Have you ever been watching a TV programme, when you mistake the sound of a door bell ringing on the soundtrack to be actually coming from your own door? Here’s a nice demo of a related effect using headphones.

At last month’s Reproduced Sound Conference in Birmingham, I heard a great talk by my colleague Bruno Fazenda related to this video. It was about headphone externalisation. Play a track of music over headphones, and the band will appear to be playing from inside your head. When listening to the band live at a gig, the brain uses cues from the acoustic waves it hears at the two ears to locate where the instruments are playing. For instance, if the guitarist is towards the right, then the guitar sound arrives later at the left ear, and the sound is also quieter at the left ear because of the shadowing effect of the head.
What does the brain do when hearing a stereo recording over headphones? A normal stereo recording is designed to be reproduced over loudspeakers, so when you listen to it over headphones, the acoustic cues are all wrong. Presented with a set of acoustic cues that don’t relate to a real situation, the brain decides the music must be coming from inside the head.

Over the years many investigations have examined ways of adding the correct acoustic cues for headphone reproduction to try and get sounds to externalise. These binaural techniques are particularly effective if headphone tracking is used so the acoustic cues change as the listener rotates their head.
Bruno is looking at a different issue, whether particular sounds are more likely to externalise. If we can be fooled by a door bell ringing on the TV, presumably there are certain sounds more likely to externalise when listening over headphones. For this, he worked with Jamie Newton, one of our final year Acoustic students at the University of Salford.
In the experiment, a person was sat at a desk in our sound proof listening room playing the snake computer game. While playing the game, the person heard a number of sounds that could have plausibly come from inside the test room. For example, there was the sound of the experimenter’s phone ringing, and to make this plausible the phone was visible to the test subjects when they entered the room. There was also a set of implausible sounds, such as a bike bell. After the test subjects were asked to say what sounds they heard and where they came from.
It is important to remember that all the sounds were actually created over the headphones.
The plausibility of a sound had a significant affect on whether sounds externalised and appeared to be outside the head. The externalisation was particularly strong if binaural location cues were added to a plausible sound. For instance, the effect was strong enough for the mobile phone ring, that a number of subjects questioned whether the experiment should continue after hearing the phone ringing, assuming it was a real call.
So it seems that beyond the usual physical reproduction accuracy that audio engineers worry about, sound event plausibility helps with headphone externalisation. Maybe this is an effect sound designers should exploit on their soundtracks?