Chris@152
|
1 /* -*- c-basic-offset: 4 indent-tabs-mode: nil -*- vi:set ts=8 sts=4 sw=4: */
|
Chris@152
|
2
|
Chris@152
|
3 /*
|
Chris@152
|
4 Sonic Visualiser
|
Chris@152
|
5 An audio file viewer and annotation editor.
|
Chris@152
|
6 Centre for Digital Music, Queen Mary, University of London.
|
Chris@152
|
7 This file copyright 2006 Chris Cannam.
|
Chris@152
|
8
|
Chris@152
|
9 This program is free software; you can redistribute it and/or
|
Chris@152
|
10 modify it under the terms of the GNU General Public License as
|
Chris@152
|
11 published by the Free Software Foundation; either version 2 of the
|
Chris@152
|
12 License, or (at your option) any later version. See the file
|
Chris@152
|
13 COPYING included with this distribution for more information.
|
Chris@152
|
14 */
|
Chris@152
|
15
|
Chris@152
|
16 #include "FFTModel.h"
|
Chris@152
|
17 #include "DenseTimeValueModel.h"
|
Chris@152
|
18
|
Chris@183
|
19 #include "base/Profiler.h"
|
Chris@275
|
20 #include "base/Pitch.h"
|
Chris@183
|
21
|
Chris@402
|
22 #include <algorithm>
|
Chris@402
|
23
|
Chris@152
|
24 #include <cassert>
|
Chris@1090
|
25 #include <deque>
|
Chris@152
|
26
|
Chris@608
|
27 #ifndef __GNUC__
|
Chris@608
|
28 #include <alloca.h>
|
Chris@608
|
29 #endif
|
Chris@608
|
30
|
Chris@1090
|
31 using namespace std;
|
Chris@1090
|
32
|
Chris@152
|
33 FFTModel::FFTModel(const DenseTimeValueModel *model,
|
Chris@152
|
34 int channel,
|
Chris@152
|
35 WindowType windowType,
|
Chris@929
|
36 int windowSize,
|
Chris@929
|
37 int windowIncrement,
|
Chris@1090
|
38 int fftSize) :
|
Chris@1090
|
39 m_model(model),
|
Chris@1090
|
40 m_channel(channel),
|
Chris@1090
|
41 m_windowType(windowType),
|
Chris@1090
|
42 m_windowSize(windowSize),
|
Chris@1090
|
43 m_windowIncrement(windowIncrement),
|
Chris@1090
|
44 m_fftSize(fftSize),
|
Chris@1091
|
45 m_windower(windowType, windowSize),
|
Chris@1093
|
46 m_fft(fftSize),
|
Chris@1093
|
47 m_cacheSize(3)
|
Chris@152
|
48 {
|
Chris@1091
|
49 if (m_windowSize > m_fftSize) {
|
Chris@1091
|
50 cerr << "ERROR: FFTModel::FFTModel: window size (" << m_windowSize
|
Chris@1091
|
51 << ") must be at least FFT size (" << m_fftSize << ")" << endl;
|
Chris@1091
|
52 throw invalid_argument("FFTModel window size must be at least FFT size");
|
Chris@1091
|
53 }
|
Chris@1133
|
54
|
Chris@1133
|
55 connect(model, SIGNAL(modelChanged()), this, SIGNAL(modelChanged()));
|
Chris@1133
|
56 connect(model, SIGNAL(modelChangedWithin(sv_frame_t, sv_frame_t)),
|
Chris@1133
|
57 this, SIGNAL(modelChangedWithin(sv_frame_t, sv_frame_t)));
|
Chris@152
|
58 }
|
Chris@152
|
59
|
Chris@152
|
60 FFTModel::~FFTModel()
