/src/mozilla-central/dom/media/webaudio/blink/Biquad.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * Copyright (C) 2010 Google Inc. All rights reserved. |
3 | | * |
4 | | * Redistribution and use in source and binary forms, with or without |
5 | | * modification, are permitted provided that the following conditions |
6 | | * are met: |
7 | | * |
8 | | * 1. Redistributions of source code must retain the above copyright |
9 | | * notice, this list of conditions and the following disclaimer. |
10 | | * 2. Redistributions in binary form must reproduce the above copyright |
11 | | * notice, this list of conditions and the following disclaimer in the |
12 | | * documentation and/or other materials provided with the distribution. |
13 | | * 3. Neither the name of Apple Computer, Inc. ("Apple") nor the names of |
14 | | * its contributors may be used to endorse or promote products derived |
15 | | * from this software without specific prior written permission. |
16 | | * |
17 | | * THIS SOFTWARE IS PROVIDED BY APPLE AND ITS CONTRIBUTORS "AS IS" AND ANY |
18 | | * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED |
19 | | * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE |
20 | | * DISCLAIMED. IN NO EVENT SHALL APPLE OR ITS CONTRIBUTORS BE LIABLE FOR ANY |
21 | | * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
22 | | * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
23 | | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND |
24 | | * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
25 | | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
26 | | * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
27 | | */ |
28 | | |
29 | | #include "Biquad.h" |
30 | | |
31 | | #include "DenormalDisabler.h" |
32 | | |
33 | | #include <float.h> |
34 | | #include <algorithm> |
35 | | #include <math.h> |
36 | | |
37 | | namespace WebCore { |
38 | | |
39 | | Biquad::Biquad() |
40 | 0 | { |
41 | 0 | // Initialize as pass-thru (straight-wire, no filter effect) |
42 | 0 | setNormalizedCoefficients(1, 0, 0, 1, 0, 0); |
43 | 0 |
|
44 | 0 | reset(); // clear filter memory |
45 | 0 | } |
46 | | |
47 | | Biquad::~Biquad() |
48 | 0 | { |
49 | 0 | } |
50 | | |
51 | | void Biquad::process(const float* sourceP, float* destP, size_t framesToProcess) |
52 | 0 | { |
53 | 0 | // Create local copies of member variables |
54 | 0 | double x1 = m_x1; |
55 | 0 | double x2 = m_x2; |
56 | 0 | double y1 = m_y1; |
57 | 0 | double y2 = m_y2; |
58 | 0 |
|
59 | 0 | double b0 = m_b0; |
60 | 0 | double b1 = m_b1; |
61 | 0 | double b2 = m_b2; |
62 | 0 | double a1 = m_a1; |
63 | 0 | double a2 = m_a2; |
64 | 0 |
|
65 | 0 | for (size_t i = 0; i < framesToProcess; ++i) { |
66 | 0 | // FIXME: this can be optimized by pipelining the multiply adds... |
67 | 0 | double x = sourceP[i]; |
68 | 0 | double y = b0*x + b1*x1 + b2*x2 - a1*y1 - a2*y2; |
69 | 0 |
|
70 | 0 | destP[i] = y; |
71 | 0 |
|
72 | 0 | // Update state variables |
73 | 0 | x2 = x1; |
74 | 0 | x1 = x; |
75 | 0 | y2 = y1; |
