A pair of virtual ears that would help architects single out a voice against background babble in a room, are being developed by scientists.
According to a report in New Scientist, the software-based ears are the brainchild of John Culling at Cardiff University in the UK, and colleagues.
They can be used to transform blueprints into "sound maps" which show how discernible speech would be, compared with background noise, at various positions in a room.
The virtual ears address a particular acoustic problem: singling out a voice against background babble.
Culling's system models how sound should travel in a room, then it works out how two competing sources would be perceived at a range of locations around the room.
The team created models of rooms and assumed their virtual ears would be 1.5 metres above the floor - to simulate human ears.
It was then used to predict whether someone standing at a given location would be audible above a noise source at a third location - a proxy for ambient chatter.
The results tally closely with those obtained when human volunteers were asked to assess audibility in test environments.