|
Chris@152
|
61 {
|
Chris@152
|
62 }
|
Chris@152
|
63
|
Chris@360
|
64 void
|
Chris@360
|
65 FFTModel::sourceModelAboutToBeDeleted()
|
Chris@360
|
66 {
|
Chris@1090
|
67 if (m_model) {
|
Chris@1090
|
68 cerr << "FFTModel[" << this << "]::sourceModelAboutToBeDeleted(" << m_model << ")" << endl;
|
Chris@1090
|
69 m_model = 0;
|
Chris@360
|
70 }
|
Chris@360
|
71 }
|
Chris@360
|
72
|
Chris@1091
|
73 int
|
Chris@1091
|
74 FFTModel::getWidth() const
|
Chris@1091
|
75 {
|
Chris@1091
|
76 if (!m_model) return 0;
|
Chris@1091
|
77 return int((m_model->getEndFrame() - m_model->getStartFrame())
|
Chris@1091
|
78 / m_windowIncrement) + 1;
|
Chris@1091
|
79 }
|
Chris@1091
|
80
|
Chris@1091
|
81 int
|
Chris@1091
|
82 FFTModel::getHeight() const
|
Chris@1091
|
83 {
|
Chris@1091
|
84 return m_fftSize / 2 + 1;
|
Chris@1091
|
85 }
|
Chris@1091
|
86
|
Chris@152
|
87 QString
|
Chris@929
|
88 FFTModel::getBinName(int n) const
|
Chris@152
|
89 {
|
Chris@1040
|
90 sv_samplerate_t sr = getSampleRate();
|
Chris@152
|
91 if (!sr) return "";
|
Chris@204
|
92 QString name = tr("%1 Hz").arg((n * sr) / ((getHeight()-1) * 2));
|
Chris@152
|
93 return name;
|
Chris@152
|
94 }
|
Chris@152
|
95
|
Chris@1091
|
96 FFTModel::Column
|
Chris@1091
|
97 FFTModel::getColumn(int x) const
|
Chris@1091
|
98 {
|
Chris@1091
|
99 auto cplx = getFFTColumn(x);
|
Chris@1091
|
100 Column col;
|
Chris@1154
|
101 col.reserve(cplx.size());
|
Chris@1091
|
102 for (auto c: cplx) col.push_back(abs(c));
|
Chris@1154
|
103 return move(col);
|
Chris@1091
|
104 }
|
Chris@1091
|
105
|
Chris@1200
|
106 FFTModel::Column
|
Chris@1200
|
107 FFTModel::getPhases(int x) const
|
Chris@1200
|
108 {
|
Chris@1200
|
109 auto cplx = getFFTColumn(x);
|
Chris@1200
|
110 Column col;
|
Chris@1200
|
111 col.reserve(cplx.size());
|
Chris@1201
|
112 for (auto c: cplx) {
|
Chris@1201
|
113 col.push_back(arg(c));
|
Chris@1201
|
114 }
|
Chris@1200
|
115 return move(col);
|
Chris@1200
|
116 }
|
Chris@1200
|
117
|
Chris@1091
|
118 float
|
Chris@1091
|
119 FFTModel::getMagnitudeAt(int x, int y) const
|
Chris@1091
|
120 {
|
Chris@1093
|
121 if (x < 0 || x >= getWidth() || y < 0 || y >= getHeight()) return 0.f;
|
Chris@1093
|
122 auto col = getFFTColumn(x);
|
Chris@1093
|
123 return abs(col[y]);
|
Chris@1091
|
124 }
|
Chris@1091
|
125
|
Chris@1091
|
126 float
|
Chris@1091
|
127 FFTModel::getMaximumMagnitudeAt(int x) const
|
Chris@1091
|
128 {
|
Chris@1091
|
129 Column col(getColumn(x));
|
Chris@1092
|
130 float max = 0.f;
|
Chris@1154
|
131 int n = int(col.size());
|
Chris@1154
|
132 for (int i = 0; i < n; ++i) {
|
Chris@1092
|
133 if (col[i] > max) max = col[i];
