1 /*
2 * Copyright (c) 2011 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
11 /* digital_agc.c
12 *
13 */
14
15 #include "modules/audio_processing/agc/legacy/digital_agc.h"
16
17 #include <string.h>
18 #ifdef WEBRTC_AGC_DEBUG_DUMP
19 #include <stdio.h>
20 #endif
21
22 #include "rtc_base/checks.h"
23 #include "modules/audio_processing/agc/legacy/gain_control.h"
24
25 // To generate the gaintable, copy&paste the following lines to a Matlab window:
26 // MaxGain = 6; MinGain = 0; CompRatio = 3; Knee = 1;
27 // zeros = 0:31; lvl = 2.^(1-zeros);
28 // A = -10*log10(lvl) * (CompRatio - 1) / CompRatio;
29 // B = MaxGain - MinGain;
30 // gains = round(2^16*10.^(0.05 * (MinGain + B * (
31 // log(exp(-Knee*A)+exp(-Knee*B)) - log(1+exp(-Knee*B)) ) /
32 // log(1/(1+exp(Knee*B))))));
33 // fprintf(1, '\t%i, %i, %i, %i,\n', gains);
34 // % Matlab code for plotting the gain and input/output level characteristic
35 // (copy/paste the following 3 lines):
36 // in = 10*log10(lvl); out = 20*log10(gains/65536);
37 // subplot(121); plot(in, out); axis([-30, 0, -5, 20]); grid on; xlabel('Input
38 // (dB)'); ylabel('Gain (dB)');
39 // subplot(122); plot(in, in+out); axis([-30, 0, -30, 5]); grid on;
40 // xlabel('Input (dB)'); ylabel('Output (dB)');
41 // zoom on;
42
43 // Generator table for y=log2(1+e^x) in Q8.
44 enum { kGenFuncTableSize = 128 };
45 static const uint16_t kGenFuncTable[kGenFuncTableSize] = {
46 256, 485, 786, 1126, 1484, 1849, 2217, 2586, 2955, 3324, 3693,
47 4063, 4432, 4801, 5171, 5540, 5909, 6279, 6648, 7017, 7387, 7756,
48 8125, 8495, 8864, 9233, 9603, 9972, 10341, 10711, 11080, 11449, 11819,
49 12188, 12557, 12927, 13296, 13665, 14035, 14404, 14773, 15143, 15512, 15881,
50 16251, 16620, 16989, 17359, 17728, 18097, 18466, 18836, 19205, 19574, 19944,
51 20313, 20682, 21052, 21421, 21790, 22160, 22529, 22898, 23268, 23637, 24006,
52 24376, 24745, 25114, 25484, 25853, 26222, 26592, 26961, 27330, 27700, 28069,
53 28438, 28808, 29177, 29546, 29916, 30285, 30654, 31024, 31393, 31762, 32132,
54 32501, 32870, 33240, 33609, 33978, 34348, 34717, 35086, 35456, 35825, 36194,
55 36564, 36933, 37302, 37672, 38041, 38410, 38780, 39149, 39518, 39888, 40257,
56 40626, 40996, 41365, 41734, 42104, 42473, 42842, 43212, 43581, 43950, 44320,
57 44689, 45058, 45428, 45797, 46166, 46536, 46905};
58
59 static const int16_t kAvgDecayTime = 250; // frames; < 3000
60
WebRtcAgc_CalculateGainTable(int32_t * gainTable,int16_t digCompGaindB,int16_t targetLevelDbfs,uint8_t limiterEnable,int16_t analogTarget)61 int32_t WebRtcAgc_CalculateGainTable(int32_t* gainTable, // Q16
62 int16_t digCompGaindB, // Q0
63 int16_t targetLevelDbfs, // Q0
64 uint8_t limiterEnable,
65 int16_t analogTarget) // Q0
66 {
67 // This function generates the compressor gain table used in the fixed digital
68 // part.
