Tristan Matthews | 0a329cc | 2013-07-17 13:20:14 -0400 | [diff] [blame^] | 1 | /* Copyright (C) 2003 Epic Games (written by Jean-Marc Valin) |
| 2 | Copyright (C) 2004-2006 Epic Games |
| 3 | |
| 4 | File: preprocess.c |
| 5 | Preprocessor with denoising based on the algorithm by Ephraim and Malah |
| 6 | |
| 7 | Redistribution and use in source and binary forms, with or without |
| 8 | modification, are permitted provided that the following conditions are |
| 9 | met: |
| 10 | |
| 11 | 1. Redistributions of source code must retain the above copyright notice, |
| 12 | this list of conditions and the following disclaimer. |
| 13 | |
| 14 | 2. Redistributions in binary form must reproduce the above copyright |
| 15 | notice, this list of conditions and the following disclaimer in the |
| 16 | documentation and/or other materials provided with the distribution. |
| 17 | |
| 18 | 3. The name of the author may not be used to endorse or promote products |
| 19 | derived from this software without specific prior written permission. |
| 20 | |
| 21 | THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR |
| 22 | IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES |
| 23 | OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE |
| 24 | DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, |
| 25 | INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES |
| 26 | (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR |
| 27 | SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 28 | HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, |
| 29 | STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN |
| 30 | ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
| 31 | POSSIBILITY OF SUCH DAMAGE. |
| 32 | */ |
| 33 | |
| 34 | |
| 35 | /* |
| 36 | Recommended papers: |
| 37 | |
| 38 | Y. Ephraim and D. Malah, "Speech enhancement using minimum mean-square error |
| 39 | short-time spectral amplitude estimator". IEEE Transactions on Acoustics, |
| 40 | Speech and Signal Processing, vol. ASSP-32, no. 6, pp. 1109-1121, 1984. |
| 41 | |
| 42 | Y. Ephraim and D. Malah, "Speech enhancement using minimum mean-square error |
| 43 | log-spectral amplitude estimator". IEEE Transactions on Acoustics, Speech and |
| 44 | Signal Processing, vol. ASSP-33, no. 2, pp. 443-445, 1985. |
| 45 | |
| 46 | I. Cohen and B. Berdugo, "Speech enhancement for non-stationary noise environments". |
| 47 | Signal Processing, vol. 81, no. 2, pp. 2403-2418, 2001. |
| 48 | |
| 49 | Stefan Gustafsson, Rainer Martin, Peter Jax, and Peter Vary. "A psychoacoustic |
| 50 | approach to combined acoustic echo cancellation and noise reduction". IEEE |
| 51 | Transactions on Speech and Audio Processing, 2002. |
| 52 | |
| 53 | J.-M. Valin, J. Rouat, and F. Michaud, "Microphone array post-filter for separation |
| 54 | of simultaneous non-stationary sources". In Proceedings IEEE International |
| 55 | Conference on Acoustics, Speech, and Signal Processing, 2004. |
| 56 | */ |
| 57 | |
| 58 | #ifdef HAVE_CONFIG_H |
| 59 | #include "config.h" |
| 60 | #endif |
| 61 | |
| 62 | #include <math.h> |
| 63 | #include "speex/speex_preprocess.h" |
| 64 | #include "speex/speex_echo.h" |
| 65 | #include "arch.h" |
| 66 | #include "fftwrap.h" |
| 67 | #include "filterbank.h" |
| 68 | #include "math_approx.h" |
| 69 | #include "os_support.h" |
| 70 | |
| 71 | #ifndef M_PI |
| 72 | #define M_PI 3.14159263 |
| 73 | #endif |
| 74 | |
| 75 | #define LOUDNESS_EXP 5.f |
| 76 | #define AMP_SCALE .001f |
| 77 | #define AMP_SCALE_1 1000.f |
| 78 | |
| 79 | #define NB_BANDS 24 |
| 80 | |
| 81 | #define SPEECH_PROB_START_DEFAULT QCONST16(0.35f,15) |
| 82 | #define SPEECH_PROB_CONTINUE_DEFAULT QCONST16(0.20f,15) |
| 83 | #define NOISE_SUPPRESS_DEFAULT -15 |
| 84 | #define ECHO_SUPPRESS_DEFAULT -40 |
| 85 | #define ECHO_SUPPRESS_ACTIVE_DEFAULT -15 |
| 86 | |
| 87 | #ifndef NULL |
| 88 | #define NULL 0 |
| 89 | #endif |
| 90 | |
| 91 | #define SQR(x) ((x)*(x)) |
| 92 | #define SQR16(x) (MULT16_16((x),(x))) |
| 93 | #define SQR16_Q15(x) (MULT16_16_Q15((x),(x))) |
| 94 | |
| 95 | #ifdef FIXED_POINT |
| 96 | static inline spx_word16_t DIV32_16_Q8(spx_word32_t a, spx_word32_t b) |
| 97 | { |
