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@1091
|
106 float
|
Chris@1091
|
107 FFTModel::getMagnitudeAt(int x, int y) const
|
Chris@1091
|
108 {
|
Chris@1093
|
109 if (x < 0 || x >= getWidth() || y < 0 || y >= getHeight()) return 0.f;
|
Chris@1093
|
110 auto col = getFFTColumn(x);
|
Chris@1093
|
111 return abs(col[y]);
|
Chris@1091
|
112 }
|
Chris@1091
|
113
|
Chris@1091
|
114 float
|
Chris@1091
|
115 FFTModel::getMaximumMagnitudeAt(int x) const
|
Chris@1091
|
116 {
|
Chris@1091
|
117 Column col(getColumn(x));
|
Chris@1092
|
118 float max = 0.f;
|
Chris@1154
|
119 int n = int(col.size());
|
Chris@1154
|
120 for (int i = 0; i < n; ++i) {
|
Chris@1092
|
121 if (col[i] > max) max = col[i];
|
Chris@1092
|
122 }
|
Chris@1092
|
123 return max;
|
Chris@1091
|
124 }
|
Chris@1091
|
125
|
Chris@1091
|
126 float
|
Chris@1091
|
127 FFTModel::getPhaseAt(int x, int y) const
|
Chris@1091
|
128 {
|
Chris@1093
|
129 if (x < 0 || x >= getWidth() || y < 0 || y >= getHeight()) return 0.f;
|
Chris@1091
|
130 return arg(getFFTColumn(x)[y]);
|
Chris@1091
|
131 }
|
Chris@1091
|
132
|
Chris@1091
|
133 void
|
Chris@1091
|
134 FFTModel::getValuesAt(int x, int y, float &re, float &im) const
|
Chris@1091
|
135 {
|
Chris@1091
|
136 auto col = getFFTColumn(x);
|
Chris@1091
|
137 re = col[y].real();
|
Chris@1091
|
138 im = col[y].imag();
|
Chris@1091
|
139 }
|
Chris@1091
|
140
|
Chris@1091
|
141 bool
|
Chris@1091
|
142 FFTModel::getMagnitudesAt(int x, float *values, int minbin, int count) const
|
Chris@1091
|
143 {
|
Chris@1091
|
144 if (count == 0) count = getHeight();
|
Chris@1091
|
145 auto col = getFFTColumn(x);
|
Chris@1091
|
146 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
147 values[i] = abs(col[minbin + i]);
|
Chris@1091
|
148 }
|
Chris@1091
|
149 return true;
|
Chris@1091
|
150 }
|
Chris@1091
|
151
|
Chris@1091
|
152 bool
|
Chris@1091
|
153 FFTModel::getPhasesAt(int x, float *values, int minbin, int count) const
|
Chris@1091
|
154 {
|
Chris@1091
|
155 if (count == 0) count = getHeight();
|
Chris@1091
|
156 auto col = getFFTColumn(x);
|
Chris@1091
|
157 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
158 values[i] = arg(col[minbin + i]);
|
Chris@1091
|
159 }
|
Chris@1091
|
160 return true;
|
Chris@1091
|
161 }
|
Chris@1091
|
162
|
Chris@1091
|
163 bool
|
Chris@1091
|
164 FFTModel::getValuesAt(int x, float *reals, float *imags, int minbin, int count) const
|
Chris@1091
|
165 {
|
Chris@1091
|
166 if (count == 0) count = getHeight();
|
Chris@1091
|
167 auto col = getFFTColumn(x);
|
Chris@1091
|
168 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
169 reals[i] = col[minbin + i].real();
|
Chris@1091
|
170 }
|
Chris@1091
|
171 for (int i = 0; i < count; ++i) {
|
Chris@1091
|
172 imags[i] = col[minbin + i].imag();
|
Chris@1091
|
173 }
|
Chris@1091
|
174 return true;
|
Chris@1091
|
175 }
|
Chris@1091
|