69 uint32_t tmpU32no1, tmpU32no2, absInLevel, logApprox;
70 int32_t inLevel, limiterLvl;
71 int32_t tmp32, tmp32no1, tmp32no2, numFIX, den, y32;
72 const uint16_t kLog10 = 54426; // log2(10) in Q14
73 const uint16_t kLog10_2 = 49321; // 10*log10(2) in Q14
74 const uint16_t kLogE_1 = 23637; // log2(e) in Q14
75 uint16_t constMaxGain;
76 uint16_t tmpU16, intPart, fracPart;
77 const int16_t kCompRatio = 3;
78 const int16_t kSoftLimiterLeft = 1;
79 int16_t limiterOffset = 0; // Limiter offset
80 int16_t limiterIdx, limiterLvlX;
81 int16_t constLinApprox, zeroGainLvl, maxGain, diffGain;
82 int16_t i, tmp16, tmp16no1;
83 int zeros, zerosScale;
84
85 // Constants
86 // kLogE_1 = 23637; // log2(e) in Q14
87 // kLog10 = 54426; // log2(10) in Q14
88 // kLog10_2 = 49321; // 10*log10(2) in Q14
89
90 // Calculate maximum digital gain and zero gain level
91 tmp32no1 = (digCompGaindB - analogTarget) * (kCompRatio - 1);
92 tmp16no1 = analogTarget - targetLevelDbfs;
93 tmp16no1 +=
94 WebRtcSpl_DivW32W16ResW16(tmp32no1 + (kCompRatio >> 1), kCompRatio);
95 maxGain = WEBRTC_SPL_MAX(tmp16no1, (analogTarget - targetLevelDbfs));
96 tmp32no1 = maxGain * kCompRatio;
97 zeroGainLvl = digCompGaindB;
98 zeroGainLvl -= WebRtcSpl_DivW32W16ResW16(tmp32no1 + ((kCompRatio - 1) >> 1),
99 kCompRatio - 1);
100 if ((digCompGaindB <= analogTarget) && (limiterEnable)) {
101 zeroGainLvl += (analogTarget - digCompGaindB + kSoftLimiterLeft);
102 limiterOffset = 0;
103 }
104
105 // Calculate the difference between maximum gain and gain at 0dB0v:
106 // diffGain = maxGain + (compRatio-1)*zeroGainLvl/compRatio
107 // = (compRatio-1)*digCompGaindB/compRatio
108 tmp32no1 = digCompGaindB * (kCompRatio - 1);
109 diffGain =
110 WebRtcSpl_DivW32W16ResW16(tmp32no1 + (kCompRatio >> 1), kCompRatio);
111 if (diffGain < 0 || diffGain >= kGenFuncTableSize) {
112 RTC_DCHECK(0);
113 return -1;
114 }
115
116 // Calculate the limiter level and index:
117 // limiterLvlX = analogTarget - limiterOffset
118 // limiterLvl = targetLevelDbfs + limiterOffset/compRatio
119 limiterLvlX = analogTarget - limiterOffset;
120 limiterIdx = 2 + WebRtcSpl_DivW32W16ResW16((int32_t)limiterLvlX * (1 << 13),
121 kLog10_2 / 2);
122 tmp16no1 =
123 WebRtcSpl_DivW32W16ResW16(limiterOffset + (kCompRatio >> 1), kCompRatio);
124 limiterLvl = targetLevelDbfs + tmp16no1;
125
126 // Calculate (through table lookup):
127 // constMaxGain = log2(1+2^(log2(e)*diffGain)); (in Q8)
128 constMaxGain = kGenFuncTable[diffGain]; // in Q8
129
130 // Calculate a parameter used to approximate the fractional part of 2^x with a
131 // piecewise linear function in Q14:
132 // constLinApprox = round(3/2*(4*(3-2*sqrt(2))/(log(2)^2)-0.5)*2^14);
133 constLinApprox = 22817; // in Q14
134
135 // Calculate a denominator used in the exponential part to convert from dB to
136 // linear scale:
137 // den = 20*constMaxGain (in Q8)
138 den = WEBRTC_SPL_MUL_16_U16(20, constMaxGain); // in Q8
139
140 for (i = 0; i < 32; i++) {
141 // Calculate scaled input level (compressor):
142 // inLevel =
143 // fix((-constLog10_2*(compRatio-1)*(1-i)+fix(compRatio/2))/compRatio)
144 tmp16 = (int16_t)((kCompRatio - 1) * (i - 1)); // Q0
145 tmp32 = WEBRTC_SPL_MUL_16_U16(tmp16, kLog10_2) + 1; // Q14
146 inLevel = WebRtcSpl_DivW32W16(tmp32, kCompRatio); // Q14
147
148 // Calculate diffGain-inLevel, to map using the genFuncTable
149 inLevel = (int32_t)diffGain * (1 << 14) - inLevel; // Q14
150
151 // Make calculations on abs(inLevel) and compensate for the sign afterwards.
