source: src/mathutils.h @ f88a326

feature/autosinkfeature/cnnfeature/cnn_orgfeature/constantqfeature/crepefeature/crepe_orgfeature/pitchshiftfeature/pydocstringsfeature/timestretchfix/ffmpeg5pitchshiftsamplertimestretchyinfft+
Last change on this file since f88a326 was a0fd4e4, checked in by Paul Brossier <piem@altern.org>, 20 years ago

merge aubionotes and aubionotesmedian

add make_audio_plot to gnuplot.py
adds autocorrelation to mathutils
adds Makefile.in to junk files

  • Property mode set to 100644
File size: 6.2 KB
Line 
1/*
2         Copyright (C) 2003 Paul Brossier
3
4         This program is free software; you can redistribute it and/or modify
5         it under the terms of the GNU General Public License as published by
6         the Free Software Foundation; either version 2 of the License, or
7         (at your option) any later version.
8
9         This program is distributed in the hope that it will be useful,
10         but WITHOUT ANY WARRANTY; without even the implied warranty of
11         MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
12         GNU General Public License for more details.
13
14         You should have received a copy of the GNU General Public License
15         along with this program; if not, write to the Free Software
16         Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
17
18*/
19
20/** @file
21 *  various math functions
22 *
23 *  \todo multichannel (each function should return -or set- an array sized to
24 *  the number of channel in the input vector)
25 *
26 *  \todo appropriate switches depending on types.h content
27 */
28
29#ifndef MATHUTILS_H
30#define MATHUTILS_H
31
32#define PI                              (M_PI)
33#define TWO_PI          (PI*2.)
34
35/* aliases to math.h functions */
36#define EXP                             expf
37#define COS                             cosf
38#define SIN                             sinf
39#define ABS                             fabsf
40#define POW                             powf
41#define SQRT                    sqrtf
42#define LOG10                   log10f
43#define LOG                       logf
44#define FLOOR                   floorf
45#define TRUNC                   truncf
46
47/* aliases to complex.h functions */
48#if defined(WIN32)
49/* mingw32 does not know about c*f functions */
50#define EXPC                    cexp
51/** complex = CEXPC(complex) */
52#define CEXPC                   cexp
53/** sample = ARGC(complex) */
54#define ARGC                    carg
55/** sample = ABSC(complex) norm */
56#define ABSC                    cabs
57/** sample = REAL(complex) */
58#define REAL                    creal
59/** sample = IMAG(complex) */
60#define IMAG                    cimag
61#else
62/** sample = EXPC(complex) */
63#define EXPC                    cexpf
64/** complex = CEXPC(complex) */
65#define CEXPC                   cexp
66/** sample = ARGC(complex) */
67#define ARGC                    cargf
68/** sample = ABSC(complex) norm */
69#define ABSC                    cabsf
70/** sample = REAL(complex) */
71#define REAL                    crealf
72/** sample = IMAG(complex) */
73#define IMAG                    cimagf
74#endif
75
76/* handy shortcuts */
77#define DB2LIN(g) (POW(10.0f,(g)*0.05f))
78#define LIN2DB(v) (20.0f*LOG10(v))
79#define SQR(_a)         (_a*_a)
80
81#define ELEM_SWAP(a,b) { register smpl_t t=(a);(a)=(b);(b)=t; }
82
83/** Window types
84 *
85 * inspired from
86 *
87 *  - dafx : http://profs.sci.univr.it/%7Edafx/Final-Papers/ps/Bernardini.ps.gz
88 *  - freqtweak : http://freqtweak.sf.net/
89 *  - extace : http://extace.sf.net/
90 */
91
92#ifdef __cplusplus
93extern "C" {
94#endif
95
96typedef enum {
97        rectangle,         
98        hamming,
99        hanning,
100        hanningz,
101        blackman,
102        blackman_harris,
103        gaussian,
104        welch,
105        parzen
106} window_type_t;
107
108/** create window */
109void window(smpl_t *w, uint_t size, window_type_t wintype);
110
111/** principal argument
112 *
113 * mod(phase+PI,-TWO_PI)+PI
