getfft.3

speech signal process tools

.\" Copyright (c) 1987-1990 Entropic Speech, Inc.
.\" Copyright (c) 1997 Entropic Research Laboratory, Inc. All rights reserved.
.\" @(#)getfft.3	1.10 18 Apr 1997 ESI/ERL
.ds ]W (c) 1997 Entropic Research Laboratory, Inc.
.TH GET_FFT 3\-ESPSsp 18 Apr 1997
.SH NAME
.nf
get_fft \- Compute the fast Fourier transform of a data sequence.
get_fftd \- Compute the fast Fourier transform of a (double) data sequence.
get_fft_inv \- Compute the inverse fast Fourier transform of a data sequence.
get_fftd_inv \- Compute the inverse fast Fourier transform of a (double)data sequence.
get_rfft \- Compute the fast Fourier transform of a real data sequence.
.SH SYNOPSIS
.ft B
void
.br
get_fft (data_real, data_imag, log_fft_size)
.br
float   *data_real, *data_imag;
.br
int     log_fft_size;
.sp
void
.br
get_fftd (data_real, data_imag, log_fft_size)
.br
double  *data_real, *data_imag;
.br
int     log_fft_size;
.sp
void
.br
get_fft_inv (data_real, data_imag, log_fft_size)
.br
float   *data_real, *data_imag;
.br
int     log_fft_size;
.sp
void
.br
get_fftd_inv (data_real, data_imag, log_fft_size)
.br
double   *data_real, *data_imag;
.br
int     log_fft_size;
.sp
void
.br
get_rfft (data_real, data_imag, log_fft_size)
.br
float   *data_real, *data_imag;
.br
int     log_fft_size;
.sp .5
.SH DESCRIPTION
The
.I get_fft
and
.I get_fft_inv
routines compute the discrete Fourier transform and the inverse discrete
Fourier transform, respectively, of a given complex data sequence using the
fast Fourier transform algorithm.
The input data is supplied through two arrays
.B
data_real
and
.B
data_imag.
The transform size is
.B
2**log_fft_size.
The FFT output is returned through the same arrays.
For transform size N, there are N/2 negative frequencies,
and N/2 positive frequencies.
The order in which the frequency components are returned is as follows:
f(0), f(1),..., f(N/2), f(\-(N/2) + 1), f(\-(N/2) + 2),..., f(\-1).
This amounts to N values, with no explicit value for f(\-(N/2)) since
f(N/2) = f(\-(N/2)).
.PP
The functions
.I get_fftd
and 
.I get_fftd_inv
are the same as 
.I get_fft
and 
.I get_fftd
except they pass data through arrays of type double.
.PP
The function
.I get_rfft
is an FFT routine similar to
.I get_fft
except that the input data is assumed to be real.  If the 
input array 
.I data_imag
is not filled with zero, 
.I get_rfft
will exit with an error message.  
.I get_rfft
function is two times faster than the
.I get_fft
routine.
.SH BUGS
None
.SH SEE ALSO
.nf
\fIget_arspect\fP(3\-ESPSsp).
.fi
.SH COMMENTS
The sine and cosine tables needed by the FFT algorithm are computed first time
these routines are called.  During the subsequent calls, these tables are 
recomputed only if the transform sizes are different from the previous calls.
This information can be exploited to implement the user's program efficiently
in certain situations.
.SH REFERENCES
[1] A. V. Oppenheim and R. W. Schafer,
.I
Digital Signal Processing,
Prentice-Hall, NJ. 1975.
.SH AUTHOR
Shankar Narayan