152 absInLevel = (uint32_t)WEBRTC_SPL_ABS_W32(inLevel); // Q14
153
154 // LUT with interpolation
155 intPart = (uint16_t)(absInLevel >> 14);
156 fracPart =
157 (uint16_t)(absInLevel & 0x00003FFF); // extract the fractional part
158 tmpU16 = kGenFuncTable[intPart + 1] - kGenFuncTable[intPart]; // Q8
159 tmpU32no1 = tmpU16 * fracPart; // Q22
160 tmpU32no1 += (uint32_t)kGenFuncTable[intPart] << 14; // Q22
161 logApprox = tmpU32no1 >> 8; // Q14
162 // Compensate for negative exponent using the relation:
163 // log2(1 + 2^-x) = log2(1 + 2^x) - x
164 if (inLevel < 0) {
165 zeros = WebRtcSpl_NormU32(absInLevel);
166 zerosScale = 0;
167 if (zeros < 15) {
168 // Not enough space for multiplication
169 tmpU32no2 = absInLevel >> (15 - zeros); // Q(zeros-1)
170 tmpU32no2 = WEBRTC_SPL_UMUL_32_16(tmpU32no2, kLogE_1); // Q(zeros+13)
171 if (zeros < 9) {
172 zerosScale = 9 - zeros;
173 tmpU32no1 >>= zerosScale; // Q(zeros+13)
174 } else {
175 tmpU32no2 >>= zeros - 9; // Q22
176 }
177 } else {
178 tmpU32no2 = WEBRTC_SPL_UMUL_32_16(absInLevel, kLogE_1); // Q28
179 tmpU32no2 >>= 6; // Q22
180 }
181 logApprox = 0;
182 if (tmpU32no2 < tmpU32no1) {
183 logApprox = (tmpU32no1 - tmpU32no2) >> (8 - zerosScale); // Q14
184 }
185 }
186 numFIX = (maxGain * constMaxGain) * (1 << 6); // Q14
187 numFIX -= (int32_t)logApprox * diffGain; // Q14
188
189 // Calculate ratio
190 // Shift |numFIX| as much as possible.
191 // Ensure we avoid wrap-around in |den| as well.
192 if (numFIX > (den >> 8) || -numFIX > (den >> 8)) // |den| is Q8.
193 {
194 zeros = WebRtcSpl_NormW32(numFIX);
195 } else {
196 zeros = WebRtcSpl_NormW32(den) + 8;
197 }
198 numFIX *= 1 << zeros; // Q(14+zeros)
199
200 // Shift den so we end up in Qy1
201 tmp32no1 = WEBRTC_SPL_SHIFT_W32(den, zeros - 9); // Q(zeros - 1)
202 y32 = numFIX / tmp32no1; // in Q15
203 // This is to do rounding in Q14.
204 y32 = y32 >= 0 ? (y32 + 1) >> 1 : -((-y32 + 1) >> 1);
205
206 if (limiterEnable && (i < limiterIdx)) {
207 tmp32 = WEBRTC_SPL_MUL_16_U16(i - 1, kLog10_2); // Q14
208 tmp32 -= limiterLvl * (1 << 14); // Q14
209 y32 = WebRtcSpl_DivW32W16(tmp32 + 10, 20);
210 }
211 if (y32 > 39000) {
212 tmp32 = (y32 >> 1) * kLog10 + 4096; // in Q27
213 tmp32 >>= 13; // In Q14.
214 } else {
215 tmp32 = y32 * kLog10 + 8192; // in Q28
216 tmp32 >>= 14; // In Q14.
