henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. |
| 3 | * |
| 4 | * Use of this source code is governed by a BSD-style license |
| 5 | * that can be found in the LICENSE file in the root of the source |
| 6 | * tree. An additional intellectual property rights grant can be found |
| 7 | * in the file PATENTS. All contributing project authors may |
| 8 | * be found in the AUTHORS file in the root of the source tree. |
| 9 | */ |
| 10 | |
henrik.lundin@webrtc.org | e5abc85 | 2014-06-09 08:10:28 | [diff] [blame] | 11 | #include "webrtc/modules/audio_coding/neteq/dsp_helper.h" |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 12 | |
| 13 | #include <assert.h> |
pbos@webrtc.org | 3f45c2e | 2013-08-05 16:22:53 | [diff] [blame] | 14 | #include <string.h> // Access to memset. |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 15 | |
| 16 | #include <algorithm> // Access to min, max. |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 17 | |
| 18 | #include "webrtc/common_audio/signal_processing/include/signal_processing_library.h" |
| 19 | |
| 20 | namespace webrtc { |
| 21 | |
| 22 | // Table of constants used in method DspHelper::ParabolicFit(). |
| 23 | const int16_t DspHelper::kParabolaCoefficients[17][3] = { |
| 24 | { 120, 32, 64 }, |
| 25 | { 140, 44, 75 }, |
| 26 | { 150, 50, 80 }, |
| 27 | { 160, 57, 85 }, |
| 28 | { 180, 72, 96 }, |
| 29 | { 200, 89, 107 }, |
| 30 | { 210, 98, 112 }, |
| 31 | { 220, 108, 117 }, |
| 32 | { 240, 128, 128 }, |
| 33 | { 260, 150, 139 }, |
| 34 | { 270, 162, 144 }, |
| 35 | { 280, 174, 149 }, |
| 36 | { 300, 200, 160 }, |
| 37 | { 320, 228, 171 }, |
| 38 | { 330, 242, 176 }, |
| 39 | { 340, 257, 181 }, |
| 40 | { 360, 288, 192 } }; |
| 41 | |
| 42 | // Filter coefficients used when downsampling from the indicated sample rates |
| 43 | // (8, 16, 32, 48 kHz) to 4 kHz. Coefficients are in Q12. The corresponding Q0 |
| 44 | // values are provided in the comments before each array. |
| 45 | |
| 46 | // Q0 values: {0.3, 0.4, 0.3}. |
| 47 | const int16_t DspHelper::kDownsample8kHzTbl[3] = { 1229, 1638, 1229 }; |
| 48 | |
| 49 | // Q0 values: {0.15, 0.2, 0.3, 0.2, 0.15}. |
| 50 | const int16_t DspHelper::kDownsample16kHzTbl[5] = { 614, 819, 1229, 819, 614 }; |
| 51 | |
| 52 | // Q0 values: {0.1425, 0.1251, 0.1525, 0.1628, 0.1525, 0.1251, 0.1425}. |
| 53 | const int16_t DspHelper::kDownsample32kHzTbl[7] = { |
| 54 | 584, 512, 625, 667, 625, 512, 584 }; |
| 55 | |
| 56 | // Q0 values: {0.2487, 0.0952, 0.1042, 0.1074, 0.1042, 0.0952, 0.2487}. |
| 57 | const int16_t DspHelper::kDownsample48kHzTbl[7] = { |
| 58 | 1019, 390, 427, 440, 427, 390, 1019 }; |
| 59 | |
| 60 | int DspHelper::RampSignal(const int16_t* input, |
| 61 | size_t length, |
| 62 | int factor, |
| 63 | int increment, |
| 64 | int16_t* output) { |
| 65 | int factor_q20 = (factor << 6) + 32; |
| 66 | // TODO(hlundin): Add 32 to factor_q20 when converting back to Q14? |
| 67 | for (size_t i = 0; i < length; ++i) { |
| 68 | output[i] = (factor * input[i] + 8192) >> 14; |
| 69 | factor_q20 += increment; |
| 70 | factor_q20 = std::max(factor_q20, 0); // Never go negative. |
| 71 | factor = std::min(factor_q20 >> 6, 16384); |
| 72 | } |
| 73 | return factor; |
| 74 | } |
| 75 | |
| 76 | int DspHelper::RampSignal(int16_t* signal, |
