustics researchers, and can be compensated for by measuring an individual or physical model (the latter may be preferable for studying the ear canal). A measurement microphone is placed at the target. The headphones are placed precisely on (or in) the ears. Calibration tones are played by the headphones and captured by the microphones. The difference between source and capture tones is calculated. An inverse filter is computed, and is applied to the headphones. A subsequent calibration should demonstrate close to zero reproduction error at the target.

The goal of headphone audio reproduction #

Assume that the goal of headphones is to reproduce sound with minimum distortion at a specific target. This target is located in between where the headphone ends and your eardrum begins. But what are the targets of the popular audio formats we’d like to reproduce?

We naturally (without headphones) hear sound that has been ‘pre-processed’ by the acoustic effects of our environment, body, and ears. Our ability to perceive where a sound is coming from is largely dependent on these effects. Lacking outer-ear spatial cues, headphones can “disembody” the listener.

For some listeners, this may be what they like about headphones: a tiny but incredibly close-up soundstage that exists only in your head. Yet, consensus exists within the hi-fi community that in-ear headphones create a narrower soundstage than over-the-ear, and that a larger soundstage is preferable.

Over-the-ears have a larger perceived soundstage because the sound wave at the ear canal entrance includes the acoustic effects of the individual’s pinna. (Not to mention, over-the-ears sidestep acoustic issues of blocking the ear canal.) But wouldn’t it be even better to then include the acoustic effects of the head and shoulders? The torso? The room?

The design problem of headphones becomes, what acoustic pre-processing is desirable?

Part II will attempt to answer this question by exploring playback formats (stereo, binaural, surround, ambisonic), and how head-related transfer functions (HRTFs), room modelling, and virtual speakers can be used to perceptually improve headphone audio reproduction of certain formats.