|
Chris@1092
|
134 }
|
Chris@1092
|
135 return max;
|
Chris@1091
|
136 }
|
Chris@1091
|
137
|
Chris@1091
|
138 float
|
Chris@1091
|
139 FFTModel::getPhaseAt(int x, int y) const
|
Chris@1091
|
140 {
|
Chris@1093
|
141 if (x < 0 || x >= getWidth() || y < 0 || y >= getHeight()) return 0.f;
|
Chris@1091
|
142 return arg(getFFTColumn(x)[y]);
|
Chris@1091
|
143 }
|
Chris@1091
|
144
|
Chris@1091
|
145 void
|
Chris@1091
|
146 FFTModel::getValuesAt(int x, int y, float &re, float &im) const
|
Chris@1091
|
147 {
|
Chris@1091
|
148 auto col = getFFTColumn(x);
|
Chris@1091
|
149 re = col[y].real();
|
Chris@1091
|
150 im = col[y].imag();
|
Chris@1091
|
151 }
|
Chris@1091
|
152
|
Chris@1091
|
153 bool
|
Chris@1091
|
154 FFTModel::getMagnitudesAt(int x, float *values, int minbin, int count) const
|
Chris@1091
|
155 {
|
Chris@1091
|
156 if (count == 0) count = getHeight();
|
Chris@1091
|
157 auto col = getFFTColumn(x);
|
Chris@1091
|
158 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
159 values[i] = abs(col[minbin + i]);
|
Chris@1091
|
160 }
|
Chris@1091
|
161 return true;
|
Chris@1091
|
162 }
|
Chris@1091
|
163
|
Chris@1091
|
164 bool
|
Chris@1091
|
165 FFTModel::getPhasesAt(int x, float *values, int minbin, int count) const
|
Chris@1091
|
166 {
|
Chris@1091
|
167 if (count == 0) count = getHeight();
|
Chris@1091
|
168 auto col = getFFTColumn(x);
|
Chris@1091
|
169 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
170 values[i] = arg(col[minbin + i]);
|
Chris@1091
|
171 }
|
Chris@1091
|
172 return true;
|
Chris@1091
|
173 }
|
Chris@1091
|
174
|
Chris@1091
|
175 bool
|
Chris@1091
|
176 FFTModel::getValuesAt(int x, float *reals, float *imags, int minbin, int count) const
|
Chris@1091
|
177 {
|
Chris@1091
|
178 if (count == 0) count = getHeight();
|
Chris@1091
|
179 auto col = getFFTColumn(x);
|
Chris@1091
|
180 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
181 reals[i] = col[minbin + i].real();
|
Chris@1091
|
182 }
|
Chris@1091
|
183 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
184 imags[i] = col[minbin + i].imag();
|
Chris@1091
|
185 }
|
Chris@1091
|
186 return true;
|
Chris@1091
|
187 }
|
Chris@1091
|
188
|
Chris@1091
|
189 vector<float>
|
Chris@1091
|
190 FFTModel::getSourceSamples(int column) const
|
Chris@1091
|
191 {
|
Chris@1094
|
192 // m_fftSize may be greater than m_windowSize, but not the reverse
|
Chris@1094
|
193
|
Chris@1094
|
194 // cerr << "getSourceSamples(" << column << ")" << endl;
|
Chris@1094
|
195
|
Chris@1091
|
196 auto range = getSourceSampleRange(column);
|
Chris@1094
|
197 auto data = getSourceData(range);
|
Chris@1094
|
198
|
Chris@1091