| 77 | size_t length, |
| 78 | int factor, |
| 79 | int increment) { |
| 80 | return RampSignal(signal, length, factor, increment, signal); |
| 81 | } |
| 82 | |
minyue-webrtc | 9fe8c58 | 2016-05-10 17:55:56 | [diff] [blame] | 83 | int DspHelper::RampSignal(AudioVector* signal, |
| 84 | size_t start_index, |
| 85 | size_t length, |
| 86 | int factor, |
| 87 | int increment) { |
| 88 | int factor_q20 = (factor << 6) + 32; |
| 89 | // TODO(hlundin): Add 32 to factor_q20 when converting back to Q14? |
| 90 | for (size_t i = start_index; i < start_index + length; ++i) { |
| 91 | (*signal)[i] = (factor * (*signal)[i] + 8192) >> 14; |
| 92 | factor_q20 += increment; |
| 93 | factor_q20 = std::max(factor_q20, 0); // Never go negative. |
| 94 | factor = std::min(factor_q20 >> 6, 16384); |
| 95 | } |
| 96 | return factor; |
| 97 | } |
| 98 | |
henrik.lundin@webrtc.org | 0e9c399 | 2013-09-30 20:38:44 | [diff] [blame] | 99 | int DspHelper::RampSignal(AudioMultiVector* signal, |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 100 | size_t start_index, |
| 101 | size_t length, |
| 102 | int factor, |
| 103 | int increment) { |
| 104 | assert(start_index + length <= signal->Size()); |
| 105 | if (start_index + length > signal->Size()) { |
| 106 | // Wrong parameters. Do nothing and return the scale factor unaltered. |
| 107 | return factor; |
| 108 | } |
| 109 | int end_factor = 0; |
| 110 | // Loop over the channels, starting at the same |factor| each time. |
| 111 | for (size_t channel = 0; channel < signal->Channels(); ++channel) { |
| 112 | end_factor = |
minyue-webrtc | 9fe8c58 | 2016-05-10 17:55:56 | [diff] [blame] | 113 | RampSignal(&(*signal)[channel], start_index, length, factor, increment); |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 114 | } |
| 115 | return end_factor; |
| 116 | } |
| 117 | |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 118 | void DspHelper::PeakDetection(int16_t* data, size_t data_length, |
| 119 | size_t num_peaks, int fs_mult, |
| 120 | size_t* peak_index, int16_t* peak_value) { |
| 121 | size_t min_index = 0; |
| 122 | size_t max_index = 0; |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 123 | |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 124 | for (size_t i = 0; i <= num_peaks - 1; i++) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 125 | if (num_peaks == 1) { |
| 126 | // Single peak. The parabola fit assumes that an extra point is |
| 127 | // available; worst case it gets a zero on the high end of the signal. |
| 128 | // TODO(hlundin): This can potentially get much worse. It breaks the |
| 129 | // API contract, that the length of |data| is |data_length|. |
| 130 | data_length++; |
| 131 | } |
| 132 | |
| 133 | peak_index[i] = WebRtcSpl_MaxIndexW16(data, data_length - 1); |
| 134 | |
| 135 | if (i != num_peaks - 1) { |
pkasting | 9edfcab | 2015-07-22 22:17:22 | [diff] [blame] | 136 | min_index = (peak_index[i] > 2) ? (peak_index[i] - 2) : 0; |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 137 | max_index = std::min(data_length - 1, peak_index[i] + 2); |
| 138 | } |
| 139 | |
| 140 | if ((peak_index[i] != 0) && (peak_index[i] != (data_length - 2))) { |
| 141 | ParabolicFit(&data[peak_index[i] - 1], fs_mult, &peak_index[i], |
| 142 | &peak_value[i]); |
| 143 | } else { |