76 | 0 | y1 = y; |
77 | 0 | } |
78 | 0 |
|
79 | 0 | // Avoid introducing a stream of subnormals when input is silent and the |
80 | 0 | // tail approaches zero. |
81 | 0 | if (x1 == 0.0 && x2 == 0.0 && (y1 != 0.0 || y2 != 0.0) && |
82 | 0 | fabs(y1) < FLT_MIN && fabs(y2) < FLT_MIN) { |
83 | 0 | // Flush future values to zero (until there is new input). |
84 | 0 | y1 = y2 = 0.0; |
85 | 0 | // Flush calculated values. |
86 | | #ifndef HAVE_DENORMAL |
87 | | for (int i = framesToProcess; i-- && fabsf(destP[i]) < FLT_MIN; ) { |
88 | | destP[i] = 0.0f; |
89 | | } |
90 | | #endif |
91 | | } |
92 | 0 | // Local variables back to member. |
93 | 0 | m_x1 = x1; |
94 | 0 | m_x2 = x2; |
95 | 0 | m_y1 = y1; |
96 | 0 | m_y2 = y2; |
97 | 0 | } |
98 | | |
99 | | void Biquad::reset() |
100 | 0 | { |
101 | 0 | m_x1 = m_x2 = m_y1 = m_y2 = 0; |
102 | 0 | } |
103 | | |
104 | | void Biquad::setLowpassParams(double cutoff, double resonance) |
105 | 0 | { |
106 | 0 | // Limit cutoff to 0 to 1. |
107 | 0 | cutoff = std::max(0.0, std::min(cutoff, 1.0)); |
108 | 0 |
|
109 | 0 | if (cutoff == 1) { |
110 | 0 | // When cutoff is 1, the z-transform is 1. |
111 | 0 | setNormalizedCoefficients(1, 0, 0, |
112 | 0 | 1, 0, 0); |
113 | 0 | } else if (cutoff > 0) { |
114 | 0 | // Compute biquad coefficients for lowpass filter |
115 | 0 | resonance = std::max(0.0, resonance); // can't go negative |
116 | 0 | double g = pow(10.0, -0.05 * resonance); |
117 | 0 | double w0 = M_PI * cutoff; |
118 | 0 | double cos_w0 = cos(w0); |
119 | 0 | double alpha = 0.5 * sin(w0) * g; |
120 | 0 |
|
121 | 0 | double b1 = 1.0 - cos_w0; |
122 | 0 | double b0 = 0.5 * b1; |
123 | 0 | double b2 = b0; |
124 | 0 | double a0 = 1.0 + alpha; |
125 | 0 | double a1 = -2.0 * cos_w0; |
126 | 0 | double a2 = 1.0 - alpha; |
127 | 0 |
|
128 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
129 | 0 | } else { |
130 | 0 | // When cutoff is zero, nothing gets through the filter, so set |
131 | 0 | // coefficients up correctly. |
132 | 0 | setNormalizedCoefficients(0, 0, 0, |
133 | 0 | 1, 0, 0); |
134 | 0 | } |
135 | 0 | } |
136 | | |
137 | | void Biquad::setHighpassParams(double cutoff, double resonance) |
138 | 0 | { |
139 | 0 | // Limit cutoff to 0 to 1. |
140 | 0 | cutoff = std::max(0.0, std::min(cutoff, 1.0)); |
141 | 0 |
|
142 | 0 | if (cutoff == 1) { |
143 | 0 | // The z-transform is 0. |
144 | 0 | setNormalizedCoefficients(0, 0, 0, |
145 | 0 | 1, 0, 0); |
146 | 0 | } else if (cutoff > 0) { |
147 | 0 | // Compute biquad coefficients for highpass filter |
148 | 0 | resonance = std::max(0.0, resonance); // can't go negative |
149 | 0 | double g = pow(10.0, -0.05 * resonance); |
150 | 0 | double w0 = M_PI * cutoff; |
151 | 0 | double cos_w0 = cos(w0); |
152 | 0 | double alpha = 0.5 * sin(w0) * g; |
153 | 0 |
|
154 | 0 | double b1 = -1.0 - cos_w0; |
155 | 0 | double b0 = -0.5 * b1; |
156 | 0 | double b2 = b0; |
157 | 0 | double a0 = 1.0 + alpha; |
158 | 0 | double a1 = -2.0 * cos_w0; |
159 | 0 | double a2 = 1.0 - alpha; |
