MzSpectrogramFFTW

SYNOPSIS

MzSpectrogramFFTW -- Demonstration of how to create spectral data using FFTW from time data supplied by a host application.

The MzSpectrogramFFTW plugin accepts time-sequence audio data as input from the host application and applies a Fourier transform to the data generate a spectrum. The plugin then calculates and outputs the non-negative frequency magnitude spectrum.

MzSpectrogramFFTW generates 1 output:

1.

The MzSpectrogramFFTW plugin produces the same output as the MzSpectrogramClient plugin. The difference between the two plugins is how the output is produced. MzSpectrogramClient calculates with a basic FFT algorithm, while MzSpectrogramFFTW calculates with a sophisticated FFT algorithm which is 3 to 4 times faster. So in the long run, you will save lots of time using the FFTW transforms rather than a basic FFT algorithm (although it will take a lot more time to get things compiled the first time using FFTW).

MzSpectogramFFT is a useful template for how to write your own frequency analysis plugin where more control over the transform process is necessary, or you want to implement a transform other than a Discrete Fourier Transform, such as a wavelet transform. Also, the analysis window applied to the audio signal in MzSpectrogramHost is controlled by the host (usually selected by the user, or set to a default type). If your plugin needs a specific analysis window which cannot be supplied by the host, then you should use a client-based transform as given in this example plugin.

Compiled versions of the MzSpectrogramFFTW plugin can be downloaded from the download page.

The source code for the plugin was last modified on 23 Jun 2006.

• Header:          [ html | PDF | text ]
• Source Code: [ html | PDF | text ]

In order of complexity:
1. MzSpectrogramHost -- demonstrates how to receive and process spectral data from the host application.
2. MzSpectrogramClient -- demonstrates how to generate and process spectral data generated from time data provided by the host application.
3. MzSpectrogramFFTW -- described on this page: demonstrates how to do three-to-four times faster Fourier transforms than MzSpectrgramClient.
4. MzNevermore -- a full-featured Spectrogram display function based on MzSpectrgramFFTW.