217 }
218 tmp32 += 16 << 14; // in Q14 (Make sure final output is in Q16)
219
220 // Calculate power
221 if (tmp32 > 0) {
222 intPart = (int16_t)(tmp32 >> 14);
223 fracPart = (uint16_t)(tmp32 & 0x00003FFF); // in Q14
224 if ((fracPart >> 13) != 0) {
225 tmp16 = (2 << 14) - constLinApprox;
226 tmp32no2 = (1 << 14) - fracPart;
227 tmp32no2 *= tmp16;
228 tmp32no2 >>= 13;
229 tmp32no2 = (1 << 14) - tmp32no2;
230 } else {
231 tmp16 = constLinApprox - (1 << 14);
232 tmp32no2 = (fracPart * tmp16) >> 13;
233 }
234 fracPart = (uint16_t)tmp32no2;
235 gainTable[i] =
236 (1 << intPart) + WEBRTC_SPL_SHIFT_W32(fracPart, intPart - 14);
237 } else {
238 gainTable[i] = 0;
239 }
240 }
241
242 return 0;
243 }
244
WebRtcAgc_InitDigital(DigitalAgc * stt,int16_t agcMode)245 int32_t WebRtcAgc_InitDigital(DigitalAgc* stt, int16_t agcMode) {
246 if (agcMode == kAgcModeFixedDigital) {
247 // start at minimum to find correct gain faster
248 stt->capacitorSlow = 0;
249 } else {
250 // start out with 0 dB gain
251 stt->capacitorSlow = 134217728; // (int32_t)(0.125f * 32768.0f * 32768.0f);
252 }
253 stt->capacitorFast = 0;
254 stt->gain = 65536;
255 stt->gatePrevious = 0;
256 stt->agcMode = agcMode;
257 #ifdef WEBRTC_AGC_DEBUG_DUMP
258 stt->frameCounter = 0;
259 #endif
260
261 // initialize VADs
262 WebRtcAgc_InitVad(&stt->vadNearend);
263 WebRtcAgc_InitVad(&stt->vadFarend);
264
265 return 0;
266 }
267
WebRtcAgc_AddFarendToDigital(DigitalAgc * stt,const int16_t * in_far,size_t nrSamples)268 int32_t WebRtcAgc_AddFarendToDigital(DigitalAgc* stt,
269 const int16_t* in_far,
270 size_t nrSamples) {
271 RTC_DCHECK(stt);
272 // VAD for far end
273 WebRtcAgc_ProcessVad(&stt->vadFarend, in_far, nrSamples);
274
275 return 0;
276 }
277
WebRtcAgc_ProcessDigital(DigitalAgc * stt,const int16_t * const * in_near,size_t num_bands,int16_t * const * out,uint32_t FS,int16_t lowlevelSignal)278 int32_t WebRtcAgc_ProcessDigital(DigitalAgc* stt,
279 const int16_t* const* in_near,
280 size_t num_bands,
281 int16_t* const* out,
282 uint32_t FS,
283 int16_t lowlevelSignal) {
284 // array for gains (one value per ms, incl start & end)
285 int32_t gains[11];
286
287 int32_t out_tmp, tmp32;
288 int32_t env[10];
289 int32_t max_nrg;
290 int32_t cur_level;
291 int32_t gain32, delta;
292 int16_t logratio;
293 int16_t lower_thr, upper_thr;
294 int16_t zeros = 0, zeros_fast, frac = 0;
295 int16_t decay;
296 int16_t gate, gain_adj;
297 int16_t k;
298 size_t n, i, L;
299 int16_t L2; // samples/subframe
300
301 // determine number of samples per ms
302 if (FS == 8000) {
303 L = 8;
304 L2 = 3;
305 } else if (FS == 16000 || FS == 32000 || FS == 48000) {
306 L = 16;
307 L2 = 4;
308 } else {
309 return -1;
310 }
311
312 for (i = 0; i < num_bands; ++i) {
313 if (in_near[i] != out[i]) {
314 // Only needed if they don't already point to the same place.
