Mercurial > hg > qm-dsp
view dsp/transforms/FFT.h @ 255:9edaa3ce62e8
* Make MFCC able to accept already-FFT'd input, and simplify API a bit
* Add log power value to MFCC, restore windowing, and avoid some heap allocs
* In HMM, bail out of iteration if loglik hits NaN
author | Chris Cannam <c.cannam@qmul.ac.uk> |
---|---|
date | Fri, 18 Jan 2008 13:24:12 +0000 |
parents | 49844bc8a895 |
children | 6cb2b3cd5356 |
line wrap: on
line source
/* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */ /* QM DSP Library Centre for Digital Music, Queen Mary, University of London. This file is based on Don Cross's public domain FFT implementation. */ #ifndef FFT_H #define FFT_H class FFT { public: static void process(unsigned int nSamples, bool bInverseTransform, const double *lpRealIn, const double *lpImagIn, double *lpRealOut, double *lpImagOut); FFT(); virtual ~FFT(); protected: static unsigned int reverseBits(unsigned int nIndex, unsigned int nBits); static unsigned int numberOfBitsNeeded( unsigned int nSamples ); static bool isPowerOfTwo( unsigned int nX ); }; #endif