| 98 | if (SHR32(a,7) >= b) |
| 99 | { |
| 100 | return 32767; |
| 101 | } else { |
| 102 | if (b>=QCONST32(1,23)) |
| 103 | { |
| 104 | a = SHR32(a,8); |
| 105 | b = SHR32(b,8); |
| 106 | } |
| 107 | if (b>=QCONST32(1,19)) |
| 108 | { |
| 109 | a = SHR32(a,4); |
| 110 | b = SHR32(b,4); |
| 111 | } |
| 112 | if (b>=QCONST32(1,15)) |
| 113 | { |
| 114 | a = SHR32(a,4); |
| 115 | b = SHR32(b,4); |
| 116 | } |
| 117 | a = SHL32(a,8); |
| 118 | return PDIV32_16(a,b); |
| 119 | } |
| 120 | |
| 121 | } |
| 122 | static inline spx_word16_t DIV32_16_Q15(spx_word32_t a, spx_word32_t b) |
| 123 | { |
| 124 | if (SHR32(a,15) >= b) |
| 125 | { |
| 126 | return 32767; |
| 127 | } else { |
| 128 | if (b>=QCONST32(1,23)) |
| 129 | { |
| 130 | a = SHR32(a,8); |
| 131 | b = SHR32(b,8); |
| 132 | } |
| 133 | if (b>=QCONST32(1,19)) |
| 134 | { |
| 135 | a = SHR32(a,4); |
| 136 | b = SHR32(b,4); |
| 137 | } |
| 138 | if (b>=QCONST32(1,15)) |
| 139 | { |
| 140 | a = SHR32(a,4); |
| 141 | b = SHR32(b,4); |
| 142 | } |
| 143 | a = SHL32(a,15)-a; |
| 144 | return DIV32_16(a,b); |
| 145 | } |
| 146 | } |
| 147 | #define SNR_SCALING 256.f |
| 148 | #define SNR_SCALING_1 0.0039062f |
| 149 | #define SNR_SHIFT 8 |
| 150 | |
| 151 | #define FRAC_SCALING 32767.f |
| 152 | #define FRAC_SCALING_1 3.0518e-05 |
| 153 | #define FRAC_SHIFT 1 |
| 154 | |
| 155 | #define EXPIN_SCALING 2048.f |
| 156 | #define EXPIN_SCALING_1 0.00048828f |
| 157 | #define EXPIN_SHIFT 11 |
| 158 | #define EXPOUT_SCALING_1 1.5259e-05 |
| 159 | |
| 160 | #define NOISE_SHIFT 7 |
| 161 | |
| 162 | #else |
| 163 | |
| 164 | #define DIV32_16_Q8(a,b) ((a)/(b)) |
| 165 | #define DIV32_16_Q15(a,b) ((a)/(b)) |
| 166 | #define SNR_SCALING 1.f |
| 167 | #define SNR_SCALING_1 1.f |
| 168 | #define SNR_SHIFT 0 |
| 169 | #define FRAC_SCALING 1.f |
| 170 | #define FRAC_SCALING_1 1.f |
| 171 | #define FRAC_SHIFT 0 |
| 172 | #define NOISE_SHIFT 0 |
| 173 | |
| 174 | #define EXPIN_SCALING 1.f |
| 175 | #define EXPIN_SCALING_1 1.f |
| 176 | #define EXPOUT_SCALING_1 1.f |
| 177 | |
| 178 | #endif |
| 179 | |
| 180 | /** Speex pre-processor state. */ |
| 181 | struct SpeexPreprocessState_ { |
| 182 | /* Basic info */ |
| 183 | int frame_size; /**< Number of samples processed each time */ |
| 184 | int ps_size; /**< Number of points in the power spectrum */ |
| 185 | int sampling_rate; /**< Sampling rate of the input/output */ |
| 186 | int nbands; |
| 187 | FilterBank *bank; |
| 188 | |
| 189 | /* Parameters */ |
| 190 | int denoise_enabled; |
| 191 | int vad_enabled; |
| 192 | int dereverb_enabled; |
| 193 | spx_word16_t reverb_decay; |
| 194 | spx_word16_t reverb_level; |
| 195 | spx_word16_t speech_prob_start; |
| 196 | spx_word16_t speech_prob_continue; |
| 197 | int noise_suppress; |
| 198 | int echo_suppress; |
| 199 | int echo_suppress_active; |
| 200 | SpeexEchoState *echo_state; |
| 201 | |
| 202 | spx_word16_t speech_prob; /**< Probability last frame was speech */ |
| 203 | |
| 204 | /* DSP-related arrays */ |
| 205 | spx_word16_t *frame; /**< Processing frame (2*ps_size) */ |
| 206 | spx_word16_t *ft; /**< Processing frame in freq domain (2*ps_size) */ |
| 207 | spx_word32_t *ps; /**< Current power spectrum */ |
| 208 | spx_word16_t *gain2; /**< Adjusted gains */ |
| 209 | spx_word16_t *gain_floor; /**< Minimum gain allowed */ |
| 210 | spx_word16_t *window; /**< Analysis/Synthesis window */ |
| 211 | spx_word32_t *noise; /**< Noise estimate */ |
| 212 | spx_word32_t *reverb_estimate; /**< Estimate of reverb energy */ |
| 213 | spx_word32_t *old_ps; /**< Power spectrum for last frame */ |
| 214 | spx_word16_t *gain; /**< Ephraim Malah gain */ |
| 215 | spx_word16_t *prior; /**< A-priori SNR */ |
| 216 | spx_word16_t *post; /**< A-posteriori SNR */ |
| 217 | |
| 218 | spx_word32_t *S; /**< Smoothed power spectrum */ |
| 219 | spx_word32_t *Smin; /**< See Cohen paper */ |
| 220 | spx_word32_t *Stmp; /**< See Cohen paper */ |
| 221 | int *update_prob; /**< Probability of speech presence for noise update */ |
| 222 | |
| 223 | spx_word16_t *zeta; /**< Smoothed a priori SNR */ |
| 224 | spx_word32_t *echo_noise; |
| 225 | spx_word32_t *residual_echo; |
| 226 | |
| 227 | /* Misc */ |