176
|
Chris@1091
|
177 vector<float>
|
Chris@1091
|
178 FFTModel::getSourceSamples(int column) const
|
Chris@1091
|
179 {
|
Chris@1094
|
180 // m_fftSize may be greater than m_windowSize, but not the reverse
|
Chris@1094
|
181
|
Chris@1094
|
182 // cerr << "getSourceSamples(" << column << ")" << endl;
|
Chris@1094
|
183
|
Chris@1091
|
184 auto range = getSourceSampleRange(column);
|
Chris@1094
|
185 auto data = getSourceData(range);
|
Chris@1094
|
186
|
Chris@1091
|
187 int off = (m_fftSize - m_windowSize) / 2;
|
Chris@1094
|
188
|
Chris@1094
|
189 if (off == 0) {
|
Chris@1094
|
190 return data;
|
Chris@1094
|
191 } else {
|
Chris@1094
|
192 vector<float> pad(off, 0.f);
|
Chris@1094
|
193 vector<float> padded;
|
Chris@1094
|
194 padded.reserve(m_fftSize);
|
Chris@1094
|
195 padded.insert(padded.end(), pad.begin(), pad.end());
|
Chris@1094
|
196 padded.insert(padded.end(), data.begin(), data.end());
|
Chris@1094
|
197 padded.insert(padded.end(), pad.begin(), pad.end());
|
Chris@1094
|
198 return padded;
|
Chris@1094
|
199 }
|
Chris@1094
|
200 }
|
Chris@1094
|
201
|
Chris@1094
|
202 vector<float>
|
Chris@1094
|
203 FFTModel::getSourceData(pair<sv_frame_t, sv_frame_t> range) const
|
Chris@1094
|
204 {
|
Chris@1094
|
205 // cerr << "getSourceData(" << range.first << "," << range.second
|
Chris@1094
|
206 // << "): saved range is (" << m_savedData.range.first
|
Chris@1094
|
207 // << "," << m_savedData.range.second << ")" << endl;
|
Chris@1094
|
208
|
Chris@1100
|
209 if (m_savedData.range == range) {
|
Chris@1100
|
210 return m_savedData.data;
|
Chris@1100
|
211 }
|
Chris@1094
|
212
|
Chris@1094
|
213 if (range.first < m_savedData.range.second &&
|
Chris@1094
|
214 range.first >= m_savedData.range.first &&
|
Chris@1094
|
215 range.second > m_savedData.range.second) {
|
Chris@1094
|
216
|
Chris@1100
|
217 sv_frame_t discard = range.first - m_savedData.range.first;
|
Chris@1100
|
218
|
Chris@1100
|
219 vector<float> acc(m_savedData.data.begin() + discard,
|
Chris@1100
|
220 m_savedData.data.end());
|
Chris@1094
|
221
|
Chris@1095
|
222 vector<float> rest =
|
Chris@1095
|
223 getSourceDataUncached({ m_savedData.range.second, range.second });
|
Chris@1100
|
224
|
Chris@1100
|
225 acc.insert(acc.end(), rest.begin(), rest.end());
|
Chris@1094
|
226
|
Chris@1095
|
227 m_savedData = { range, acc };
|
Chris@1095
|
228 return acc;
|
Chris@1095
|
229
|
Chris@1095
|
230 } else {
|
Chris@1095
|
231
|
Chris@1095
|
232 auto data = getSourceDataUncached(range);
|
Chris@1095
|
233 m_savedData = { range, data };
|
Chris@1095
|
234 return data;
|
Chris@1094
|
235 }
|
Chris@1095
|
236 }
|
Chris@1094
|
237
|
Chris@1095
|
238 vector<float>
|
Chris@1095
|
239 FFTModel::getSourceDataUncached(pair<sv_frame_t, sv_frame_t> range) const
|
Chris@1095
|
240 {
|
Chris@1091
|
241 decltype(range.first) pfx = 0;
|
Chris@1091
|
242 if (range.first < 0) {