160 | 0 |
|
161 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
162 | 0 | } else { |
163 | 0 | // When cutoff is zero, we need to be careful because the above |
164 | 0 | // gives a quadratic divided by the same quadratic, with poles |
165 | 0 | // and zeros on the unit circle in the same place. When cutoff |
166 | 0 | // is zero, the z-transform is 1. |
167 | 0 | setNormalizedCoefficients(1, 0, 0, |
168 | 0 | 1, 0, 0); |
169 | 0 | } |
170 | 0 | } |
171 | | |
172 | | void Biquad::setNormalizedCoefficients(double b0, double b1, double b2, double a0, double a1, double a2) |
173 | 0 | { |
174 | 0 | double a0Inverse = 1 / a0; |
175 | 0 |
|
176 | 0 | m_b0 = b0 * a0Inverse; |
177 | 0 | m_b1 = b1 * a0Inverse; |
178 | 0 | m_b2 = b2 * a0Inverse; |
179 | 0 | m_a1 = a1 * a0Inverse; |
180 | 0 | m_a2 = a2 * a0Inverse; |
181 | 0 | } |
182 | | |
183 | | void Biquad::setLowShelfParams(double frequency, double dbGain) |
184 | 0 | { |
185 | 0 | // Clip frequencies to between 0 and 1, inclusive. |
186 | 0 | frequency = std::max(0.0, std::min(frequency, 1.0)); |
187 | 0 |
|
188 | 0 | double A = pow(10.0, dbGain / 40); |
189 | 0 |
|
190 | 0 | if (frequency == 1) { |
191 | 0 | // The z-transform is a constant gain. |
192 | 0 | setNormalizedCoefficients(A * A, 0, 0, |
193 | 0 | 1, 0, 0); |
194 | 0 | } else if (frequency > 0) { |
195 | 0 | double w0 = M_PI * frequency; |
196 | 0 | double S = 1; // filter slope (1 is max value) |
197 | 0 | double alpha = 0.5 * sin(w0) * sqrt((A + 1 / A) * (1 / S - 1) + 2); |
198 | 0 | double k = cos(w0); |
199 | 0 | double k2 = 2 * sqrt(A) * alpha; |
200 | 0 | double aPlusOne = A + 1; |
201 | 0 | double aMinusOne = A - 1; |
202 | 0 |
|
203 | 0 | double b0 = A * (aPlusOne - aMinusOne * k + k2); |
204 | 0 | double b1 = 2 * A * (aMinusOne - aPlusOne * k); |
205 | 0 | double b2 = A * (aPlusOne - aMinusOne * k - k2); |
206 | 0 | double a0 = aPlusOne + aMinusOne * k + k2; |
207 | 0 | double a1 = -2 * (aMinusOne + aPlusOne * k); |
208 | 0 | double a2 = aPlusOne + aMinusOne * k - k2; |
209 | 0 |
|
210 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
211 | 0 | } else { |
212 | 0 | // When frequency is 0, the z-transform is 1. |
213 | 0 | setNormalizedCoefficients(1, 0, 0, |
214 | 0 | 1, 0, 0); |
215 | 0 | } |
216 | 0 | } |
217 | | |
218 | | void Biquad::setHighShelfParams(double frequency, double dbGain) |
219 | 0 | { |
220 | 0 | // Clip frequencies to between 0 and 1, inclusive. |
221 | 0 | frequency = std::max(0.0, std::min(frequency, 1.0)); |
222 | 0 |
|
223 | 0 | double A = pow(10.0, dbGain / 40); |
224 | 0 |
|
225 | 0 | if (frequency == 1) { |
226 | 0 | // The z-transform is 1. |
227 | 0 | setNormalizedCoefficients(1, 0, 0, |
228 | 0 | 1, 0, 0); |
229 | 0 | } else if (frequency > 0) { |
230 | 0 | double w0 = M_PI * frequency; |
231 | 0 | double S = 1; // filter slope (1 is max value) |
232 | 0 | double alpha = 0.5 * sin(w0) * sqrt((A + 1 / A) * (1 / S - 1) + 2); |
233 | 0 | double k = cos(w0); |
234 | 0 | double k2 = 2 * sqrt(A) * alpha; |