|
199 int off = (m_fftSize - m_windowSize) / 2;
|
Chris@1094
|
200
|
Chris@1094
|
201 if (off == 0) {
|
Chris@1094
|
202 return data;
|
Chris@1094
|
203 } else {
|
Chris@1094
|
204 vector<float> pad(off, 0.f);
|
Chris@1094
|
205 vector<float> padded;
|
Chris@1094
|
206 padded.reserve(m_fftSize);
|
Chris@1094
|
207 padded.insert(padded.end(), pad.begin(), pad.end());
|
Chris@1094
|
208 padded.insert(padded.end(), data.begin(), data.end());
|
Chris@1094
|
209 padded.insert(padded.end(), pad.begin(), pad.end());
|
Chris@1094
|
210 return padded;
|
Chris@1094
|
211 }
|
Chris@1094
|
212 }
|
Chris@1094
|
213
|
Chris@1094
|
214 vector<float>
|
Chris@1094
|
215 FFTModel::getSourceData(pair<sv_frame_t, sv_frame_t> range) const
|
Chris@1094
|
216 {
|
Chris@1094
|
217 // cerr << "getSourceData(" << range.first << "," << range.second
|
Chris@1094
|
218 // << "): saved range is (" << m_savedData.range.first
|
Chris@1094
|
219 // << "," << m_savedData.range.second << ")" << endl;
|
Chris@1094
|
220
|
Chris@1100
|
221 if (m_savedData.range == range) {
|
Chris@1100
|
222 return m_savedData.data;
|
Chris@1100
|
223 }
|
Chris@1094
|
224
|
Chris@1094
|
225 if (range.first < m_savedData.range.second &&
|
Chris@1094
|
226 range.first >= m_savedData.range.first &&
|
Chris@1094
|
227 range.second > m_savedData.range.second) {
|
Chris@1094
|
228
|
Chris@1100
|
229 sv_frame_t discard = range.first - m_savedData.range.first;
|
Chris@1100
|
230
|
Chris@1100
|
231 vector<float> acc(m_savedData.data.begin() + discard,
|
Chris@1100
|
232 m_savedData.data.end());
|
Chris@1094
|
233
|
Chris@1095
|
234 vector<float> rest =
|
Chris@1095
|
235 getSourceDataUncached({ m_savedData.range.second, range.second });
|
Chris@1100
|
236
|
Chris@1100
|
237 acc.insert(acc.end(), rest.begin(), rest.end());
|
Chris@1094
|
238
|
Chris@1095
|
239 m_savedData = { range, acc };
|
Chris@1095
|
240 return acc;
|
Chris@1095
|
241
|
Chris@1095
|
242 } else {
|
Chris@1095
|
243
|
Chris@1095
|
244 auto data = getSourceDataUncached(range);
|
Chris@1095
|
245 m_savedData = { range, data };
|
Chris@1095
|
246 return data;
|
Chris@1094
|
247 }
|
Chris@1095
|
248 }
|
Chris@1094
|
249
|
Chris@1095
|
250 vector<float>
|
Chris@1095
|
251 FFTModel::getSourceDataUncached(pair<sv_frame_t, sv_frame_t> range) const
|
Chris@1095
|
252 {
|
Chris@1091
|
253 decltype(range.first) pfx = 0;
|
Chris@1091
|
254 if (range.first < 0) {
|
Chris@1091
|
255 pfx = -range.first;
|
Chris@1091
|
256 range = { 0, range.second };
|
Chris@1091
|
257 }
|
Chris@1096
|
258
|
Chris@1096
|
259 auto data = m_model->getData(m_channel,
|
Chris@1096
|
260 range.first,
|