| 144 | if (peak_index[i] == data_length - 2) { |
| 145 | if (data[peak_index[i]] > data[peak_index[i] + 1]) { |
| 146 | ParabolicFit(&data[peak_index[i] - 1], fs_mult, &peak_index[i], |
| 147 | &peak_value[i]); |
| 148 | } else if (data[peak_index[i]] <= data[peak_index[i] + 1]) { |
| 149 | // Linear approximation. |
| 150 | peak_value[i] = (data[peak_index[i]] + data[peak_index[i] + 1]) >> 1; |
| 151 | peak_index[i] = (peak_index[i] * 2 + 1) * fs_mult; |
| 152 | } |
| 153 | } else { |
| 154 | peak_value[i] = data[peak_index[i]]; |
| 155 | peak_index[i] = peak_index[i] * 2 * fs_mult; |
| 156 | } |
| 157 | } |
| 158 | |
| 159 | if (i != num_peaks - 1) { |
| 160 | memset(&data[min_index], 0, |
| 161 | sizeof(data[0]) * (max_index - min_index + 1)); |
| 162 | } |
| 163 | } |
| 164 | } |
| 165 | |
| 166 | void DspHelper::ParabolicFit(int16_t* signal_points, int fs_mult, |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 167 | size_t* peak_index, int16_t* peak_value) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 168 | uint16_t fit_index[13]; |
| 169 | if (fs_mult == 1) { |
| 170 | fit_index[0] = 0; |
| 171 | fit_index[1] = 8; |
| 172 | fit_index[2] = 16; |
| 173 | } else if (fs_mult == 2) { |
| 174 | fit_index[0] = 0; |
| 175 | fit_index[1] = 4; |
| 176 | fit_index[2] = 8; |
| 177 | fit_index[3] = 12; |
| 178 | fit_index[4] = 16; |
| 179 | } else if (fs_mult == 4) { |
| 180 | fit_index[0] = 0; |
| 181 | fit_index[1] = 2; |
| 182 | fit_index[2] = 4; |
| 183 | fit_index[3] = 6; |
| 184 | fit_index[4] = 8; |
| 185 | fit_index[5] = 10; |
| 186 | fit_index[6] = 12; |
| 187 | fit_index[7] = 14; |
| 188 | fit_index[8] = 16; |
| 189 | } else { |
| 190 | fit_index[0] = 0; |
| 191 | fit_index[1] = 1; |
| 192 | fit_index[2] = 3; |
| 193 | fit_index[3] = 4; |
| 194 | fit_index[4] = 5; |
| 195 | fit_index[5] = 7; |
| 196 | fit_index[6] = 8; |
| 197 | fit_index[7] = 9; |
| 198 | fit_index[8] = 11; |
| 199 | fit_index[9] = 12; |
| 200 | fit_index[10] = 13; |
| 201 | fit_index[11] = 15; |
| 202 | fit_index[12] = 16; |
| 203 | } |
| 204 | |
| 205 | // num = -3 * signal_points[0] + 4 * signal_points[1] - signal_points[2]; |
| 206 | // den = signal_points[0] - 2 * signal_points[1] + signal_points[2]; |
| 207 | int32_t num = (signal_points[0] * -3) + (signal_points[1] * 4) |
| 208 | - signal_points[2]; |
| 209 | int32_t den = signal_points[0] + (signal_points[1] * -2) + signal_points[2]; |
| 210 | int32_t temp = num * 120; |
| 211 | int flag = 1; |
| 212 | int16_t stp = kParabolaCoefficients[fit_index[fs_mult]][0] |
| 213 | - kParabolaCoefficients[fit_index[fs_mult - 1]][0]; |
| 214 | int16_t strt = (kParabolaCoefficients[fit_index[fs_mult]][0] |
| 215 | + kParabolaCoefficients[fit_index[fs_mult - 1]][0]) / 2; |
| 216 | int16_t lmt; |
| 217 | if (temp < -den * strt) { |
| 218 | lmt = strt - stp; |
| 219 | while (flag) { |
| 220 | if ((flag == fs_mult) || (temp > -den * lmt)) { |
| 221 | *peak_value = (den * kParabolaCoefficients[fit_index[fs_mult - flag]][1] |
| 222 | + num * kParabolaCoefficients[fit_index[fs_mult - flag]][2] |
| 223 | + signal_points[0] * 256) / 256; |
| 224 | *peak_index = *peak_index * 2 * fs_mult - flag; |
| 225 | flag = 0; |
| 226 | } else { |
| 227 | flag++; |
| 228 | lmt -= stp; |
| 229 | } |
| 230 | } |