| 228 | spx_word16_t *inbuf; /**< Input buffer (overlapped analysis) */ |
| 229 | spx_word16_t *outbuf; /**< Output buffer (for overlap and add) */ |
| 230 | |
| 231 | /* AGC stuff, only for floating point for now */ |
| 232 | #ifndef FIXED_POINT |
| 233 | int agc_enabled; |
| 234 | float agc_level; |
| 235 | float loudness_accum; |
| 236 | float *loudness_weight; /**< Perceptual loudness curve */ |
| 237 | float loudness; /**< Loudness estimate */ |
| 238 | float agc_gain; /**< Current AGC gain */ |
| 239 | float max_gain; /**< Maximum gain allowed */ |
| 240 | float max_increase_step; /**< Maximum increase in gain from one frame to another */ |
| 241 | float max_decrease_step; /**< Maximum decrease in gain from one frame to another */ |
| 242 | float prev_loudness; /**< Loudness of previous frame */ |
| 243 | float init_max; /**< Current gain limit during initialisation */ |
| 244 | #endif |
| 245 | int nb_adapt; /**< Number of frames used for adaptation so far */ |
| 246 | int was_speech; |
| 247 | int min_count; /**< Number of frames processed so far */ |
| 248 | void *fft_lookup; /**< Lookup table for the FFT */ |
| 249 | #ifdef FIXED_POINT |
| 250 | int frame_shift; |
| 251 | #endif |
| 252 | }; |
| 253 | |
| 254 | |
| 255 | static void conj_window(spx_word16_t *w, int len) |
| 256 | { |
| 257 | int i; |
| 258 | for (i=0;i<len;i++) |
| 259 | { |
| 260 | spx_word16_t tmp; |
| 261 | #ifdef FIXED_POINT |
| 262 | spx_word16_t x = DIV32_16(MULT16_16(32767,i),len); |
| 263 | #else |
| 264 | spx_word16_t x = DIV32_16(MULT16_16(QCONST16(4.f,13),i),len); |
| 265 | #endif |
| 266 | int inv=0; |
| 267 | if (x<QCONST16(1.f,13)) |
| 268 | { |
| 269 | } else if (x<QCONST16(2.f,13)) |
| 270 | { |
| 271 | x=QCONST16(2.f,13)-x; |
| 272 | inv=1; |
| 273 | } else if (x<QCONST16(3.f,13)) |
| 274 | { |
| 275 | x=x-QCONST16(2.f,13); |
| 276 | inv=1; |
| 277 | } else { |
| 278 | x=QCONST16(2.f,13)-x+QCONST16(2.f,13); /* 4 - x */ |
| 279 | } |
| 280 | x = MULT16_16_Q14(QCONST16(1.271903f,14), x); |
| 281 | tmp = SQR16_Q15(QCONST16(.5f,15)-MULT16_16_P15(QCONST16(.5f,15),spx_cos_norm(SHL32(EXTEND32(x),2)))); |
| 282 | if (inv) |
| 283 | tmp=SUB16(Q15_ONE,tmp); |
| 284 | w[i]=spx_sqrt(SHL32(EXTEND32(tmp),15)); |
| 285 | } |
| 286 | } |
| 287 | |
| 288 | |
| 289 | #ifdef FIXED_POINT |
| 290 | /* This function approximates the gain function |
| 291 | y = gamma(1.25)^2 * M(-.25;1;-x) / sqrt(x) |
| 292 | which multiplied by xi/(1+xi) is the optimal gain |
| 293 | in the loudness domain ( sqrt[amplitude] ) |
| 294 | Input in Q11 format, output in Q15 |
| 295 | */ |
| 296 | static inline spx_word32_t hypergeom_gain(spx_word32_t xx) |
| 297 | { |
| 298 | int ind; |
| 299 | spx_word16_t frac; |
| 300 | /* Q13 table */ |
| 301 | static const spx_word16_t table[21] = { |
| 302 | 6730, 8357, 9868, 11267, 12563, 13770, 14898, |
| 303 | 15959, 16961, 17911, 18816, 19682, 20512, 21311, |
| 304 | 22082, 22827, 23549, 24250, 24931, 25594, 26241}; |
| 305 | ind = SHR32(xx,10); |
| 306 | if (ind<0) |
| 307 | return Q15_ONE; |
| 308 | if (ind>19) |
| 309 | return ADD32(EXTEND32(Q15_ONE),EXTEND32(DIV32_16(QCONST32(.1296,23), SHR32(xx,EXPIN_SHIFT-SNR_SHIFT)))); |
| 310 | frac = SHL32(xx-SHL32(ind,10),5); |
| 311 | return SHL32(DIV32_16(PSHR32(MULT16_16(Q15_ONE-frac,table[ind]) + MULT16_16(frac,table[ind+1]),7),(spx_sqrt(SHL32(xx,15)+6711))),7); |
| 312 | } |
| 313 | |
| 314 | static inline spx_word16_t qcurve(spx_word16_t x) |
| 315 | { |
| 316 | x = MAX16(x, 1); |
| 317 | return DIV32_16(SHL32(EXTEND32(32767),9),ADD16(512,MULT16_16_Q15(QCONST16(.60f,15),DIV32_16(32767,x)))); |
| 318 | } |
| 319 | |
| 320 | /* Compute the gain floor based on different floors for the background noise and residual echo */ |
| 321 | static void compute_gain_floor(int noise_suppress, int effective_echo_suppress, spx_word32_t *noise, spx_word32_t *echo, spx_word16_t *gain_floor, int len) |
| 322 | { |
| 323 | int i; |
| 324 | |
| 325 | if (noise_suppress > effective_echo_suppress) |
| 326 | { |
| 327 | spx_word16_t noise_gain, gain_ratio; |
| 328 | noise_gain = EXTRACT16(MIN32(Q15_ONE,SHR32(spx_exp(MULT16_16(QCONST16(0.11513,11),noise_suppress)),1))); |