Chris@1096
|
261 range.second - range.first);
|
Chris@1096
|
262
|
Chris@1096
|
263 // don't return a partial frame
|
Chris@1096
|
264 data.resize(range.second - range.first, 0.f);
|
Chris@1096
|
265
|
Chris@1096
|
266 if (pfx > 0) {
|
Chris@1096
|
267 vector<float> pad(pfx, 0.f);
|
Chris@1096
|
268 data.insert(data.begin(), pad.begin(), pad.end());
|
Chris@1096
|
269 }
|
Chris@1096
|
270
|
Chris@1091
|
271 if (m_channel == -1) {
|
Chris@1091
|
272 int channels = m_model->getChannelCount();
|
Chris@1091
|
273 if (channels > 1) {
|
Chris@1096
|
274 int n = int(data.size());
|
Chris@1096
|
275 float factor = 1.f / float(channels);
|
Chris@1100
|
276 // use mean instead of sum for fft model input
|
Chris@1096
|
277 for (int i = 0; i < n; ++i) {
|
Chris@1096
|
278 data[i] *= factor;
|
Chris@1091
|
279 }
|
Chris@1091
|
280 }
|
Chris@1091
|
281 }
|
Chris@1094
|
282
|
Chris@1094
|
283 return data;
|
Chris@1091
|
284 }
|
Chris@1091
|
285
|
Chris@1091
|
286 vector<complex<float>>
|
Chris@1093
|
287 FFTModel::getFFTColumn(int n) const
|
Chris@1091
|
288 {
|
Chris@1093
|
289 for (auto &incache : m_cached) {
|
Chris@1093
|
290 if (incache.n == n) {
|
Chris@1093
|
291 return incache.col;
|
Chris@1093
|
292 }
|
Chris@1093
|
293 }
|
Chris@1093
|
294
|
Chris@1093
|
295 auto samples = getSourceSamples(n);
|
Chris@1100
|
296 m_windower.cut(samples.data());
|
Chris@1093
|
297 auto col = m_fft.process(samples);
|
Chris@1093
|
298
|
Chris@1093
|
299 SavedColumn sc { n, col };
|
Chris@1093
|
300 if (m_cached.size() >= m_cacheSize) {
|
Chris@1093
|
301 m_cached.pop_front();
|
Chris@1093
|
302 }
|
Chris@1093
|
303 m_cached.push_back(sc);
|
Chris@1093
|
304
|
Chris@1154
|
305 return move(col);
|
Chris@1091
|
306 }
|
Chris@1091
|
307
|
Chris@275
|
308 bool
|
Chris@1045
|
309 FFTModel::estimateStableFrequency(int x, int y, double &frequency)
|
Chris@275
|
310 {
|
Chris@275
|
311 if (!isOK()) return false;
|
Chris@275
|
312
|
Chris@1090
|
313 frequency = double(y * getSampleRate()) / m_fftSize;
|
Chris@275
|
314
|
Chris@275
|
315 if (x+1 >= getWidth()) return false;
|
Chris@275
|
316
|
Chris@275
|
317 // At frequency f, a phase shift of 2pi (one cycle) happens in 1/f sec.
|
Chris@275
|
318 // At hopsize h and sample rate sr, one hop happens in h/sr sec.
|
Chris@275
|
319 // At window size w, for bin b, f is b*sr/w.
|
Chris@275
|
320 // thus 2pi phase shift happens in w/(b*sr) sec.
|
Chris@275
|
321 // We need to know what phase shift we expect from h/sr sec.
|
Chris@275
|
322 // -> 2pi * ((h/sr) / (w/(b*sr)))
|
Chris@275
|
323 // = 2pi * ((h * b * sr) / (w * sr))
|
Chris@275
|
324 // = 2pi * (h * b) / w.
