SUpDEq - Spatio-temporal spectral equalization of spherically measured head-related transfer functions
Keywords: Binaural, 3D-Audio, Kopfhörer, Virtual Reality, Augmented Reality, Silent Disco, Quiet Clubbing, Headphone
The use of binaural technology is made possible by head-related transfer functions (HRTF). These HRTFs can be measured using an artificial head. However, an individual personal measurement provides a better “acoustic fingerprint” of the spatial sound field arriving at a person's ears, which results in a more plausible presentation and improved localization. The measurement of an HRTF set is usually performed by microphones placed in the ears and measurement of the sound incidence from a variety of directions. These “acoustic fingerprints” represented by the HRTF set are then imprinted on an audio signal via a convolution, thus creating a spatial perception from the corresponding direction during playback. A high number of measuring positions (up to 2700 measuring positions) is required for HRTFs in full resolution. The reason is that the sound field can change considerably even between relatively close directions. Mathematically speaking, the sound field has high spatial orders that make interpolation difficult in practical applications.
The innovation SUpDEq (Spatial Upsampling by directional Equalization) developed by the Institute for Telecommunications Technology of TH Köln - University of Applied Sciences - is based on the spatial-temporal-spectral equalization of the spherical data sets. This shifts the energy of the HRTF sets in the SH domain to much smaller orders. A much lower spatial order can be used for measurement, storage and interpolation. The number of measured data sets is therefore reduced from approx. 2700 to only 100 - 300, without the errors in interpolation increasing significantly or artifacts becoming audible when using the data sets. The effort for the measurement and storage of an HRTF record is thus significantly reduced. The audio signal presented through the headphones contains the original spectral and temporal characteristics again. The spatially higher orders are also essentially restored.
- Drastic reduction of the measuring effort and the measuring duration
- Drastic reduction of computational effort in binaural synthesis
- Outstanding playback quality
- User-customized individualization of head geometries possible
- Headphone-based 3D audio
- Head-mounted displays
- Lifelike audio reproduction of sound events in clubs, concert halls, opera houses and sports stadiums
- Silent Disco / Quiet Clubbing
- Virtual Reality
- Augmented Reality
A patent application has been filed for the invention. A fully functional demonstrator in the form of a Matlab toolbox is available. We would be pleased to inform you about the status of the proceedings. The functionality was completely algorithmically realized and proven. On behalf of TH Köln - University of Applied Sciences, PROvendis offers interested companies licenses for the invention and the patent application.
An invention of TH Köln.