Linux Audio projects at Kokkini Zita

Jconvolver is a Convolution Engine for JACK, based on FFT convolution and using non-uniform partition sizes: small ones at the start of the IR and building up to the most efficient size further on. It can perform zero-delay processing with moderate CPU load. Jconvolver uses the convolution engine designed for Aella, a convolution application for reverberation processing (to be announced later). This distributes the calculation over up to five threads, one for each partition size, running at priorities just below the the one of JACK's processing thread. This engine is a separate library that will be documented as soon as I can find the time.

Main features:

• Any matrix of convolutions between up to up 64 inputs and 64 outputs, as long as your CPU(s) can handle it.
• Allows trading off CPU load to processing delay, and remains efficient even when configured for zero delay.
• Sparse and diagonal matrices are handled as efficiently as dense ones. No CPU cycles or memory resources are wasted on empty cells in the matrix, nor on empty partitions if IRs are of different length.

Libzita-resampler is a C++ library for resampling audio signals. It is designed to be used within a real-time processing context, to be fast, and to provide high-quality sample rate conversion.

The library operates on signals represented in single-precision floating point format. For multichannel operation both the input and output signals are assumed to be stored as interleaved samples.

The API allows a trade-off between quality and CPU load. For the latter a range of approximately 1:6 is available. Even at the highest quality setting libzita-resampler will be faster than most similar libraries, e.g. libsamplerate.

The source distribution includes the resample application. Input format is any file readable by libsndfile, output is either WAV (WAVEX for more than 2 channels) or CAF. Apart from resampling you can change the sample format to 16-bit, 24-bit or float, and for 16-bit output, add dithering. Available dithering types are rectangular, triangular, and Lipschitz' optimised error feedback filter. Some examples of dithering can be seen here.

First presented at the 2007 Linux Audio Conference.

The paper and pressentation can be found here.

TetraProc converts the A-format signals from a tetrahedral Ambisonic microphone into B-format signals ready for recording. Main features:

• A-B conversion using a classic scalar matrix and minimum phase filters, or
• A-B conversion using a 4 by 4 convolution matrix using measured or computed impulse responses, or a combination of both.
• Individual microphone calibration facilities.
• 24 dB/oct higpass filters.
• Metering, monitoring and test facilities.
• Virtual stereo mic for stereo monitoring or recording.
• Unlimited number of stored configurations.
• Jack client with graphical user interface.

TetraCal is the microphone calibration application for TetraProc. Given a number of impulse response measurements of a tetrahedral microphome (captured using Aliki, see below) it computes the matching configuration file for TetraProc. Main features:

• Automatic determination of gain and directivity errors of the A-format microphones.
• Automatic computation of an A-B matrix compensating for those errors.
• Frequency and phase response display of all A and B format signals.
• Interactive adjustment of pre- and post matrix equalisers.

Future versions of TetraCal will also exploit the convolution matrix implemented in TetraProc.

TetraProc is available now, please contact me by e-mail. Release of TetraCal awaits some required changes in Aliki, and completion of the manual that explains the complete calibration procedure.

As one result of a NDA signed with Core Sound, I can offer a free calibration service for anyone using their TetraMic on a Linux system. Given the serial number, I will provide a TetraProc configuration based on the impulse response measurements performed by Core Sound on your microphone. Currently this configuration does not use the convolution capabilities of TetraProc, but the results are excellent even without that. The combination of a very well designed microphone with IR-based calibration provides an affordable Ambisonic recording system of really suberb quality.

• 1st, 2nd and 3rd order 2-D or 3-D decoding.
• Up to 36 speakers (could be extended).
• Optional dual frequency band decoding.
• Optional speaker delay and gain compensation.
• Optional Near-Field effect compensation.
• Built-in test and Mute/Solo for each speaker.
• Unlimited number of presets.
• Jack client with graphical user interface.

Avaiable filters:

• Flat - no filtering.
• 20 kHz. Lowpass with a noise bandwidth of 20 kHz.
• A and C weighting filters.
• ITU-R468.
• ITU-R468, Dolby variant.

Avaiable Detectors:

• RMS. Root-mean-square (i.e. 'power') meter. The time constant is 125 ms as per IEC standard, or 1 second in slow mode.
• Average. This measures the average of the absolute value. The one used in jnoisemeter has actually the exact ballistics of a VU meter. A 10 times slower version is also provided.
• ITU-R468. This is a 'pseudo-peak' detector designed specifically to measure noise and S/N ratios. For a peak meter it is quite slow, as it should be for noise measurements, but at the same time it is much more sensitive to short impulsive noise than its speed would suggest.