235 | 0 | double aPlusOne = A + 1; |
236 | 0 | double aMinusOne = A - 1; |
237 | 0 |
|
238 | 0 | double b0 = A * (aPlusOne + aMinusOne * k + k2); |
239 | 0 | double b1 = -2 * A * (aMinusOne + aPlusOne * k); |
240 | 0 | double b2 = A * (aPlusOne + aMinusOne * k - k2); |
241 | 0 | double a0 = aPlusOne - aMinusOne * k + k2; |
242 | 0 | double a1 = 2 * (aMinusOne - aPlusOne * k); |
243 | 0 | double a2 = aPlusOne - aMinusOne * k - k2; |
244 | 0 |
|
245 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
246 | 0 | } else { |
247 | 0 | // When frequency = 0, the filter is just a gain, A^2. |
248 | 0 | setNormalizedCoefficients(A * A, 0, 0, |
249 | 0 | 1, 0, 0); |
250 | 0 | } |
251 | 0 | } |
252 | | |
253 | | void Biquad::setPeakingParams(double frequency, double Q, double dbGain) |
254 | 0 | { |
255 | 0 | // Clip frequencies to between 0 and 1, inclusive. |
256 | 0 | frequency = std::max(0.0, std::min(frequency, 1.0)); |
257 | 0 |
|
258 | 0 | // Don't let Q go negative, which causes an unstable filter. |
259 | 0 | Q = std::max(0.0, Q); |
260 | 0 |
|
261 | 0 | double A = pow(10.0, dbGain / 40); |
262 | 0 |
|
263 | 0 | if (frequency > 0 && frequency < 1) { |
264 | 0 | if (Q > 0) { |
265 | 0 | double w0 = M_PI * frequency; |
266 | 0 | double alpha = sin(w0) / (2 * Q); |
267 | 0 | double k = cos(w0); |
268 | 0 |
|
269 | 0 | double b0 = 1 + alpha * A; |
270 | 0 | double b1 = -2 * k; |
271 | 0 | double b2 = 1 - alpha * A; |
272 | 0 | double a0 = 1 + alpha / A; |
273 | 0 | double a1 = -2 * k; |
274 | 0 | double a2 = 1 - alpha / A; |
275 | 0 |
|
276 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
277 | 0 | } else { |
278 | 0 | // When Q = 0, the above formulas have problems. If we look at |
279 | 0 | // the z-transform, we can see that the limit as Q->0 is A^2, so |
280 | 0 | // set the filter that way. |
281 | 0 | setNormalizedCoefficients(A * A, 0, 0, |
282 | 0 | 1, 0, 0); |
283 | 0 | } |
284 | 0 | } else { |
285 | 0 | // When frequency is 0 or 1, the z-transform is 1. |
286 | 0 | setNormalizedCoefficients(1, 0, 0, |
287 | 0 | 1, 0, 0); |
288 | 0 | } |
289 | 0 | } |
290 | | |
291 | | void Biquad::setAllpassParams(double frequency, double Q) |
292 | 0 | { |
293 | 0 | // Clip frequencies to between 0 and 1, inclusive. |
294 | 0 | frequency = std::max(0.0, std::min(frequency, 1.0)); |
295 | 0 |
|
296 | 0 | // Don't let Q go negative, which causes an unstable filter. |
297 | 0 | Q = std::max(0.0, Q); |
298 | 0 |
|
299 | 0 | if (frequency > 0 && frequency < 1) { |
300 | 0 | if (Q > 0) { |
301 | 0 | double w0 = M_PI * frequency; |
302 | 0 | double alpha = sin(w0) / (2 * Q); |
303 | 0 | double k = cos(w0); |
304 | 0 |
|
305 | 0 | double b0 = 1 - alpha; |
306 | 0 | double b1 = -2 * k; |
307 | 0 | double b2 = 1 + alpha; |
308 | 0 | double a0 = 1 + alpha; |
309 | 0 | double a1 = -2 * k; |
310 | 0 | double a2 = 1 - alpha; |
311 | 0 |
|
312 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
313 | 0 | } else { |
314 | 0 | // When Q = 0, the above formulas have problems. If we look at |