| 329 | gain_ratio = EXTRACT16(MIN32(Q15_ONE,SHR32(spx_exp(MULT16_16(QCONST16(.2302585f,11),effective_echo_suppress-noise_suppress)),1))); |
| 330 | |
| 331 | /* gain_floor = sqrt [ (noise*noise_floor + echo*echo_floor) / (noise+echo) ] */ |
| 332 | for (i=0;i<len;i++) |
| 333 | gain_floor[i] = MULT16_16_Q15(noise_gain, |
| 334 | spx_sqrt(SHL32(EXTEND32(DIV32_16_Q15(PSHR32(noise[i],NOISE_SHIFT) + MULT16_32_Q15(gain_ratio,echo[i]), |
| 335 | (1+PSHR32(noise[i],NOISE_SHIFT) + echo[i]) )),15))); |
| 336 | } else { |
| 337 | spx_word16_t echo_gain, gain_ratio; |
| 338 | echo_gain = EXTRACT16(MIN32(Q15_ONE,SHR32(spx_exp(MULT16_16(QCONST16(0.11513,11),effective_echo_suppress)),1))); |
| 339 | gain_ratio = EXTRACT16(MIN32(Q15_ONE,SHR32(spx_exp(MULT16_16(QCONST16(.2302585f,11),noise_suppress-effective_echo_suppress)),1))); |
| 340 | |
| 341 | /* gain_floor = sqrt [ (noise*noise_floor + echo*echo_floor) / (noise+echo) ] */ |
| 342 | for (i=0;i<len;i++) |
| 343 | gain_floor[i] = MULT16_16_Q15(echo_gain, |
| 344 | spx_sqrt(SHL32(EXTEND32(DIV32_16_Q15(MULT16_32_Q15(gain_ratio,PSHR32(noise[i],NOISE_SHIFT)) + echo[i], |
| 345 | (1+PSHR32(noise[i],NOISE_SHIFT) + echo[i]) )),15))); |
| 346 | } |
| 347 | } |
| 348 | |
| 349 | #else |
| 350 | /* This function approximates the gain function |
| 351 | y = gamma(1.25)^2 * M(-.25;1;-x) / sqrt(x) |
| 352 | which multiplied by xi/(1+xi) is the optimal gain |
| 353 | in the loudness domain ( sqrt[amplitude] ) |
| 354 | */ |
| 355 | static inline spx_word32_t hypergeom_gain(spx_word32_t xx) |
| 356 | { |
| 357 | int ind; |
| 358 | float integer, frac; |
| 359 | float x; |
| 360 | static const float table[21] = { |
| 361 | 0.82157f, 1.02017f, 1.20461f, 1.37534f, 1.53363f, 1.68092f, 1.81865f, |
| 362 | 1.94811f, 2.07038f, 2.18638f, 2.29688f, 2.40255f, 2.50391f, 2.60144f, |
| 363 | 2.69551f, 2.78647f, 2.87458f, 2.96015f, 3.04333f, 3.12431f, 3.20326f}; |
| 364 | x = EXPIN_SCALING_1*xx; |
| 365 | integer = floor(2*x); |
| 366 | ind = (int)integer; |
| 367 | if (ind<0) |
| 368 | return FRAC_SCALING; |
| 369 | if (ind>19) |
| 370 | return FRAC_SCALING*(1+.1296/x); |
| 371 | frac = 2*x-integer; |
| 372 | return FRAC_SCALING*((1-frac)*table[ind] + frac*table[ind+1])/sqrt(x+.0001f); |
| 373 | } |
| 374 | |
| 375 | static inline spx_word16_t qcurve(spx_word16_t x) |
| 376 | { |
| 377 | return 1.f/(1.f+.15f/(SNR_SCALING_1*x)); |
| 378 | } |
| 379 | |
| 380 | static void compute_gain_floor(int noise_suppress, int effective_echo_suppress, spx_word32_t *noise, spx_word32_t *echo, spx_word16_t *gain_floor, int len) |
| 381 | { |
| 382 | int i; |
| 383 | float echo_floor; |
| 384 | float noise_floor; |
| 385 | |
| 386 | noise_floor = exp(.2302585f*noise_suppress); |
| 387 | echo_floor = exp(.2302585f*effective_echo_suppress); |
| 388 | |
| 389 | /* Compute the gain floor based on different floors for the background noise and residual echo */ |
| 390 | for (i=0;i<len;i++) |
| 391 | gain_floor[i] = FRAC_SCALING*sqrt(noise_floor*PSHR32(noise[i],NOISE_SHIFT) + echo_floor*echo[i])/sqrt(1+PSHR32(noise[i],NOISE_SHIFT) + echo[i]); |
| 392 | } |
| 393 | |
| 394 | #endif |
| 395 | EXPORT SpeexPreprocessState *speex_preprocess_state_init(int frame_size, int sampling_rate) |
| 396 | { |
| 397 | int i; |
| 398 | int N, N3, N4, M; |
| 399 | |
| 400 | SpeexPreprocessState *st = (SpeexPreprocessState *)speex_alloc(sizeof(SpeexPreprocessState)); |
| 401 | st->frame_size = frame_size; |
| 402 | |
| 403 | /* Round ps_size down to the nearest power of two */ |
| 404 | #if 0 |
| 405 | i=1; |
| 406 | st->ps_size = st->frame_size; |
| 407 | while(1) |
| 408 | { |
| 409 | if (st->ps_size & ~i) |
| 410 | { |
| 411 | st->ps_size &= ~i; |
| 412 | i<<=1; |
| 413 | } else { |
| 414 | break; |
| 415 | } |
| 416 | } |
| 417 | |
| 418 | |
| 419 | if (st->ps_size < 3*st->frame_size/4) |
| 420 | st->ps_size = st->ps_size * 3 / 2; |
| 421 | #else |
| 422 | st->ps_size = st->frame_size; |
| 423 | #endif |
| 424 | |
| 425 | N = st->ps_size; |
| 426 | N3 = 2*N - st->frame_size; |
| 427 | N4 = st->frame_size - N3; |
| 428 | |
| 429 | st->sampling_rate = sampling_rate; |
| 430 | st->denoise_enabled = 1; |
| 431 | st->vad_enabled = 0; |
| 432 | st->dereverb_enabled = 0; |
| 433 | st->reverb_decay = 0; |