|
Chris@275
|
325
|
Chris@1038
|
326 double oldPhase = getPhaseAt(x, y);
|
Chris@1038
|
327 double newPhase = getPhaseAt(x+1, y);
|
Chris@275
|
328
|
Chris@929
|
329 int incr = getResolution();
|
Chris@275
|
330
|
Chris@1090
|
331 double expectedPhase = oldPhase + (2.0 * M_PI * y * incr) / m_fftSize;
|
Chris@275
|
332
|
Chris@1038
|
333 double phaseError = princarg(newPhase - expectedPhase);
|
Chris@275
|
334
|
Chris@275
|
335 // The new frequency estimate based on the phase error resulting
|
Chris@275
|
336 // from assuming the "native" frequency of this bin
|
Chris@275
|
337
|
Chris@275
|
338 frequency =
|
Chris@1090
|
339 (getSampleRate() * (expectedPhase + phaseError - oldPhase)) /
|
Chris@1045
|
340 (2.0 * M_PI * incr);
|
Chris@275
|
341
|
Chris@275
|
342 return true;
|
Chris@275
|
343 }
|
Chris@275
|
344
|
Chris@275
|
345 FFTModel::PeakLocationSet
|
Chris@1191
|
346 FFTModel::getPeaks(PeakPickType type, int x, int ymin, int ymax) const
|
Chris@275
|
347 {
|
Chris@551
|
348 Profiler profiler("FFTModel::getPeaks");
|
Chris@551
|
349
|
Chris@275
|
350 FFTModel::PeakLocationSet peaks;
|
Chris@275
|
351 if (!isOK()) return peaks;
|
Chris@275
|
352
|
Chris@275
|
353 if (ymax == 0 || ymax > getHeight() - 1) {
|
Chris@275
|
354 ymax = getHeight() - 1;
|
Chris@275
|
355 }
|
Chris@275
|
356
|
Chris@275
|
357 if (type == AllPeaks) {
|
Chris@551
|
358 int minbin = ymin;
|
Chris@551
|
359 if (minbin > 0) minbin = minbin - 1;
|
Chris@551
|
360 int maxbin = ymax;
|
Chris@551
|
361 if (maxbin < getHeight() - 1) maxbin = maxbin + 1;
|
Chris@551
|
362 const int n = maxbin - minbin + 1;
|
Chris@608
|
363 #ifdef __GNUC__
|
Chris@551
|
364 float values[n];
|
Chris@608
|
365 #else
|
Chris@608
|
366 float *values = (float *)alloca(n * sizeof(float));
|
Chris@608
|
367 #endif
|
Chris@551
|
368 getMagnitudesAt(x, values, minbin, maxbin - minbin + 1);
|
Chris@929
|
369 for (int bin = ymin; bin <= ymax; ++bin) {
|
Chris@551
|
370 if (bin == minbin || bin == maxbin) continue;
|
Chris@551
|
371 if (values[bin - minbin] > values[bin - minbin - 1] &&
|
Chris@551
|
372 values[bin - minbin] > values[bin - minbin + 1]) {
|
Chris@275
|
373 peaks.insert(bin);
|
Chris@275
|
374 }
|
Chris@275
|
375 }
|
Chris@275
|
376 return peaks;
|
Chris@275
|
377 }
|
Chris@275
|
378
|
Chris@551
|
379 Column values = getColumn(x);
|
Chris@1154
|
380 int nv = int(values.size());
|
Chris@275
|
381
|
Chris@500
|
382 float mean = 0.f;
|
Chris@1154
|
383 for (int i = 0; i < nv; ++i) mean += values[i];
|
Chris@1154
|
384 if (nv > 0) mean = mean / float(values.size());
|
Chris@1038
|
385
|
Chris@275
|
386 // For peak picking we use a moving median window, picking the
|
Chris@275
|
387 // highest value within each continuous region of values that
|
Chris@275
|
388 // exceed the median. For pitch adaptivity, we adjust the window
|
Chris@275
|
389 // size to a roughly constant pitch range (about four tones).