315 | 0 | // the z-transform, we can see that the limit as Q->0 is -1, so |
316 | 0 | // set the filter that way. |
317 | 0 | setNormalizedCoefficients(-1, 0, 0, |
318 | 0 | 1, 0, 0); |
319 | 0 | } |
320 | 0 | } else { |
321 | 0 | // When frequency is 0 or 1, the z-transform is 1. |
322 | 0 | setNormalizedCoefficients(1, 0, 0, |
323 | 0 | 1, 0, 0); |
324 | 0 | } |
325 | 0 | } |
326 | | |
327 | | void Biquad::setNotchParams(double frequency, double Q) |
328 | 0 | { |
329 | 0 | // Clip frequencies to between 0 and 1, inclusive. |
330 | 0 | frequency = std::max(0.0, std::min(frequency, 1.0)); |
331 | 0 |
|
332 | 0 | // Don't let Q go negative, which causes an unstable filter. |
333 | 0 | Q = std::max(0.0, Q); |
334 | 0 |
|
335 | 0 | if (frequency > 0 && frequency < 1) { |
336 | 0 | if (Q > 0) { |
337 | 0 | double w0 = M_PI * frequency; |
338 | 0 | double alpha = sin(w0) / (2 * Q); |
339 | 0 | double k = cos(w0); |
340 | 0 |
|
341 | 0 | double b0 = 1; |
342 | 0 | double b1 = -2 * k; |
343 | 0 | double b2 = 1; |
344 | 0 | double a0 = 1 + alpha; |
345 | 0 | double a1 = -2 * k; |
346 | 0 | double a2 = 1 - alpha; |
347 | 0 |
|
348 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
349 | 0 | } else { |
350 | 0 | // When Q = 0, the above formulas have problems. If we look at |
351 | 0 | // the z-transform, we can see that the limit as Q->0 is 0, so |
352 | 0 | // set the filter that way. |
353 | 0 | setNormalizedCoefficients(0, 0, 0, |
354 | 0 | 1, 0, 0); |
355 | 0 | } |
356 | 0 | } else { |
357 | 0 | // When frequency is 0 or 1, the z-transform is 1. |
358 | 0 | setNormalizedCoefficients(1, 0, 0, |
359 | 0 | 1, 0, 0); |
360 | 0 | } |
361 | 0 | } |
362 | | |
363 | | void Biquad::setBandpassParams(double frequency, double Q) |
364 | 0 | { |
365 | 0 | // No negative frequencies allowed. |
366 | 0 | frequency = std::max(0.0, frequency); |
367 | 0 |
|
368 | 0 | // Don't let Q go negative, which causes an unstable filter. |
369 | 0 | Q = std::max(0.0, Q); |
370 | 0 |
|
371 | 0 | if (frequency > 0 && frequency < 1) { |
372 | 0 | double w0 = M_PI * frequency; |
373 | 0 | if (Q > 0) { |
374 | 0 | double alpha = sin(w0) / (2 * Q); |
375 | 0 | double k = cos(w0); |
376 | 0 |
|
377 | 0 | double b0 = alpha; |
378 | 0 | double b1 = 0; |
379 | 0 | double b2 = -alpha; |
380 | 0 | double a0 = 1 + alpha; |
381 | 0 | double a1 = -2 * k; |
382 | 0 | double a2 = 1 - alpha; |
383 | 0 |
|
384 | 0 | setNormalizedCoefficients(b0, b1, b2, a0, a1, a2); |
385 | 0 | } else { |
386 | 0 | // When Q = 0, the above formulas have problems. If we look at |
387 | 0 | // the z-transform, we can see that the limit as Q->0 is 1, so |
388 | 0 | // set the filter that way. |
389 | 0 | setNormalizedCoefficients(1, 0, 0, |
390 | 0 | 1, 0, 0); |
391 | 0 | } |
392 | 0 | } else { |
393 | 0 | // When the cutoff is zero, the z-transform approaches 0, if Q |
394 | 0 | // > 0. When both Q and cutoff are zero, the z-transform is |
395 | 0 | // pretty much undefined. What should we do in this case? |