315 memcpy(out[i], in_near[i], 10 * L * sizeof(in_near[i][0]));
316 }
317 }
318 // VAD for near end
319 logratio = WebRtcAgc_ProcessVad(&stt->vadNearend, out[0], L * 10);
320
321 // Account for far end VAD
322 if (stt->vadFarend.counter > 10) {
323 tmp32 = 3 * logratio;
324 logratio = (int16_t)((tmp32 - stt->vadFarend.logRatio) >> 2);
325 }
326
327 // Determine decay factor depending on VAD
328 // upper_thr = 1.0f;
329 // lower_thr = 0.25f;
330 upper_thr = 1024; // Q10
331 lower_thr = 0; // Q10
332 if (logratio > upper_thr) {
333 // decay = -2^17 / DecayTime; -> -65
334 decay = -65;
335 } else if (logratio < lower_thr) {
336 decay = 0;
337 } else {
338 // decay = (int16_t)(((lower_thr - logratio)
339 // * (2^27/(DecayTime*(upper_thr-lower_thr)))) >> 10);
340 // SUBSTITUTED: 2^27/(DecayTime*(upper_thr-lower_thr)) -> 65
341 tmp32 = (lower_thr - logratio) * 65;
342 decay = (int16_t)(tmp32 >> 10);
343 }
344
345 // adjust decay factor for long silence (detected as low standard deviation)
346 // This is only done in the adaptive modes
347 if (stt->agcMode != kAgcModeFixedDigital) {
348 if (stt->vadNearend.stdLongTerm < 4000) {
349 decay = 0;
350 } else if (stt->vadNearend.stdLongTerm < 8096) {
351 // decay = (int16_t)(((stt->vadNearend.stdLongTerm - 4000) * decay) >>
352 // 12);
353 tmp32 = (stt->vadNearend.stdLongTerm - 4000) * decay;
354 decay = (int16_t)(tmp32 >> 12);
355 }
356
357 if (lowlevelSignal != 0) {
358 decay = 0;
359 }
360 }
361 #ifdef WEBRTC_AGC_DEBUG_DUMP
362 stt->frameCounter++;
363 fprintf(stt->logFile, "%5.2f\t%d\t%d\t%d\t", (float)(stt->frameCounter) / 100,
364 logratio, decay, stt->vadNearend.stdLongTerm);
365 #endif
366 // Find max amplitude per sub frame
367 // iterate over sub frames
368 for (k = 0; k < 10; k++) {
369 // iterate over samples
370 max_nrg = 0;
371 for (n = 0; n < L; n++) {
372 int32_t nrg = out[0][k * L + n] * out[0][k * L + n];
373 if (nrg > max_nrg) {
374 max_nrg = nrg;
375 }
376 }
377 env[k] = max_nrg;
378 }
379
380 // Calculate gain per sub frame
381 gains[0] = stt->gain;
382 for (k = 0; k < 10; k++) {
383 // Fast envelope follower
384 // decay time = -131000 / -1000 = 131 (ms)
385 stt->capacitorFast =
386 AGC_SCALEDIFF32(-1000, stt->capacitorFast, stt->capacitorFast);
387 if (env[k] > stt->capacitorFast) {
388 stt->capacitorFast = env[k];
389 }
390 // Slow envelope follower
391 if (env[k] > stt->capacitorSlow) {
392 // increase capacitorSlow
393 stt->capacitorSlow = AGC_SCALEDIFF32(500, (env[k] - stt->capacitorSlow),
394 stt->capacitorSlow);
395 } else {
396 // decrease capacitorSlow
397 stt->capacitorSlow =
398 AGC_SCALEDIFF32(decay, stt->capacitorSlow, stt->capacitorSlow);
399 }
400
401 // use maximum of both capacitors as current level
402 if (stt->capacitorFast > stt->capacitorSlow) {
403 cur_level = stt->capacitorFast;
404 } else {
405 cur_level = stt->capacitorSlow;
406 }
407 // Translate signal level into gain, using a piecewise linear approximation
408 // find number of leading zeros
409 zeros = WebRtcSpl_NormU32((uint32_t)cur_level);
410 if (cur_level == 0) {
411 zeros = 31;
412 }
413 tmp32 = ((uint32_t)cur_level << zeros) & 0x7FFFFFFF;
414 frac = (int16_t)(tmp32 >> 19); // Q12.
