FFANA FFSYN

FFANA / FFSYN - filter I/O modules - simple Wiener filter implementation

Analyse a signal using a simple Wiener filter.

Usage:

FFANA WIN L TYPE FMT

Inputs:

WIN

The name of a shell wave item for signal input (mono).

L

The filter/frame length in samples.

TYPE

The analysis/resynthesis filter type. The following values are supported:

NOOVERLAP|0 - a frame with 2.n samples is used for analysis, where the first half is filled with a signal frame and the 2nd half is set to 0. This method minimizes the effects (distortions) of the cyclic convolution. The input frames are not overlapped. The length of A and B is 2.(n/2+1).

HALFOVERLAP|1 - the input signal is multiplied with a Hanning window and half overlapped. The length of A and B is n/2+1.

FMT

The format of the output spectrum The following values are supported:

COMPLEX|0 - A is the real part and B the imaginary part of the complex spectrum.

MAGNITUDE|1 - A is the magnitude spectrum and B the phase spectrum.

POWER|2 - A is the power spectrum and B the phase spectrum.

Outputs:

A

First part of the signal spectrum (dependant on FMT).

B

Second part of the signal spectrum (dependant on FMT).

SR

The sampling rate.

Synthesise a signal using a simple Wiener filter.

Usage:

FFSYN WOUT L TYPE FMT A B

Inputs:

WOUT

The name of a shell wave item for the signal output (compatible to WIN).

Outputs:

none

Function:

The two atoms FFANA and FFSYN implement the analysis and re-synthesis modules for signal processing methods using the fast-convolution technique. The FFANA atom partitions the signal into frames and performs the Fourier transform. The synthesis module FFSYN uses the inverse Fourier transform and an overlap-add method to reconstruct the signal stream.

The FFT length n is set via the input L. If L is not a power of 2, the FFT length is set to the next power of 2 greater than or equal to L. The spectrum of the current frame is stored in the outputs A and B.

The source and target wave items must be mono signals with equal sampling rates and lengths. At the end of the process, the analyser atom appends zero-frames to the signal, while the synthesis-atom empties the delay buffers.

In the following example of a filter SPU, the magnitude spectrum is processed by the modification module and the phase spectrum is left untouched. For the filter/modification section any kind of stationary or adaptive process can be used.

Functional diagram of FFANA and FFSYN SPAtoms in the framework of the STx analysis/synthesis model.