| 231 | } else if (temp > -den * (strt + stp)) { |
| 232 | lmt = strt + 2 * stp; |
| 233 | while (flag) { |
| 234 | if ((flag == fs_mult) || (temp < -den * lmt)) { |
| 235 | int32_t temp_term_1 = |
| 236 | den * kParabolaCoefficients[fit_index[fs_mult+flag]][1]; |
| 237 | int32_t temp_term_2 = |
| 238 | num * kParabolaCoefficients[fit_index[fs_mult+flag]][2]; |
| 239 | int32_t temp_term_3 = signal_points[0] * 256; |
| 240 | *peak_value = (temp_term_1 + temp_term_2 + temp_term_3) / 256; |
| 241 | *peak_index = *peak_index * 2 * fs_mult + flag; |
| 242 | flag = 0; |
| 243 | } else { |
| 244 | flag++; |
| 245 | lmt += stp; |
| 246 | } |
| 247 | } |
| 248 | } else { |
| 249 | *peak_value = signal_points[1]; |
| 250 | *peak_index = *peak_index * 2 * fs_mult; |
| 251 | } |
| 252 | } |
| 253 | |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 254 | size_t DspHelper::MinDistortion(const int16_t* signal, size_t min_lag, |
| 255 | size_t max_lag, size_t length, |
| 256 | int32_t* distortion_value) { |
| 257 | size_t best_index = 0; |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 258 | int32_t min_distortion = WEBRTC_SPL_WORD32_MAX; |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 259 | for (size_t i = min_lag; i <= max_lag; i++) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 260 | int32_t sum_diff = 0; |
| 261 | const int16_t* data1 = signal; |
| 262 | const int16_t* data2 = signal - i; |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 263 | for (size_t j = 0; j < length; j++) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 264 | sum_diff += WEBRTC_SPL_ABS_W32(data1[j] - data2[j]); |
| 265 | } |
| 266 | // Compare with previous minimum. |
| 267 | if (sum_diff < min_distortion) { |
| 268 | min_distortion = sum_diff; |
| 269 | best_index = i; |
| 270 | } |
| 271 | } |
| 272 | *distortion_value = min_distortion; |
| 273 | return best_index; |
| 274 | } |
| 275 | |
| 276 | void DspHelper::CrossFade(const int16_t* input1, const int16_t* input2, |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 277 | size_t length, int16_t* mix_factor, |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 278 | int16_t factor_decrement, int16_t* output) { |
| 279 | int16_t factor = *mix_factor; |
| 280 | int16_t complement_factor = 16384 - factor; |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 281 | for (size_t i = 0; i < length; i++) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 282 | output[i] = |
| 283 | (factor * input1[i] + complement_factor * input2[i] + 8192) >> 14; |
| 284 | factor -= factor_decrement; |
| 285 | complement_factor += factor_decrement; |
| 286 | } |
| 287 | *mix_factor = factor; |
| 288 | } |
| 289 | |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 290 | void DspHelper::UnmuteSignal(const int16_t* input, size_t length, |
Peter Kasting | 93ec7bc | 2015-06-12 02:57:18 | [diff] [blame] | 291 | int16_t* factor, int increment, |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 292 | int16_t* output) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 293 | uint16_t factor_16b = *factor; |
| 294 | int32_t factor_32b = (static_cast<int32_t>(factor_16b) << 6) + 32; |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 295 | for (size_t i = 0; i < length; i++) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 296 | output[i] = (factor_16b * input[i] + 8192) >> 14; |
