This example illustrates the use of the binauralModeled node to spatialize in real-time an incoming audio stream around the head of the listener. The process can be easily replicated for multiple sources or multichannel audio streams. In this implementation, the binaural process relies on the filtering of the incoming audio stream with HRTF (Head Related Transfer Function) measured on a human head in free field conditions (anechoïc room). In order to reduce the processing cost, each HRTF is modeled by the combination of a fractional delay that accounts for the Interaural Time Delay (ITD) and a series of biquad filters that approximate the magnitude spectrum of the HRTFs. See [1] for implementation details.
The basic user interface provides control on the rotation in the horizontal plane (azimuth) and on the vertical dimension (elevation). In the current version, the set of HRTFs is imposed. In a future release, the user interface will propose to download different sets of HRTFs filters from a public server that will host a database of individual HRTFs.
[1] T. Carpentier, Binaural Synthesis with the Web Audio API, 1st Web Audio Conference (WAC15), Paris 2015
Azimuth
Elevation