| 434 | st->reverb_level = 0; |
| 435 | st->noise_suppress = NOISE_SUPPRESS_DEFAULT; |
| 436 | st->echo_suppress = ECHO_SUPPRESS_DEFAULT; |
| 437 | st->echo_suppress_active = ECHO_SUPPRESS_ACTIVE_DEFAULT; |
| 438 | |
| 439 | st->speech_prob_start = SPEECH_PROB_START_DEFAULT; |
| 440 | st->speech_prob_continue = SPEECH_PROB_CONTINUE_DEFAULT; |
| 441 | |
| 442 | st->echo_state = NULL; |
| 443 | |
| 444 | st->nbands = NB_BANDS; |
| 445 | M = st->nbands; |
| 446 | st->bank = filterbank_new(M, sampling_rate, N, 1); |
| 447 | |
| 448 | st->frame = (spx_word16_t*)speex_alloc(2*N*sizeof(spx_word16_t)); |
| 449 | st->window = (spx_word16_t*)speex_alloc(2*N*sizeof(spx_word16_t)); |
| 450 | st->ft = (spx_word16_t*)speex_alloc(2*N*sizeof(spx_word16_t)); |
| 451 | |
| 452 | st->ps = (spx_word32_t*)speex_alloc((N+M)*sizeof(spx_word32_t)); |
| 453 | st->noise = (spx_word32_t*)speex_alloc((N+M)*sizeof(spx_word32_t)); |
| 454 | st->echo_noise = (spx_word32_t*)speex_alloc((N+M)*sizeof(spx_word32_t)); |
| 455 | st->residual_echo = (spx_word32_t*)speex_alloc((N+M)*sizeof(spx_word32_t)); |
| 456 | st->reverb_estimate = (spx_word32_t*)speex_alloc((N+M)*sizeof(spx_word32_t)); |
| 457 | st->old_ps = (spx_word32_t*)speex_alloc((N+M)*sizeof(spx_word32_t)); |
| 458 | st->prior = (spx_word16_t*)speex_alloc((N+M)*sizeof(spx_word16_t)); |
| 459 | st->post = (spx_word16_t*)speex_alloc((N+M)*sizeof(spx_word16_t)); |
| 460 | st->gain = (spx_word16_t*)speex_alloc((N+M)*sizeof(spx_word16_t)); |
| 461 | st->gain2 = (spx_word16_t*)speex_alloc((N+M)*sizeof(spx_word16_t)); |
| 462 | st->gain_floor = (spx_word16_t*)speex_alloc((N+M)*sizeof(spx_word16_t)); |
| 463 | st->zeta = (spx_word16_t*)speex_alloc((N+M)*sizeof(spx_word16_t)); |
| 464 | |
| 465 | st->S = (spx_word32_t*)speex_alloc(N*sizeof(spx_word32_t)); |
| 466 | st->Smin = (spx_word32_t*)speex_alloc(N*sizeof(spx_word32_t)); |
| 467 | st->Stmp = (spx_word32_t*)speex_alloc(N*sizeof(spx_word32_t)); |
| 468 | st->update_prob = (int*)speex_alloc(N*sizeof(int)); |
| 469 | |
| 470 | st->inbuf = (spx_word16_t*)speex_alloc(N3*sizeof(spx_word16_t)); |
| 471 | st->outbuf = (spx_word16_t*)speex_alloc(N3*sizeof(spx_word16_t)); |
| 472 | |
| 473 | conj_window(st->window, 2*N3); |
| 474 | for (i=2*N3;i<2*st->ps_size;i++) |
| 475 | st->window[i]=Q15_ONE; |
| 476 | |
| 477 | if (N4>0) |
| 478 | { |
| 479 | for (i=N3-1;i>=0;i--) |
| 480 | { |
| 481 | st->window[i+N3+N4]=st->window[i+N3]; |
| 482 | st->window[i+N3]=1; |
| 483 | } |
| 484 | } |
| 485 | for (i=0;i<N+M;i++) |
| 486 | { |
| 487 | st->noise[i]=QCONST32(1.f,NOISE_SHIFT); |
| 488 | st->reverb_estimate[i]=0; |
| 489 | st->old_ps[i]=1; |
| 490 | st->gain[i]=Q15_ONE; |
| 491 | st->post[i]=SHL16(1, SNR_SHIFT); |
| 492 | st->prior[i]=SHL16(1, SNR_SHIFT); |
| 493 | } |
| 494 | |
| 495 | for (i=0;i<N;i++) |
| 496 | st->update_prob[i] = 1; |
| 497 | for (i=0;i<N3;i++) |
| 498 | { |
| 499 | st->inbuf[i]=0; |
| 500 | st->outbuf[i]=0; |
| 501 | } |
| 502 | #ifndef FIXED_POINT |
| 503 | st->agc_enabled = 0; |
| 504 | st->agc_level = 8000; |
| 505 | st->loudness_weight = (float*)speex_alloc(N*sizeof(float)); |
| 506 | for (i=0;i<N;i++) |
| 507 | { |
| 508 | float ff=((float)i)*.5*sampling_rate/((float)N); |
| 509 | /*st->loudness_weight[i] = .5f*(1.f/(1.f+ff/8000.f))+1.f*exp(-.5f*(ff-3800.f)*(ff-3800.f)/9e5f);*/ |
| 510 | st->loudness_weight[i] = .35f-.35f*ff/16000.f+.73f*exp(-.5f*(ff-3800)*(ff-3800)/9e5f); |
| 511 | if (st->loudness_weight[i]<.01f) |
| 512 | st->loudness_weight[i]=.01f; |
| 513 | st->loudness_weight[i] *= st->loudness_weight[i]; |
| 514 | } |
| 515 | /*st->loudness = pow(AMP_SCALE*st->agc_level,LOUDNESS_EXP);*/ |
| 516 | st->loudness = 1e-15; |
| 517 | st->agc_gain = 1; |
| 518 | st->max_gain = 30; |
| 519 | st->max_increase_step = exp(0.11513f * 12.*st->frame_size / st->sampling_rate); |
| 520 | st->max_decrease_step = exp(-0.11513f * 40.*st->frame_size / st->sampling_rate); |
| 521 | st->prev_loudness = 1; |
| 522 | st->init_max = 1; |
| 523 | #endif |
| 524 | st->was_speech = 0; |
| 525 | |
| 526 | st->fft_lookup = spx_fft_init(2*N); |
| 527 | |
| 528 | st->nb_adapt=0; |
| 529 | st->min_count=0; |
| 530 | return st; |
| 531 | } |
| 532 | |