396 | 0 | // For now, just make the filter 0. When the cutoff is 1, the |
397 | 0 | // z-transform also approaches 0. |
398 | 0 | setNormalizedCoefficients(0, 0, 0, |
399 | 0 | 1, 0, 0); |
400 | 0 | } |
401 | 0 | } |
402 | | |
403 | | void Biquad::setZeroPolePairs(const Complex &zero, const Complex &pole) |
404 | 0 | { |
405 | 0 | double b0 = 1; |
406 | 0 | double b1 = -2 * zero.real(); |
407 | 0 |
|
408 | 0 | double zeroMag = abs(zero); |
409 | 0 | double b2 = zeroMag * zeroMag; |
410 | 0 |
|
411 | 0 | double a1 = -2 * pole.real(); |
412 | 0 |
|
413 | 0 | double poleMag = abs(pole); |
414 | 0 | double a2 = poleMag * poleMag; |
415 | 0 | setNormalizedCoefficients(b0, b1, b2, 1, a1, a2); |
416 | 0 | } |
417 | | |
418 | | void Biquad::setAllpassPole(const Complex &pole) |
419 | 0 | { |
420 | 0 | Complex zero = Complex(1, 0) / pole; |
421 | 0 | setZeroPolePairs(zero, pole); |
422 | 0 | } |
423 | | |
424 | | void Biquad::getFrequencyResponse(int nFrequencies, |
425 | | const float* frequency, |
426 | | float* magResponse, |
427 | | float* phaseResponse) |
428 | 0 | { |
429 | 0 | // Evaluate the Z-transform of the filter at given normalized |
430 | 0 | // frequency from 0 to 1. (1 corresponds to the Nyquist |
431 | 0 | // frequency.) |
432 | 0 | // |
433 | 0 | // The z-transform of the filter is |
434 | 0 | // |
435 | 0 | // H(z) = (b0 + b1*z^(-1) + b2*z^(-2))/(1 + a1*z^(-1) + a2*z^(-2)) |
436 | 0 | // |
437 | 0 | // Evaluate as |
438 | 0 | // |
439 | 0 | // b0 + (b1 + b2*z1)*z1 |
440 | 0 | // -------------------- |
441 | 0 | // 1 + (a1 + a2*z1)*z1 |
442 | 0 | // |
443 | 0 | // with z1 = 1/z and z = exp(j*pi*frequency). Hence z1 = exp(-j*pi*frequency) |
444 | 0 |
|
445 | 0 | // Make local copies of the coefficients as a micro-optimization. |
446 | 0 | double b0 = m_b0; |
447 | 0 | double b1 = m_b1; |
448 | 0 | double b2 = m_b2; |
449 | 0 | double a1 = m_a1; |
450 | 0 | double a2 = m_a2; |
451 | 0 |
|
452 | 0 | for (int k = 0; k < nFrequencies; ++k) { |
453 | 0 | double omega = -M_PI * frequency[k]; |
454 | 0 | Complex z = Complex(cos(omega), sin(omega)); |
455 | 0 | Complex numerator = b0 + (b1 + b2 * z) * z; |
456 | 0 | Complex denominator = Complex(1, 0) + (a1 + a2 * z) * z; |
457 | 0 | // Strangely enough, using complex division: |
458 | 0 | // e.g. Complex response = numerator / denominator; |
459 | 0 | // fails on our test machines, yielding infinities and NaNs, so we do |
460 | 0 | // things the long way here. |
461 | 0 | double n = norm(denominator); |
462 | 0 | double r = (real(numerator)*real(denominator) + imag(numerator)*imag(denominator)) / n; |
463 | 0 | double i = (imag(numerator)*real(denominator) - real(numerator)*imag(denominator)) / n; |
464 | 0 | std::complex<double> response = std::complex<double>(r, i); |
465 | 0 |
|
466 | 0 | magResponse[k] = static_cast<float>(abs(response)); |
467 | 0 | phaseResponse[k] = static_cast<float>(atan2(imag(response), real(response))); |
468 | 0 | } |
469 | 0 | } |
470 | | |
471 | | } // namespace WebCore |
472 | | |