| 297 | factor_32b = std::max(factor_32b + increment, 0); |
| 298 | factor_16b = std::min(16384, factor_32b >> 6); |
| 299 | } |
| 300 | *factor = factor_16b; |
| 301 | } |
| 302 | |
Peter Kasting | 93ec7bc | 2015-06-12 02:57:18 | [diff] [blame] | 303 | void DspHelper::MuteSignal(int16_t* signal, int mute_slope, size_t length) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 304 | int32_t factor = (16384 << 6) + 32; |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 305 | for (size_t i = 0; i < length; i++) { |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 306 | signal[i] = ((factor >> 6) * signal[i] + 8192) >> 14; |
| 307 | factor -= mute_slope; |
| 308 | } |
| 309 | } |
| 310 | |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 311 | int DspHelper::DownsampleTo4kHz(const int16_t* input, size_t input_length, |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 312 | size_t output_length, int input_rate_hz, |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 313 | bool compensate_delay, int16_t* output) { |
| 314 | // Set filter parameters depending on input frequency. |
| 315 | // NOTE: The phase delay values are wrong compared to the true phase delay |
| 316 | // of the filters. However, the error is preserved (through the +1 term) for |
| 317 | // consistency. |
| 318 | const int16_t* filter_coefficients; // Filter coefficients. |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 319 | size_t filter_length; // Number of coefficients. |
| 320 | size_t filter_delay; // Phase delay in samples. |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 321 | int16_t factor; // Conversion rate (inFsHz / 8000). |
| 322 | switch (input_rate_hz) { |
| 323 | case 8000: { |
| 324 | filter_length = 3; |
| 325 | factor = 2; |
| 326 | filter_coefficients = kDownsample8kHzTbl; |
| 327 | filter_delay = 1 + 1; |
| 328 | break; |
| 329 | } |
| 330 | case 16000: { |
| 331 | filter_length = 5; |
| 332 | factor = 4; |
| 333 | filter_coefficients = kDownsample16kHzTbl; |
| 334 | filter_delay = 2 + 1; |
| 335 | break; |
| 336 | } |
| 337 | case 32000: { |
| 338 | filter_length = 7; |
| 339 | factor = 8; |
| 340 | filter_coefficients = kDownsample32kHzTbl; |
| 341 | filter_delay = 3 + 1; |
| 342 | break; |
| 343 | } |
| 344 | case 48000: { |
| 345 | filter_length = 7; |
| 346 | factor = 12; |
| 347 | filter_coefficients = kDownsample48kHzTbl; |
| 348 | filter_delay = 3 + 1; |
| 349 | break; |
| 350 | } |
| 351 | default: { |
| 352 | assert(false); |
| 353 | return -1; |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | if (!compensate_delay) { |
| 358 | // Disregard delay compensation. |
| 359 | filter_delay = 0; |
| 360 | } |
| 361 | |
| 362 | // Returns -1 if input signal is too short; 0 otherwise. |
turaj@webrtc.org | 045e45e | 2013-09-20 16:25:28 | [diff] [blame] | 363 | return WebRtcSpl_DownsampleFast( |
Peter Kasting | a0ad248 | 2015-08-24 21:52:23 | [diff] [blame] | 364 | &input[filter_length - 1], input_length - filter_length + 1, output, |
| 365 | output_length, filter_coefficients, filter_length, factor, filter_delay); |
henrik.lundin@webrtc.org | 9a40081 | 2013-01-29 12:09:21 | [diff] [blame] | 366 | } |
| 367 | |
| 368 | } // namespace webrtc |