| 533 | EXPORT void speex_preprocess_state_destroy(SpeexPreprocessState *st) |
| 534 | { |
| 535 | speex_free(st->frame); |
| 536 | speex_free(st->ft); |
| 537 | speex_free(st->ps); |
| 538 | speex_free(st->gain2); |
| 539 | speex_free(st->gain_floor); |
| 540 | speex_free(st->window); |
| 541 | speex_free(st->noise); |
| 542 | speex_free(st->reverb_estimate); |
| 543 | speex_free(st->old_ps); |
| 544 | speex_free(st->gain); |
| 545 | speex_free(st->prior); |
| 546 | speex_free(st->post); |
| 547 | #ifndef FIXED_POINT |
| 548 | speex_free(st->loudness_weight); |
| 549 | #endif |
| 550 | speex_free(st->echo_noise); |
| 551 | speex_free(st->residual_echo); |
| 552 | |
| 553 | speex_free(st->S); |
| 554 | speex_free(st->Smin); |
| 555 | speex_free(st->Stmp); |
| 556 | speex_free(st->update_prob); |
| 557 | speex_free(st->zeta); |
| 558 | |
| 559 | speex_free(st->inbuf); |
| 560 | speex_free(st->outbuf); |
| 561 | |
| 562 | spx_fft_destroy(st->fft_lookup); |
| 563 | filterbank_destroy(st->bank); |
| 564 | speex_free(st); |
| 565 | } |
| 566 | |
| 567 | /* FIXME: The AGC doesn't work yet with fixed-point*/ |
| 568 | #ifndef FIXED_POINT |
| 569 | static void speex_compute_agc(SpeexPreprocessState *st, spx_word16_t Pframe, spx_word16_t *ft) |
| 570 | { |
| 571 | int i; |
| 572 | int N = st->ps_size; |
| 573 | float target_gain; |
| 574 | float loudness=1.f; |
| 575 | float rate; |
| 576 | |
| 577 | for (i=2;i<N;i++) |
| 578 | { |
| 579 | loudness += 2.f*N*st->ps[i]* st->loudness_weight[i]; |
| 580 | } |
| 581 | loudness=sqrt(loudness); |
| 582 | /*if (loudness < 2*pow(st->loudness, 1.0/LOUDNESS_EXP) && |
| 583 | loudness*2 > pow(st->loudness, 1.0/LOUDNESS_EXP))*/ |
| 584 | if (Pframe>.3f) |
| 585 | { |
| 586 | /*rate=2.0f*Pframe*Pframe/(1+st->nb_loudness_adapt);*/ |
| 587 | rate = .03*Pframe*Pframe; |
| 588 | st->loudness = (1-rate)*st->loudness + (rate)*pow(AMP_SCALE*loudness, LOUDNESS_EXP); |
| 589 | st->loudness_accum = (1-rate)*st->loudness_accum + rate; |
| 590 | if (st->init_max < st->max_gain && st->nb_adapt > 20) |
| 591 | st->init_max *= 1.f + .1f*Pframe*Pframe; |
| 592 | } |
| 593 | /*printf ("%f %f %f %f\n", Pframe, loudness, pow(st->loudness, 1.0f/LOUDNESS_EXP), st->loudness2);*/ |
| 594 | |
| 595 | target_gain = AMP_SCALE*st->agc_level*pow(st->loudness/(1e-4+st->loudness_accum), -1.0f/LOUDNESS_EXP); |
| 596 | |
| 597 | if ((Pframe>.5 && st->nb_adapt > 20) || target_gain < st->agc_gain) |
| 598 | { |
| 599 | if (target_gain > st->max_increase_step*st->agc_gain) |
| 600 | target_gain = st->max_increase_step*st->agc_gain; |
| 601 | if (target_gain < st->max_decrease_step*st->agc_gain && loudness < 10*st->prev_loudness) |
| 602 | target_gain = st->max_decrease_step*st->agc_gain; |
| 603 | if (target_gain > st->max_gain) |
| 604 | target_gain = st->max_gain; |
| 605 | if (target_gain > st->init_max) |
| 606 | target_gain = st->init_max; |
| 607 | |
| 608 | st->agc_gain = target_gain; |
| 609 | } |
| 610 | /*fprintf (stderr, "%f %f %f\n", loudness, (float)AMP_SCALE_1*pow(st->loudness, 1.0f/LOUDNESS_EXP), st->agc_gain);*/ |
| 611 | |
| 612 | for (i=0;i<2*N;i++) |
| 613 | ft[i] *= st->agc_gain; |
| 614 | st->prev_loudness = loudness; |
| 615 | } |
| 616 | #endif |
| 617 | |
| 618 | static void preprocess_analysis(SpeexPreprocessState *st, spx_int16_t *x) |
| 619 | { |
| 620 | int i; |
| 621 | int N = st->ps_size; |
| 622 | int N3 = 2*N - st->frame_size; |
| 623 | int N4 = st->frame_size - N3; |
| 624 | spx_word32_t *ps=st->ps; |
| 625 | |
| 626 | /* 'Build' input frame */ |
| 627 | for (i=0;i<N3;i++) |
| 628 | st->frame[i]=st->inbuf[i]; |
| 629 | for (i=0;i<st->frame_size;i++) |
| 630 | st->frame[N3+i]=x[i]; |
| 631 | |
| 632 | /* Update inbuf */ |
| 633 | for (i=0;i<N3;i++) |
| 634 | st->inbuf[i]=x[N4+i]; |
| 635 | |
| 636 | /* Windowing */ |
| 637 | for (i=0;i<2*N;i++) |
| 638 | st->frame[i] = MULT16_16_Q15(st->frame[i], st->window[i]); |
| 639 | |
| 640 | #ifdef FIXED_POINT |
| 641 | { |
| 642 | spx_word16_t max_val=0; |
| 643 | for (i=0;i<2*N;i++) |
| 644 | max_val = MAX16(max_val, ABS16(st->frame[i])); |
| 645 | st->frame_shift = 14-spx_ilog2(EXTEND32(max_val)); |
| 646 | for (i=0;i<2*N;i++) |
| 647 | st->frame[i] = SHL16(st->frame[i], st->frame_shift); |