|
Chris@1091
|
243 pfx = -range.first;
|
Chris@1091
|
244 range = { 0, range.second };
|
Chris@1091
|
245 }
|
Chris@1096
|
246
|
Chris@1096
|
247 auto data = m_model->getData(m_channel,
|
Chris@1096
|
248 range.first,
|
Chris@1096
|
249 range.second - range.first);
|
Chris@1096
|
250
|
Chris@1096
|
251 // don't return a partial frame
|
Chris@1096
|
252 data.resize(range.second - range.first, 0.f);
|
Chris@1096
|
253
|
Chris@1096
|
254 if (pfx > 0) {
|
Chris@1096
|
255 vector<float> pad(pfx, 0.f);
|
Chris@1096
|
256 data.insert(data.begin(), pad.begin(), pad.end());
|
Chris@1096
|
257 }
|
Chris@1096
|
258
|
Chris@1091
|
259 if (m_channel == -1) {
|
Chris@1091
|
260 int channels = m_model->getChannelCount();
|
Chris@1091
|
261 if (channels > 1) {
|
Chris@1096
|
262 int n = int(data.size());
|
Chris@1096
|
263 float factor = 1.f / float(channels);
|
Chris@1100
|
264 // use mean instead of sum for fft model input
|
Chris@1096
|
265 for (int i = 0; i < n; ++i) {
|
Chris@1096
|
266 data[i] *= factor;
|
Chris@1091
|
267 }
|
Chris@1091
|
268 }
|
Chris@1091
|
269 }
|
Chris@1094
|
270
|
Chris@1094
|
271 return data;
|
Chris@1091
|
272 }
|
Chris@1091
|
273
|
Chris@1091
|
274 vector<complex<float>>
|
Chris@1093
|
275 FFTModel::getFFTColumn(int n) const
|
Chris@1091
|
276 {
|
Chris@1093
|
277 for (auto &incache : m_cached) {
|
Chris@1093
|
278 if (incache.n == n) {
|
Chris@1093
|
279 return incache.col;
|
Chris@1093
|
280 }
|
Chris@1093
|
281 }
|
Chris@1093
|
282
|
Chris@1093
|
283 auto samples = getSourceSamples(n);
|
Chris@1100
|
284 m_windower.cut(samples.data());
|
Chris@1093
|
285 auto col = m_fft.process(samples);
|
Chris@1093
|
286
|
Chris@1093
|
287 SavedColumn sc { n, col };
|
Chris@1093
|
288 if (m_cached.size() >= m_cacheSize) {
|
Chris@1093
|
289 m_cached.pop_front();
|
Chris@1093
|
290 }
|
Chris@1093
|
291 m_cached.push_back(sc);
|
Chris@1093
|
292
|
Chris@1154
|
293 return move(col);
|
Chris@1091
|
294 }
|
Chris@1091
|
295
|
Chris@275
|
296 bool
|
Chris@1045
|
297 FFTModel::estimateStableFrequency(int x, int y, double &frequency)
|
Chris@275
|
298 {
|
Chris@275
|
299 if (!isOK()) return false;
|
Chris@275
|
300
|
Chris@1090
|
301 frequency = double(y * getSampleRate()) / m_fftSize;
|
Chris@275
|
302
|
Chris@275
|
303 if (x+1 >= getWidth()) return false;
|
Chris@275
|
304
|
Chris@275
|
305 // At frequency f, a phase shift of 2pi (one cycle) happens in 1/f sec.
|
Chris@275
|
306 // At hopsize h and sample rate sr, one hop happens in h/sr sec.
|
Chris@275
|
307 // At window size w, for bin b, f is b*sr/w.
|
Chris@275
|
308 // thus 2pi phase shift happens in w/(b*sr) sec.
|
Chris@275
|
309 // We need to know what phase shift we expect from h/sr sec.
|
Chris@275
|
310 // -> 2pi * ((h/sr) / (w/(b*sr)))
|
Chris@275
|
311 // = 2pi * ((h * b * sr) / (w * sr))
|
Chris@275
|
312 // = 2pi * (h * b) / w.