415 // Interpolate between gainTable[zeros] and gainTable[zeros-1].
416 tmp32 = ((stt->gainTable[zeros - 1] - stt->gainTable[zeros]) *
417 (int64_t)frac) >> 12;
418 gains[k + 1] = stt->gainTable[zeros] + tmp32;
419 #ifdef WEBRTC_AGC_DEBUG_DUMP
420 if (k == 0) {
421 fprintf(stt->logFile, "%d\t%d\t%d\t%d\t%d\n", env[0], cur_level,
422 stt->capacitorFast, stt->capacitorSlow, zeros);
423 }
424 #endif
425 }
426
427 // Gate processing (lower gain during absence of speech)
428 zeros = (zeros << 9) - (frac >> 3);
429 // find number of leading zeros
430 zeros_fast = WebRtcSpl_NormU32((uint32_t)stt->capacitorFast);
431 if (stt->capacitorFast == 0) {
432 zeros_fast = 31;
433 }
434 tmp32 = ((uint32_t)stt->capacitorFast << zeros_fast) & 0x7FFFFFFF;
435 zeros_fast <<= 9;
436 zeros_fast -= (int16_t)(tmp32 >> 22);
437
438 gate = 1000 + zeros_fast - zeros - stt->vadNearend.stdShortTerm;
439
440 if (gate < 0) {
441 stt->gatePrevious = 0;
442 } else {
443 tmp32 = stt->gatePrevious * 7;
444 gate = (int16_t)((gate + tmp32) >> 3);
445 stt->gatePrevious = gate;
446 }
447 // gate < 0 -> no gate
448 // gate > 2500 -> max gate
449 if (gate > 0) {
450 if (gate < 2500) {
451 gain_adj = (2500 - gate) >> 5;
452 } else {
453 gain_adj = 0;
454 }
455 for (k = 0; k < 10; k++) {
456 if ((gains[k + 1] - stt->gainTable[0]) > 8388608) {
457 // To prevent wraparound
458 tmp32 = (gains[k + 1] - stt->gainTable[0]) >> 8;
459 tmp32 *= 178 + gain_adj;
460 } else {
461 tmp32 = (gains[k + 1] - stt->gainTable[0]) * (178 + gain_adj);
462 tmp32 >>= 8;
463 }
464 gains[k + 1] = stt->gainTable[0] + tmp32;
465 }
466 }
467
468 // Limit gain to avoid overload distortion
469 for (k = 0; k < 10; k++) {
470 // Find a shift of gains[k + 1] such that it can be squared without
471 // overflow, but at least by 10 bits.
472 zeros = 10;
473 if (gains[k + 1] > 47452159) {
474 zeros = 16 - WebRtcSpl_NormW32(gains[k + 1]);
475 }
476 gain32 = (gains[k + 1] >> zeros) + 1;
477 gain32 *= gain32;
478 // check for overflow
479 while (AGC_MUL32((env[k] >> 12) + 1, gain32) >
480 WEBRTC_SPL_SHIFT_W32((int32_t)32767, 2 * (1 - zeros + 10))) {
481 // multiply by 253/256 ==> -0.1 dB
482 if (gains[k + 1] > 8388607) {
483 // Prevent wrap around
484 gains[k + 1] = (gains[k + 1] / 256) * 253;
485 } else {
486 gains[k + 1] = (gains[k + 1] * 253) / 256;
487 }
488 gain32 = (gains[k + 1] >> zeros) + 1;
489 gain32 *= gain32;
490 }
491 }
492 // gain reductions should be done 1 ms earlier than gain increases
493 for (k = 1; k < 10; k++) {
494 if (gains[k] > gains[k + 1]) {
495 gains[k] = gains[k + 1];
496 }
497 }
498 // save start gain for next frame
499 stt->gain = gains[10];
500
501 // Apply gain
502 // handle first sub frame separately
503 delta = (gains[1] - gains[0]) * (1 << (4 - L2));
504 gain32 = gains[0] * (1 << 4);
505 // iterate over samples
506 for (n = 0; n < L; n++) {
507 for (i = 0; i < num_bands; ++i) {
508 out_tmp = (int64_t)out[i][n] * ((gain32 + 127) >> 7) >> 16;
509 if (out_tmp > 4095) {