|
Chris@275
|
390
|
Chris@1040
|
391 sv_samplerate_t sampleRate = getSampleRate();
|
Chris@275
|
392
|
Chris@1090
|
393 deque<float> window;
|
Chris@1090
|
394 vector<int> inrange;
|
Chris@280
|
395 float dist = 0.5;
|
Chris@500
|
396
|
Chris@929
|
397 int medianWinSize = getPeakPickWindowSize(type, sampleRate, ymin, dist);
|
Chris@929
|
398 int halfWin = medianWinSize/2;
|
Chris@275
|
399
|
Chris@929
|
400 int binmin;
|
Chris@275
|
401 if (ymin > halfWin) binmin = ymin - halfWin;
|
Chris@275
|
402 else binmin = 0;
|
Chris@275
|
403
|
Chris@929
|
404 int binmax;
|
Chris@1154
|
405 if (ymax + halfWin < nv) binmax = ymax + halfWin;
|
Chris@1154
|
406 else binmax = nv - 1;
|
Chris@275
|
407
|
Chris@929
|
408 int prevcentre = 0;
|
Chris@500
|
409
|
Chris@929
|
410 for (int bin = binmin; bin <= binmax; ++bin) {
|
Chris@275
|
411
|
Chris@275
|
412 float value = values[bin];
|
Chris@275
|
413
|
Chris@275
|
414 window.push_back(value);
|
Chris@275
|
415
|
Chris@280
|
416 // so-called median will actually be the dist*100'th percentile
|
Chris@280
|
417 medianWinSize = getPeakPickWindowSize(type, sampleRate, bin, dist);
|
Chris@275
|
418 halfWin = medianWinSize/2;
|
Chris@275
|
419
|
Chris@929
|
420 while ((int)window.size() > medianWinSize) {
|
Chris@500
|
421 window.pop_front();
|
Chris@500
|
422 }
|
Chris@500
|
423
|
Chris@1038
|
424 int actualSize = int(window.size());
|
Chris@275
|
425
|
Chris@275
|
426 if (type == MajorPitchAdaptivePeaks) {
|
Chris@1154
|
427 if (ymax + halfWin < nv) binmax = ymax + halfWin;
|
Chris@1154
|
428 else binmax = nv - 1;
|
Chris@275
|
429 }
|
Chris@275
|
430
|
Chris@1090
|
431 deque<float> sorted(window);
|
Chris@1090
|
432 sort(sorted.begin(), sorted.end());
|
Chris@1038
|
433 float median = sorted[int(float(sorted.size()) * dist)];
|
Chris@275
|
434
|
Chris@929
|
435 int centrebin = 0;
|
Chris@500
|
436 if (bin > actualSize/2) centrebin = bin - actualSize/2;
|
Chris@500
|
437
|
Chris@500
|
438 while (centrebin > prevcentre || bin == binmin) {
|
Chris@275
|
439
|
Chris@500
|
440 if (centrebin > prevcentre) ++prevcentre;
|
Chris@500
|
441
|
Chris@500
|
442 float centre = values[prevcentre];
|
Chris@500
|
443
|
Chris@500
|
444 if (centre > median) {
|
Chris@500
|
445 inrange.push_back(centrebin);
|
Chris@500
|
446 }
|
Chris@500
|
447
|
Chris@1154
|
448 if (centre <= median || centrebin+1 == nv) {
|
Chris@500
|
449 if (!inrange.empty()) {
|
Chris@929
|
450 int peakbin = 0;
|
Chris@500
|
451 float peakval = 0.f;
|
Chris@929
|
452 for (int i = 0; i < (int)inrange.size(); ++i) {
|
Chris@500
|
453 if (i == 0 || values[inrange[i]] > peakval) {
|
Chris@500
|
454 peakval = values[inrange[i]];
|
Chris@500
|
455 peakbin = inrange[i];
|
Chris@500
|
456 }
|
Chris@500
|
457 }
|
Chris@500
|
458 inrange.clear();
|
Chris@500
|
459 if (peakbin >= ymin && peakbin <= ymax) {
|
Chris@500
|
460 peaks.insert(peakbin);
|
Chris@275
|
461 }
|
Chris@275
|
462 }
|
Chris@275
|
463 }
|
Chris@500
|
464
|
Chris@500
|