|
Chris@275
|
313
|
Chris@1038
|
314 double oldPhase = getPhaseAt(x, y);
|
Chris@1038
|
315 double newPhase = getPhaseAt(x+1, y);
|
Chris@275
|
316
|
Chris@929
|
317 int incr = getResolution();
|
Chris@275
|
318
|
Chris@1090
|
319 double expectedPhase = oldPhase + (2.0 * M_PI * y * incr) / m_fftSize;
|
Chris@275
|
320
|
Chris@1038
|
321 double phaseError = princarg(newPhase - expectedPhase);
|
Chris@275
|
322
|
Chris@275
|
323 // The new frequency estimate based on the phase error resulting
|
Chris@275
|
324 // from assuming the "native" frequency of this bin
|
Chris@275
|
325
|
Chris@275
|
326 frequency =
|
Chris@1090
|
327 (getSampleRate() * (expectedPhase + phaseError - oldPhase)) /
|
Chris@1045
|
328 (2.0 * M_PI * incr);
|
Chris@275
|
329
|
Chris@275
|
330 return true;
|
Chris@275
|
331 }
|
Chris@275
|
332
|
Chris@275
|
333 FFTModel::PeakLocationSet
|
Chris@1191
|
334 FFTModel::getPeaks(PeakPickType type, int x, int ymin, int ymax) const
|
Chris@275
|
335 {
|
Chris@551
|
336 Profiler profiler("FFTModel::getPeaks");
|
Chris@551
|
337
|
Chris@275
|
338 FFTModel::PeakLocationSet peaks;
|
Chris@275
|
339 if (!isOK()) return peaks;
|
Chris@275
|
340
|
Chris@275
|
341 if (ymax == 0 || ymax > getHeight() - 1) {
|
Chris@275
|
342 ymax = getHeight() - 1;
|
Chris@275
|
343 }
|
Chris@275
|
344
|
Chris@275
|
345 if (type == AllPeaks) {
|
Chris@551
|
346 int minbin = ymin;
|
Chris@551
|
347 if (minbin > 0) minbin = minbin - 1;
|
Chris@551
|
348 int maxbin = ymax;
|
Chris@551
|
349 if (maxbin < getHeight() - 1) maxbin = maxbin + 1;
|
Chris@551
|
350 const int n = maxbin - minbin + 1;
|
Chris@608
|
351 #ifdef __GNUC__
|
Chris@551
|
352 float values[n];
|
Chris@608
|
353 #else
|
Chris@608
|
354 float *values = (float *)alloca(n * sizeof(float));
|
Chris@608
|
355 #endif
|
Chris@551
|
356 getMagnitudesAt(x, values, minbin, maxbin - minbin + 1);
|
Chris@929
|
357 for (int bin = ymin; bin <= ymax; ++bin) {
|
Chris@551
|
358 if (bin == minbin || bin == maxbin) continue;
|
Chris@551
|
359 if (values[bin - minbin] > values[bin - minbin - 1] &&
|
Chris@551
|
360 values[bin - minbin] > values[bin - minbin + 1]) {
|
Chris@275
|
361 peaks.insert(bin);
|
Chris@275
|
362 }
|
Chris@275
|
363 }
|
Chris@275
|
364 return peaks;
|
Chris@275
|
365 }
|
Chris@275
|
366
|
Chris@551
|
367 Column values = getColumn(x);
|
Chris@1154
|
368 int nv = int(values.size());
|
Chris@275
|
369
|
Chris@500
|
370 float mean = 0.f;
|
Chris@1154
|
371 for (int i = 0; i < nv; ++i) mean += values[i];
|
Chris@1154
|
372 if (nv > 0) mean = mean / float(values.size());
|
Chris@1038
|
373
|
Chris@275
|
374 // For peak picking we use a moving median window, picking the
|
Chris@275
|
375 // highest value within each continuous region of values that
|
Chris@275
|
376 // exceed the median. For pitch adaptivity, we adjust the window
|
Chris@275
|
377 // size to a roughly constant pitch range (about four tones).
