Spatial audio teleconferencing

1. A spatial sound conferencing server comprising:

an input processor for receiving audio and position and orientation information from at least one network client;

a conference processor for spatializing the audio according to the received position and orientation information;

an output interface for transmitting the spatialized audio to the at least one network client; and

a connection manager, in operative communication with the conference processor, for defining conferences associated with each client, the audio mix transmitted to a client corresponding to a conference associated with that client, wherein:

the input processor receives audio and position and orientation information from a plurality of network clients, the conference processor creating, for each client, a mix of audio from other clients spatialized according to the received position and orientation information;

the positional information is represented in a world model;

each client corresponds to a user;

for a first client, the associated conference comprises audio from other clients corresponding to users who are proximate to the first-client user in the world model; and

the conference processor comprises means for spatializing the audio by:

modeling a Doppler motion effect by delaying sound to the first client by a time proportional to a distance between the first-client user and a sound source as represented in the world model;

filtering to model object occlusion;

scaling by a gain inversely proportional to the distance between the listener and the sound source to model distance cues;

filtering by a pair of HRTFs to encode localization cues; and

summing the resulting stereo signal with stereo reverberation.

2. The conference server of claim 1 wherein the conference processor comprises a single spatializer per network client.

3. The conference server of claim 1 wherein the conference processor comprises a dedicated spatializer for each participant in each conference handled by the conference server.

4. The conference server of claim 1 comprising a plurality of blade servers, each blade server being dedicated to one or more conferences and comprising (i) a local world model containing position and orientation information received from participants in the one or more conferences; and (ii) a conference processor for spatializing the audio according to the local world model.

5. A method of spatializing sound, the method comprising the steps of:

receiving, on a server, audio and position and orientation information over a network from at least one network client;

spatializing the audio according to the received position and orientation information;

transmitting the spatialized audio to the at least one network client;

receiving audio and position and orientation information from a plurality of network clients and creating, for each client, a mix of audio from other clients spatialized according to the received position and orientation information; and

defining conferences associated with each client, the audio mix transmitted to a client corresponding to a conference associated with that client, wherein

positional information is represented in a world model;

each client corresponds to a user;

the positional information is represented in a world model;

for a first client, the associated conference comprises audio from other clients corresponding to users who are proximate to the first-client user in the world model; and

audio is spatialized by:

modeling a Doppler motion effect by delaying sound to the first client by a time proportional to a distance between the first-client user and a sound source as represented in the world model;

filtering to model object occlusion;

scaling by a gain inversely proportional to the distance between the listener and the sound source to model distance cues;

filtering by a pair of HRTFs to encode localization cues; and

summing the resulting stereo signal with stereo reverberation.