| 648 | } |
| 649 | #endif |
| 650 | |
| 651 | /* Perform FFT */ |
| 652 | spx_fft(st->fft_lookup, st->frame, st->ft); |
| 653 | |
| 654 | /* Power spectrum */ |
| 655 | ps[0]=MULT16_16(st->ft[0],st->ft[0]); |
| 656 | for (i=1;i<N;i++) |
| 657 | ps[i]=MULT16_16(st->ft[2*i-1],st->ft[2*i-1]) + MULT16_16(st->ft[2*i],st->ft[2*i]); |
| 658 | for (i=0;i<N;i++) |
| 659 | st->ps[i] = PSHR32(st->ps[i], 2*st->frame_shift); |
| 660 | |
| 661 | filterbank_compute_bank32(st->bank, ps, ps+N); |
| 662 | } |
| 663 | |
| 664 | static void update_noise_prob(SpeexPreprocessState *st) |
| 665 | { |
| 666 | int i; |
| 667 | int min_range; |
| 668 | int N = st->ps_size; |
| 669 | |
| 670 | for (i=1;i<N-1;i++) |
| 671 | st->S[i] = MULT16_32_Q15(QCONST16(.8f,15),st->S[i]) + MULT16_32_Q15(QCONST16(.05f,15),st->ps[i-1]) |
| 672 | + MULT16_32_Q15(QCONST16(.1f,15),st->ps[i]) + MULT16_32_Q15(QCONST16(.05f,15),st->ps[i+1]); |
| 673 | st->S[0] = MULT16_32_Q15(QCONST16(.8f,15),st->S[0]) + MULT16_32_Q15(QCONST16(.2f,15),st->ps[0]); |
| 674 | st->S[N-1] = MULT16_32_Q15(QCONST16(.8f,15),st->S[N-1]) + MULT16_32_Q15(QCONST16(.2f,15),st->ps[N-1]); |
| 675 | |
| 676 | if (st->nb_adapt==1) |
| 677 | { |
| 678 | for (i=0;i<N;i++) |
| 679 | st->Smin[i] = st->Stmp[i] = 0; |
| 680 | } |
| 681 | |
| 682 | if (st->nb_adapt < 100) |
| 683 | min_range = 15; |
| 684 | else if (st->nb_adapt < 1000) |
| 685 | min_range = 50; |
| 686 | else if (st->nb_adapt < 10000) |
| 687 | min_range = 150; |
| 688 | else |
| 689 | min_range = 300; |
| 690 | if (st->min_count > min_range) |
| 691 | { |
| 692 | st->min_count = 0; |
| 693 | for (i=0;i<N;i++) |
| 694 | { |
| 695 | st->Smin[i] = MIN32(st->Stmp[i], st->S[i]); |
| 696 | st->Stmp[i] = st->S[i]; |
| 697 | } |
| 698 | } else { |
| 699 | for (i=0;i<N;i++) |
| 700 | { |
| 701 | st->Smin[i] = MIN32(st->Smin[i], st->S[i]); |
| 702 | st->Stmp[i] = MIN32(st->Stmp[i], st->S[i]); |
| 703 | } |
| 704 | } |
| 705 | for (i=0;i<N;i++) |
| 706 | { |
| 707 | if (MULT16_32_Q15(QCONST16(.4f,15),st->S[i]) > st->Smin[i]) |
| 708 | st->update_prob[i] = 1; |
| 709 | else |
| 710 | st->update_prob[i] = 0; |
| 711 | /*fprintf (stderr, "%f ", st->S[i]/st->Smin[i]);*/ |
| 712 | /*fprintf (stderr, "%f ", st->update_prob[i]);*/ |
| 713 | } |
| 714 | |
| 715 | } |
| 716 | |
| 717 | #define NOISE_OVERCOMPENS 1. |
| 718 | |
| 719 | void speex_echo_get_residual(SpeexEchoState *st, spx_word32_t *Yout, int len); |
| 720 | |
| 721 | EXPORT int speex_preprocess(SpeexPreprocessState *st, spx_int16_t *x, spx_int32_t *echo) |
| 722 | { |
| 723 | return speex_preprocess_run(st, x); |
| 724 | } |
| 725 | |
| 726 | EXPORT int speex_preprocess_run(SpeexPreprocessState *st, spx_int16_t *x) |
| 727 | { |
| 728 | int i; |
| 729 | int M; |
| 730 | int N = st->ps_size; |
| 731 | int N3 = 2*N - st->frame_size; |
| 732 | int N4 = st->frame_size - N3; |
| 733 | spx_word32_t *ps=st->ps; |
| 734 | spx_word32_t Zframe; |
| 735 | spx_word16_t Pframe; |
| 736 | spx_word16_t beta, beta_1; |
| 737 | spx_word16_t effective_echo_suppress; |
| 738 | |
| 739 | st->nb_adapt++; |
| 740 | if (st->nb_adapt>20000) |
| 741 | st->nb_adapt = 20000; |
| 742 | st->min_count++; |
| 743 | |
| 744 | beta = MAX16(QCONST16(.03,15),DIV32_16(Q15_ONE,st->nb_adapt)); |
| 745 | beta_1 = Q15_ONE-beta; |
| 746 | M = st->nbands; |
| 747 | /* Deal with residual echo if provided */ |
| 748 | if (st->echo_state) |
| 749 | { |
| 750 | speex_echo_get_residual(st->echo_state, st->residual_echo, N); |
| 751 | #ifndef FIXED_POINT |
| 752 | /* If there are NaNs or ridiculous values, it'll show up in the DC and we just reset everything to zero */ |
| 753 | if (!(st->residual_echo[0] >=0 && st->residual_echo[0]<N*1e9f)) |
| 754 | { |
| 755 | for (i=0;i<N;i++) |
| 756 | st->residual_echo[i] = 0; |
| 757 | } |
| 758 | #endif |
| 759 | for (i=0;i<N;i++) |
| 760 | st->echo_noise[i] = MAX32(MULT16_32_Q15(QCONST16(.6f,15),st->echo_noise[i]), st->residual_echo[i]); |
| 761 | filterbank_compute_bank32(st->bank, st->echo_noise, st->echo_noise+N); |
| 762 | } else { |
| 763 | for (i=0;i<N+M;i++) |
| 764 | st->echo_noise[i] = 0; |
| 765 | } |
| 766 | preprocess_analysis(st, x); |
| 767 | |
| 768 | update_noise_prob(st); |