|
Chris@275
|
378
|
Chris@1040
|
379 sv_samplerate_t sampleRate = getSampleRate();
|
Chris@275
|
380
|
Chris@1090
|
381 deque<float> window;
|
Chris@1090
|
382 vector<int> inrange;
|
Chris@280
|
383 float dist = 0.5;
|
Chris@500
|
384
|
Chris@929
|
385 int medianWinSize = getPeakPickWindowSize(type, sampleRate, ymin, dist);
|
Chris@929
|
386 int halfWin = medianWinSize/2;
|
Chris@275
|
387
|
Chris@929
|
388 int binmin;
|
Chris@275
|
389 if (ymin > halfWin) binmin = ymin - halfWin;
|
Chris@275
|
390 else binmin = 0;
|
Chris@275
|
391
|
Chris@929
|
392 int binmax;
|
Chris@1154
|
393 if (ymax + halfWin < nv) binmax = ymax + halfWin;
|
Chris@1154
|
394 else binmax = nv - 1;
|
Chris@275
|
395
|
Chris@929
|
396 int prevcentre = 0;
|
Chris@500
|
397
|
Chris@929
|
398 for (int bin = binmin; bin <= binmax; ++bin) {
|
Chris@275
|
399
|
Chris@275
|
400 float value = values[bin];
|
Chris@275
|
401
|
Chris@275
|
402 window.push_back(value);
|
Chris@275
|
403
|
Chris@280
|
404 // so-called median will actually be the dist*100'th percentile
|
Chris@280
|
405 medianWinSize = getPeakPickWindowSize(type, sampleRate, bin, dist);
|
Chris@275
|
406 halfWin = medianWinSize/2;
|
Chris@275
|
407
|
Chris@929
|
408 while ((int)window.size() > medianWinSize) {
|
Chris@500
|
409 window.pop_front();
|
Chris@500
|
410 }
|
Chris@500
|
411
|
Chris@1038
|
412 int actualSize = int(window.size());
|
Chris@275
|
413
|
Chris@275
|
414 if (type == MajorPitchAdaptivePeaks) {
|
Chris@1154
|
415 if (ymax + halfWin < nv) binmax = ymax + halfWin;
|
Chris@1154
|
416 else binmax = nv - 1;
|
Chris@275
|
417 }
|
Chris@275
|
418
|
Chris@1090
|
419 deque<float> sorted(window);
|
Chris@1090
|
420 sort(sorted.begin(), sorted.end());
|
Chris@1038
|
421 float median = sorted[int(float(sorted.size()) * dist)];
|
Chris@275
|
422
|
Chris@929
|
423 int centrebin = 0;
|
Chris@500
|
424 if (bin > actualSize/2) centrebin = bin - actualSize/2;
|
Chris@500
|
425
|
Chris@500
|
426 while (centrebin > prevcentre || bin == binmin) {
|
Chris@275
|
427
|
Chris@500
|
428 if (centrebin > prevcentre) ++prevcentre;
|
Chris@500
|
429
|
Chris@500
|
430 float centre = values[prevcentre];
|
Chris@500
|
431
|
Chris@500
|
432 if (centre > median) {
|
Chris@500
|
433 inrange.push_back(centrebin);
|
Chris@500
|
434 }
|
Chris@500
|
435
|
Chris@1154
|
436 if (centre <= median || centrebin+1 == nv) {
|
Chris@500
|
437 if (!inrange.empty()) {
|
Chris@929
|
438 int peakbin = 0;
|
Chris@500
|
439 float peakval = 0.f;
|
Chris@929
|
440 for (int i = 0; i < (int)inrange.size(); ++i) {
|
Chris@500
|
441 if (i == 0 || values[inrange[i]] > peakval) {
|
Chris@500
|
442 peakval = values[inrange[i]];
|
Chris@500
|
443 peakbin = inrange[i];
|
Chris@500
|
444 }
|
Chris@500
|
445 }
|
Chris@500
|
446 inrange.clear();
|
Chris@500
|
447 if (peakbin >= ymin && peakbin <= ymax) {
|
Chris@500
|
448 peaks.insert(peakbin);
|
Chris@275
|
449 }
|
Chris@275
|
450 }
|
Chris@275
|
451 }
|
Chris@500
|
452
|
Chris@500
|