465 if (bin == binmin) break;
|
Chris@275
|
466 }
|
Chris@275
|
467 }
|
Chris@275
|
468
|
Chris@275
|
469 return peaks;
|
Chris@275
|
470 }
|
Chris@275
|
471
|
Chris@929
|
472 int
|
Chris@1040
|
473 FFTModel::getPeakPickWindowSize(PeakPickType type, sv_samplerate_t sampleRate,
|
Chris@929
|
474 int bin, float &percentile) const
|
Chris@275
|
475 {
|
Chris@280
|
476 percentile = 0.5;
|
Chris@275
|
477 if (type == MajorPeaks) return 10;
|
Chris@275
|
478 if (bin == 0) return 3;
|
Chris@280
|
479
|
Chris@1091
|
480 double binfreq = (sampleRate * bin) / m_fftSize;
|
Chris@1038
|
481 double hifreq = Pitch::getFrequencyForPitch(73, 0, binfreq);
|
Chris@280
|
482
|
Chris@1091
|
483 int hibin = int(lrint((hifreq * m_fftSize) / sampleRate));
|
Chris@275
|
484 int medianWinSize = hibin - bin;
|
Chris@275
|
485 if (medianWinSize < 3) medianWinSize = 3;
|
Chris@280
|
486
|
Chris@1091
|
487 percentile = 0.5f + float(binfreq / sampleRate);
|
Chris@280
|
488
|
Chris@275
|
489 return medianWinSize;
|
Chris@275
|
490 }
|
Chris@275
|
491
|
Chris@275
|
492 FFTModel::PeakSet
|
Chris@929
|
493 FFTModel::getPeakFrequencies(PeakPickType type, int x,
|
Chris@1191
|
494 int ymin, int ymax) const
|
Chris@275
|
495 {
|
Chris@551
|
496 Profiler profiler("FFTModel::getPeakFrequencies");
|
Chris@551
|
497
|
Chris@275
|
498 PeakSet peaks;
|
Chris@275
|
499 if (!isOK()) return peaks;
|
Chris@275
|
500 PeakLocationSet locations = getPeaks(type, x, ymin, ymax);
|
Chris@275
|
501
|
Chris@1040
|
502 sv_samplerate_t sampleRate = getSampleRate();
|
Chris@929
|
503 int incr = getResolution();
|
Chris@275
|
504
|
Chris@275
|
505 // This duplicates some of the work of estimateStableFrequency to
|
Chris@275
|
506 // allow us to retrieve the phases in two separate vertical
|
Chris@275
|
507 // columns, instead of jumping back and forth between columns x and
|
Chris@275
|
508 // x+1, which may be significantly slower if re-seeking is needed
|
Chris@275
|
509
|
Chris@1090
|
510 vector<float> phases;
|
Chris@275
|
511 for (PeakLocationSet::iterator i = locations.begin();
|
Chris@275
|
512 i != locations.end(); ++i) {
|
Chris@275
|
513 phases.push_back(getPhaseAt(x, *i));
|
Chris@275
|
514 }
|
Chris@275
|
515
|
Chris@929
|
516 int phaseIndex = 0;
|
Chris@275
|
517 for (PeakLocationSet::iterator i = locations.begin();
|
Chris@275
|
518 i != locations.end(); ++i) {
|
Chris@1038
|
519 double oldPhase = phases[phaseIndex];
|
Chris@1038
|
520 double newPhase = getPhaseAt(x+1, *i);
|
Chris@1090
|
521 double expectedPhase = oldPhase + (2.0 * M_PI * *i * incr) / m_fftSize;
|
Chris@1038
|
522 double phaseError = princarg(newPhase - expectedPhase);
|
Chris@1038
|
523 double frequency =
|
Chris@275
|
524 (sampleRate * (expectedPhase + phaseError - oldPhase))
|
Chris@275
|
525 / (2 * M_PI * incr);
|
Chris@1045
|
526 peaks[*i] = frequency;
|
Chris@275
|
527 ++phaseIndex;
|
Chris@275
|
528 }
|
Chris@275
|
529
|
Chris@275
|
530 return peaks;
|
Chris@275
|
531 }
|
Chris@275
|
532
|