510 out[i][n] = (int16_t)32767;
511 } else if (out_tmp < -4096) {
512 out[i][n] = (int16_t)-32768;
513 } else {
514 tmp32 = ((int64_t)out[i][n] * (gain32 >> 4)) >> 16;
515 out[i][n] = (int16_t)tmp32;
516 }
517 }
518
519 gain32 += delta;
520 }
521 // iterate over subframes
522 for (k = 1; k < 10; k++) {
523 delta = (gains[k + 1] - gains[k]) * (1 << (4 - L2));
524 gain32 = gains[k] * (1 << 4);
525 // iterate over samples
526 for (n = 0; n < L; n++) {
527 for (i = 0; i < num_bands; ++i) {
528 int64_t tmp64 = ((int64_t)(out[i][k * L + n])) * (gain32 >> 4);
529 tmp64 = tmp64 >> 16;
530 if (tmp64 > 32767) {
531 out[i][k * L + n] = 32767;
532 }
533 else if (tmp64 < -32768) {
534 out[i][k * L + n] = -32768;
535 }
536 else {
537 out[i][k * L + n] = (int16_t)(tmp64);
538 }
539 }
540 gain32 += delta;
541 }
542 }
543
544 return 0;
545 }
546
WebRtcAgc_InitVad(AgcVad * state)547 void WebRtcAgc_InitVad(AgcVad* state) {
548 int16_t k;
549
550 state->HPstate = 0; // state of high pass filter
551 state->logRatio = 0; // log( P(active) / P(inactive) )
552 // average input level (Q10)
553 state->meanLongTerm = 15 << 10;
554
555 // variance of input level (Q8)
556 state->varianceLongTerm = 500 << 8;
557
558 state->stdLongTerm = 0; // standard deviation of input level in dB
559 // short-term average input level (Q10)
560 state->meanShortTerm = 15 << 10;
561
562 // short-term variance of input level (Q8)
563 state->varianceShortTerm = 500 << 8;
564
565 state->stdShortTerm =
566 0; // short-term standard deviation of input level in dB
567 state->counter = 3; // counts updates
568 for (k = 0; k < 8; k++) {
569 // downsampling filter
570 state->downState[k] = 0;
571 }
572 }
573
WebRtcAgc_ProcessVad(AgcVad * state,const int16_t * in,size_t nrSamples)574 int16_t WebRtcAgc_ProcessVad(AgcVad* state, // (i) VAD state
575 const int16_t* in, // (i) Speech signal
576 size_t nrSamples) // (i) number of samples
577 {
578 uint32_t nrg;
579 int32_t out, tmp32, tmp32b;
580 uint16_t tmpU16;
581 int16_t k, subfr, tmp16;
582 int16_t buf1[8];
583 int16_t buf2[4];
584 int16_t HPstate;
585 int16_t zeros, dB;
586 int64_t tmp64;
587
588 // process in 10 sub frames of 1 ms (to save on memory)
589 nrg = 0;
590 HPstate = state->HPstate;
591 for (subfr = 0; subfr < 10; subfr++) {
592 // downsample to 4 kHz
593 if (nrSamples == 160) {
594 for (k = 0; k < 8; k++) {
595 tmp32 = (int32_t)in[2 * k] + (int32_t)in[2 * k + 1];
596 tmp32 >>= 1;
597 buf1[k] = (int16_t)tmp32;
598 }
599 in += 16;
600
601 WebRtcSpl_DownsampleBy2(buf1, 8, buf2, state->downState);
602 } else {
603 WebRtcSpl_DownsampleBy2(in, 8, buf2, state->downState);
604 in += 8;
605 }
606
607 // high pass filter and compute energy
608 for (k = 0; k < 4; k++) {
609 out = buf2[k] + HPstate;
610 tmp32 = 600 * out;
611 HPstate = (int16_t)((tmp32 >> 10) - buf2[k]);
612
613 // Add 'out * out / 2**6' to 'nrg' in a non-overflowing
614 // way. Guaranteed to work as long as 'out * out / 2**6' fits in
615 // an int32_t.