453 if (bin == binmin) break;
|
Chris@275
|
454 }
|
Chris@275
|
455 }
|
Chris@275
|
456
|
Chris@275
|
457 return peaks;
|
Chris@275
|
458 }
|
Chris@275
|
459
|
Chris@929
|
460 int
|
Chris@1040
|
461 FFTModel::getPeakPickWindowSize(PeakPickType type, sv_samplerate_t sampleRate,
|
Chris@929
|
462 int bin, float &percentile) const
|
Chris@275
|
463 {
|
Chris@280
|
464 percentile = 0.5;
|
Chris@275
|
465 if (type == MajorPeaks) return 10;
|
Chris@275
|
466 if (bin == 0) return 3;
|
Chris@280
|
467
|
Chris@1091
|
468 double binfreq = (sampleRate * bin) / m_fftSize;
|
Chris@1038
|
469 double hifreq = Pitch::getFrequencyForPitch(73, 0, binfreq);
|
Chris@280
|
470
|
Chris@1091
|
471 int hibin = int(lrint((hifreq * m_fftSize) / sampleRate));
|
Chris@275
|
472 int medianWinSize = hibin - bin;
|
Chris@275
|
473 if (medianWinSize < 3) medianWinSize = 3;
|
Chris@280
|
474
|
Chris@1091
|
475 percentile = 0.5f + float(binfreq / sampleRate);
|
Chris@280
|
476
|
Chris@275
|
477 return medianWinSize;
|
Chris@275
|
478 }
|
Chris@275
|
479
|
Chris@275
|
480 FFTModel::PeakSet
|
Chris@929
|
481 FFTModel::getPeakFrequencies(PeakPickType type, int x,
|
Chris@1191
|
482 int ymin, int ymax) const
|
Chris@275
|
483 {
|
Chris@551
|
484 Profiler profiler("FFTModel::getPeakFrequencies");
|
Chris@551
|
485
|
Chris@275
|
486 PeakSet peaks;
|
Chris@275
|
487 if (!isOK()) return peaks;
|
Chris@275
|
488 PeakLocationSet locations = getPeaks(type, x, ymin, ymax);
|
Chris@275
|
489
|
Chris@1040
|
490 sv_samplerate_t sampleRate = getSampleRate();
|
Chris@929
|
491 int incr = getResolution();
|
Chris@275
|
492
|
Chris@275
|
493 // This duplicates some of the work of estimateStableFrequency to
|
Chris@275
|
494 // allow us to retrieve the phases in two separate vertical
|
Chris@275
|
495 // columns, instead of jumping back and forth between columns x and
|
Chris@275
|
496 // x+1, which may be significantly slower if re-seeking is needed
|
Chris@275
|
497
|
Chris@1090
|
498 vector<float> phases;
|
Chris@275
|
499 for (PeakLocationSet::iterator i = locations.begin();
|
Chris@275
|
500 i != locations.end(); ++i) {
|
Chris@275
|
501 phases.push_back(getPhaseAt(x, *i));
|
Chris@275
|
502 }
|
Chris@275
|
503
|
Chris@929
|
504 int phaseIndex = 0;
|
Chris@275
|
505 for (PeakLocationSet::iterator i = locations.begin();
|
Chris@275
|
506 i != locations.end(); ++i) {
|
Chris@1038
|
507 double oldPhase = phases[phaseIndex];
|
Chris@1038
|
508 double newPhase = getPhaseAt(x+1, *i);
|
Chris@1090
|
509 double expectedPhase = oldPhase + (2.0 * M_PI * *i * incr) / m_fftSize;
|
Chris@1038
|
510 double phaseError = princarg(newPhase - expectedPhase);
|
Chris@1038
|
511 double frequency =
|
Chris@275
|
512 (sampleRate * (expectedPhase + phaseError - oldPhase))
|
Chris@275
|
513 / (2 * M_PI * incr);
|
Chris@1045
|
514 peaks[*i] = frequency;
|
Chris@275
|
515 ++phaseIndex;
|
Chris@275
|
516 }
|
Chris@275
|
517
|
Chris@275
|
518 return peaks;
|
Chris@275
|
519 }
|
Chris@275
|
520
|