A set of Finite Impulse Response (FIR) filter kernels for use by convolution-based encoders and decoders.
The ATK's FIR filter kernels are designed to support a variety of useful encoding and decoding strategies. Encoding means, converting from some other signal format, e.g., monophonic or stereophonic, to Ambisonic B-format. Decoding means converting from Ambisonic B-format.
All kernels are constructed so that the zero order harmonic, W, is linear phase. This gives all encoder and decoder designs a constant delay in samples of:
delay = (kernel_size - 1) / 2
The soundfield's pressure component is delayed by delay
, and the phase
response of the higher order harmonics then referenced (offset) to preserve
the required relationships. The result is, all kernels of the same kernel_size
will produce a resulting signal delayed by the same number of samples.
SuperStereo - Classic Ambisonic Super Stereo encoding. Converts standard two channel stereo to pantophonic (Horizontal-Only) B-format by wrapping the input across a wide frontal stage.
Note: We strongly advise using this encoder as the default stereo encoder.
UHJ - Classic Ambisonic UHJ format encoding. Converts stereo compatible two channel Ambisonic UHJ format to pantophonic (Horizontal-Only) B-format. Shelf filtering optimized for dual-band psycho-acoustic decoding is included for convenience.
Spectral Spreading - The frequency spreading technique encodes a monophonic (single channel) input into B-format by smoothly rotating the signal across the soundfield sphere, by frequency. The spectrum is decomposed and then wrapped as planewaves.
Diffuse Field - Encodes a monophonic (single channel) input into B-format through frequency dependent phase diffusion. Equivalent to applying the spatial characteristics of a diffuse soundfield to the input sound. In the natural world, diffuse soundfields are found as the very late, immersive part of reverberation.
UHJ - Classic Ambisonic UHJ format decoding. Converts B-format to stereo compatible two channel Ambisonic UHJ format.
Note: We strongly advise using this decoder as the default stereo decoder.
Binaural - Binaural decoding. Converts B-format to two channel binaural output. Three full-band diffuse field equalized models are implemented:
Discussion: The Synthetic and Measured decoders are implemented as frequency dependent multi-band decoders, and can be regarded as spectrally optimized for hearing in different ranges.
The Synthetic decoder is 4-in-1:
Synthetic spherical head models a sphere only, and is equivalent to placing a pair of spaced, sphere baffled microphones in the soundfield. With no pinnae and no body, this spherical head model is symmetric across elevation. As elevation cues are not present, there is no Z component. That is, this decoder is pantophonic (Horizontal-Only).
The Measured decoders are 5-in-1:
LF = 35Hz & 60Hz for CIPIC & Listen.
Measured decoders are designed to compensate for the proximity (near-field effect) of the two HRTF measurement systems. Asymmetry, across all axes, found in the measured HRTFs is preserved.
Note: You should have received a license notice with these kernels. Please see the included Third Party Notices for further details on the CIPIC and Listen databases.
The kernels are already included as part of the ATK for Reaper install, and there is no need to download and install the kernels seperately.
Install ATK for SuperCollider. Launch SuperCollider3, and run the following code:
ATK Kernel Installation
(
// Create ATK support directory
// Place unzipped kernels in the directory opened
Atk.createUserSupportDir;
Atk.openUserSupportDir;
)
If ATK for SuperCollider has been correctly installed, the above code will open the ATK's user support directory. Place the downloaded kernels here.
If you use ATK for Reaper as well as ATK for SuperCollider on Mac OSX, the kernels are shared between the two programs, and they will reside in the same location. We do not expect this to cause any problems.
Known issues are logged at GitHub.
If you experience problems or have questions pertaining to the ATK Kernels, please create an issue in the ATK-Kernels issue tracker.
If you use the kernels for an artistic project, please let us know. We plan on adding a gallery of example artistic and creative projects that make use of the Ambisonic Toolkit.
If you wish to use the kernels as part of a technical or software project, we'd like to know about that too. We're planning to link to other projects making use of Ambisonic Toolkit assets.
Version 1.2.1
Changes:
Bug Fixes:
Version 1.2.0
New Features:
Bug Fixes:
Version 1.1.0
Version 1.0.1
Version 1.0.0
The filter kernels distributed with the Ambisonic Toolkit are licensed under a Creative Commons Attribution-Share Alike 3.0 Unported (CC BY-SA 3.0) License and are copyright the Ambisonic Toolkit Community and Joseph Anderson, 2011, 2016.