616 nrg += out * (out / (1 << 6));
617 nrg += out * (out % (1 << 6)) / (1 << 6);
618 }
619 }
620 state->HPstate = HPstate;
621
622 // find number of leading zeros
623 if (!(0xFFFF0000 & nrg)) {
624 zeros = 16;
625 } else {
626 zeros = 0;
627 }
628 if (!(0xFF000000 & (nrg << zeros))) {
629 zeros += 8;
630 }
631 if (!(0xF0000000 & (nrg << zeros))) {
632 zeros += 4;
633 }
634 if (!(0xC0000000 & (nrg << zeros))) {
635 zeros += 2;
636 }
637 if (!(0x80000000 & (nrg << zeros))) {
638 zeros += 1;
639 }
640
641 // energy level (range {-32..30}) (Q10)
642 dB = (15 - zeros) * (1 << 11);
643
644 // Update statistics
645
646 if (state->counter < kAvgDecayTime) {
647 // decay time = AvgDecTime * 10 ms
648 state->counter++;
649 }
650
651 // update short-term estimate of mean energy level (Q10)
652 tmp32 = state->meanShortTerm * 15 + dB;
653 state->meanShortTerm = (int16_t)(tmp32 >> 4);
654
655 // update short-term estimate of variance in energy level (Q8)
656 tmp32 = (dB * dB) >> 12;
657 tmp32 += state->varianceShortTerm * 15;
658 state->varianceShortTerm = tmp32 / 16;
659
660 // update short-term estimate of standard deviation in energy level (Q10)
661 tmp32 = state->meanShortTerm * state->meanShortTerm;
662 tmp32 = (state->varianceShortTerm << 12) - tmp32;
663 state->stdShortTerm = (int16_t)WebRtcSpl_Sqrt(tmp32);
664
665 // update long-term estimate of mean energy level (Q10)
666 tmp32 = state->meanLongTerm * state->counter + dB;
667 state->meanLongTerm =
668 WebRtcSpl_DivW32W16ResW16(tmp32, WebRtcSpl_AddSatW16(state->counter, 1));
669
670 // update long-term estimate of variance in energy level (Q8)
671 tmp32 = (dB * dB) >> 12;
672 tmp32 += state->varianceLongTerm * state->counter;
673 state->varianceLongTerm =
674 WebRtcSpl_DivW32W16(tmp32, WebRtcSpl_AddSatW16(state->counter, 1));
675
676 // update long-term estimate of standard deviation in energy level (Q10)
677 tmp32 = state->meanLongTerm * state->meanLongTerm;
678 tmp32 = (state->varianceLongTerm << 12) - tmp32;
679 state->stdLongTerm = (int16_t)WebRtcSpl_Sqrt(tmp32);
680
681 // update voice activity measure (Q10)
682 tmp16 = 3 << 12;
683 // TODO(bjornv): (dB - state->meanLongTerm) can overflow, e.g., in
684 // ApmTest.Process unit test. Previously the macro WEBRTC_SPL_MUL_16_16()
685 // was used, which did an intermediate cast to (int16_t), hence losing
686 // significant bits. This cause logRatio to max out positive, rather than
687 // negative. This is a bug, but has very little significance.
688 tmp32 = tmp16 * (int16_t)(dB - state->meanLongTerm);
689 tmp32 = WebRtcSpl_DivW32W16(tmp32, state->stdLongTerm);
690 tmpU16 = (13 << 12);
691 tmp32b = WEBRTC_SPL_MUL_16_U16(state->logRatio, tmpU16);
692 tmp64 = tmp32;
693 tmp64 += tmp32b >> 10;
694 tmp64 >>= 6;
695
696 // limit
697 if (tmp64 > 2048) {
698 tmp64 = 2048;
699 } else if (tmp64 < -2048) {
700 tmp64 = -2048;
701 }
702 state->logRatio = (int16_t)tmp64;
703
704 return state->logRatio; // Q10
705 }
706