114 */
115smpl_t unwrap2pi (smpl_t phase);
116
117/** calculates the mean of a vector
118 *
119 * \bug mono
120 */
121smpl_t vec_mean(fvec_t *s);
122/** returns the max of a vector
123 *
124 * \bug mono
125 */
126smpl_t vec_max(fvec_t *s);
127/** returns the min of a vector
128 *
129 * \bug mono
130 */
131smpl_t vec_min(fvec_t *s);
132/** returns the index of the min of a vector
133 *
134 * \bug mono
135 */
136uint_t vec_min_elem(fvec_t *s);
137/** returns the index of the max of a vector
138 *
139 * \bug mono
140 */
141uint_t vec_max_elem(fvec_t *s);
142/** implement 'fftshift' like function
143 *
144 * a[0]...,a[n/2],a[n/2+1],...a[n]
145 *
146 * becomes
147 *
148 * a[n/2+1],...a[n],a[0]...,a[n/2]
149 */
150void vec_shift(fvec_t *s);
151/** returns sum */
152smpl_t vec_sum(fvec_t *s);
153/** returns energy
154 *
155 * \bug mono
156 */
157smpl_t vec_local_energy(fvec_t * f);
158/** returns High Frequency Energy Content
159 *
160 * \bug mono */
161smpl_t vec_local_hfc(fvec_t * f);
162/** return alpha norm.
163 *
164 * alpha=2 means normalise variance.
165 * alpha=1 means normalise abs value.
166 * as alpha goes large, tends to normalisation
167 * by max value.
168 *
169 * \bug should not use POW :(
170 */
171smpl_t vec_alpha_norm(fvec_t * DF, smpl_t alpha);
172/*  dc(min) removal */
173void vec_dc_removal(fvec_t * mag);
174/**  alpha normalisation */
175void vec_alpha_normalise(fvec_t * mag, uint_t alpha);
176
177void vec_add(fvec_t * mag, smpl_t threshold);
178
179void vec_adapt_thres(fvec_t * vec, fvec_t * tmp, 
180    uint_t win_post, uint_t win_pre);
181/**  adaptative thresholding
182 *
183 * y=fn_thresh(fn,x,post,pre)
184 * compute adaptive threshold at each time
185 *    fn : a function name or pointer, eg 'median'
186 *    x:   signal vector
187 *    post: window length, causal part
188 *    pre: window length, anti-causal part
189 * Returns:
190 *    y:   signal the same length as x
191 *
192 * Formerly median_thresh, used compute median over a
193 * window of post+pre+1 samples, but now works with any
194 * function that takes a vector or matrix and returns a
195 * 'representative' value for each column, eg
196 *    medians=fn_thresh(median,x,8,8) 
197 *    minima=fn_thresh(min,x,8,8) 
198 * see SPARMS for explanation of post and pre
199 */
200smpl_t vec_moving_thres(fvec_t * vec, fvec_t * tmp, 
201    uint_t win_post, uint_t win_pre, uint_t win_pos);
202
203/** returns the median of the vector
204 *
205 *  This Quickselect routine is based on the algorithm described in
206 *  "Numerical recipes in C", Second Edition,
207 *  Cambridge University Press, 1992, Section 8.5, ISBN 0-521-43108-5
208 *
209 *  This code by Nicolas Devillard - 1998. Public domain,
210 *  available at http://ndevilla.free.fr/median/median/
211 */
212smpl_t vec_median(fvec_t * input);
213
214/** finds exact peak position by quadratic interpolation*/
215smpl_t vec_quadint(fvec_t * x,uint_t pos);
216
217/** Quadratic interpolation using Lagrange polynomial.
218 *
219 * inspired from ``Comparison of interpolation algorithms in real-time sound
220 * processing'', Vladimir Arnost,
221 *
222 * estimate = s0 + (pf/2.)*((pf-3.)*s0-2.*(pf-2.)*s1+(pf-1.)*s2);
223 *    where
224 *    \param s0,s1,s2 are 3 known points on the curve,
225 *    \param pf is the floating point index [0;2]
226 */
227smpl_t quadfrac(smpl_t s0, smpl_t s1, smpl_t s2, smpl_t pf);
228
229/** returns 1 if X1 is a peak and positive */
230uint_t vec_peakpick(fvec_t * input, uint_t pos);
231
232smpl_t bintomidi(smpl_t bin, smpl_t samplerate, smpl_t fftsize);
233smpl_t bintofreq(smpl_t bin, smpl_t samplerate, smpl_t fftsize);
234smpl_t freqtomidi(smpl_t freq);
235
236uint_t aubio_silence_detection(fvec_t * ibuf, smpl_t threshold);
237smpl_t aubio_level_detection(fvec_t * ibuf, smpl_t threshold);
238/**
239 * calculate normalised autocorrelation function
240 */
241void aubio_autocorr(fvec_t * input, fvec_t * output);
242
243#ifdef __cplusplus
244}
245#endif
246
247#endif
248
Note: See TracBrowser for help on using the repository browser.