A C++ toolbox for computing Discrete and Fast Fourier Transforms (DFT,FFT), Power Spectral Density (PSD) estimates, and the sound pressure level (SPL) in (dB).
If you have a data file with column formats:
time signal_1 signal_2 signal_3
line1 0.0 0.0 1.0 -0.3
line2 0.1 2.5 2.0 0.7
.... ... ... ... ...
Linux
--------
Inside the cloned directory $ mkdir obj , $ mkdir bin, before compile.
1-compile the code by running $ make , inside the code directory after cloning.
2-run the code as follows:
$ ./bin/fftpsd -i datafile.dat -n 2048
for fft only.
or
$ ./bin/fftpsd -i datafile.dat -n 2048 -spl
for computing the sound pressure level
The above commands by default use the first two columns in the input file only.
If you want to compute fft/spl for other signals use the following:
$ ./bin/fftpsd -i datafile.dat -n 2048 -spl -c 3 -r 10 -q 100
This means use signal_2 and start from row 10 to row 100.
** An update will be pushed soon to create the obj and bin directories automatically.
For more options please use the following table:
help with implemented options
-------------------------------
minimum arguments for fft computations
--------------------------------------
[-i datafile]
[-n N]
minimum arguments for psd, power & spl computations
-----------------------------------------------------
[-i datafile]
[-n N]
[-psd] or [-pow] or [-spl]
If DFT is preferred just add to your arguments the flag [-dft]
Detailed option list and discription
---------------------------------------
[-i datafile] name of the data file with its address
[-o directory] output directory, default is the same as input
[-spec] for output files with headers and spec suffix
[-r first_row] of the signal data in the data file, default 1
[-q last_row ] of the signal data in the data file, default is the end of the file
[-c column] of the signal data in the data file, default 2, assuming time is at the first column
[-n N] number of data in one window subset, or in the whole sample if no shifting/averaging, must be 2^{k}, k is an integer
[-l Lt_sub] window length in sec or the whole sample length if no shifting/averaging
[-s shift] a value in [0.,1.] to indicate the ratio of data to be shifted
shift defaults: 0.5 for psd/pow/spl, 0. for fft
[-dt time_step] of the window subset or the whole sample in sec
[-w window] name of the window functions, for psd default is hann, for fft default is rectangular
windows: hann,hamm,bartlett,welch,blackman,triangular,rectangular
[-m number] either 1 or 0 for mean substract, default is 1 to substract the mean
[-psd filename] compute the power spectral density (psd), filename is optional
[-pow filename] compute the power spectrum, filename is optional
[-spl filename] compute the sound pressure level (SPL), filename is optional
[-peak] peak preserving averaging fft mode, useful for ensemble averaging with shift=1.
[-variance] variance preserving averaging fft mode for conserving the signal energy, by default it is forced for psd, power and spl computations
[-pref pressure] reference pressure for SPL computation, default is 2e-5 Pa
[-bin] bin-averaging for psd and spl
[-oct number] the number nth octave averaging that is required, default is 1