| 769 | |
| 770 | /* Noise estimation always updated for the 10 first frames */ |
| 771 | /*if (st->nb_adapt<10) |
| 772 | { |
| 773 | for (i=1;i<N-1;i++) |
| 774 | st->update_prob[i] = 0; |
| 775 | } |
| 776 | */ |
| 777 | |
| 778 | /* Update the noise estimate for the frequencies where it can be */ |
| 779 | for (i=0;i<N;i++) |
| 780 | { |
| 781 | if (!st->update_prob[i] || st->ps[i] < PSHR32(st->noise[i], NOISE_SHIFT)) |
| 782 | st->noise[i] = MAX32(EXTEND32(0),MULT16_32_Q15(beta_1,st->noise[i]) + MULT16_32_Q15(beta,SHL32(st->ps[i],NOISE_SHIFT))); |
| 783 | } |
| 784 | filterbank_compute_bank32(st->bank, st->noise, st->noise+N); |
| 785 | |
| 786 | /* Special case for first frame */ |
| 787 | if (st->nb_adapt==1) |
| 788 | for (i=0;i<N+M;i++) |
| 789 | st->old_ps[i] = ps[i]; |
| 790 | |
| 791 | /* Compute a posteriori SNR */ |
| 792 | for (i=0;i<N+M;i++) |
| 793 | { |
| 794 | spx_word16_t gamma; |
| 795 | |
| 796 | /* Total noise estimate including residual echo and reverberation */ |
| 797 | spx_word32_t tot_noise = ADD32(ADD32(ADD32(EXTEND32(1), PSHR32(st->noise[i],NOISE_SHIFT)) , st->echo_noise[i]) , st->reverb_estimate[i]); |
| 798 | |
| 799 | /* A posteriori SNR = ps/noise - 1*/ |
| 800 | st->post[i] = SUB16(DIV32_16_Q8(ps[i],tot_noise), QCONST16(1.f,SNR_SHIFT)); |
| 801 | st->post[i]=MIN16(st->post[i], QCONST16(100.f,SNR_SHIFT)); |
| 802 | |
| 803 | /* Computing update gamma = .1 + .9*(old/(old+noise))^2 */ |
| 804 | gamma = QCONST16(.1f,15)+MULT16_16_Q15(QCONST16(.89f,15),SQR16_Q15(DIV32_16_Q15(st->old_ps[i],ADD32(st->old_ps[i],tot_noise)))); |
| 805 | |
| 806 | /* A priori SNR update = gamma*max(0,post) + (1-gamma)*old/noise */ |
| 807 | st->prior[i] = EXTRACT16(PSHR32(ADD32(MULT16_16(gamma,MAX16(0,st->post[i])), MULT16_16(Q15_ONE-gamma,DIV32_16_Q8(st->old_ps[i],tot_noise))), 15)); |
| 808 | st->prior[i]=MIN16(st->prior[i], QCONST16(100.f,SNR_SHIFT)); |
| 809 | } |
| 810 | |
| 811 | /*print_vec(st->post, N+M, "");*/ |
| 812 | |
| 813 | /* Recursive average of the a priori SNR. A bit smoothed for the psd components */ |
| 814 | st->zeta[0] = PSHR32(ADD32(MULT16_16(QCONST16(.7f,15),st->zeta[0]), MULT16_16(QCONST16(.3f,15),st->prior[0])),15); |
| 815 | for (i=1;i<N-1;i++) |
| 816 | st->zeta[i] = PSHR32(ADD32(ADD32(ADD32(MULT16_16(QCONST16(.7f,15),st->zeta[i]), MULT16_16(QCONST16(.15f,15),st->prior[i])), |
| 817 | MULT16_16(QCONST16(.075f,15),st->prior[i-1])), MULT16_16(QCONST16(.075f,15),st->prior[i+1])),15); |
| 818 | for (i=N-1;i<N+M;i++) |
| 819 | st->zeta[i] = PSHR32(ADD32(MULT16_16(QCONST16(.7f,15),st->zeta[i]), MULT16_16(QCONST16(.3f,15),st->prior[i])),15); |
| 820 | |
| 821 | /* Speech probability of presence for the entire frame is based on the average filterbank a priori SNR */ |
| 822 | Zframe = 0; |
| 823 | for (i=N;i<N+M;i++) |
| 824 | Zframe = ADD32(Zframe, EXTEND32(st->zeta[i])); |
| 825 | Pframe = QCONST16(.1f,15)+MULT16_16_Q15(QCONST16(.899f,15),qcurve(DIV32_16(Zframe,st->nbands))); |
| 826 | |
| 827 | effective_echo_suppress = EXTRACT16(PSHR32(ADD32(MULT16_16(SUB16(Q15_ONE,Pframe), st->echo_suppress), MULT16_16(Pframe, st->echo_suppress_active)),15)); |
| 828 | |
| 829 | compute_gain_floor(st->noise_suppress, effective_echo_suppress, st->noise+N, st->echo_noise+N, st->gain_floor+N, M); |
| 830 | |
| 831 | /* Compute Ephraim & Malah gain speech probability of presence for each critical band (Bark scale) |
| 832 | Technically this is actually wrong because the EM gaim assumes a slightly different probability |
| 833 | distribution */ |
| 834 | for (i=N;i<N+M;i++) |
| 835 | { |
| 836 | /* See EM and Cohen papers*/ |
| 837 | spx_word32_t theta; |
| 838 | /* Gain from hypergeometric function */ |
| 839 | spx_word32_t MM; |
| 840 | /* Weiner filter gain */ |
| 841 | spx_word16_t prior_ratio; |
| 842 | /* a priority probability of speech presence based on Bark sub-band alone */ |
| 843 | spx_word16_t P1; |
| 844 | /* Speech absence a priori probability (considering sub-band and frame) */ |
| 845 | spx_word16_t q; |
| 846 | #ifdef FIXED_POINT |
| 847 | spx_word16_t tmp; |
| 848 | #endif |
| 849 | |
| 850 | prior_ratio = PDIV32_16(SHL32(EXTEND32(st->prior[i]), 15), ADD16(st->prior[i], SHL32(1,SNR_SHIFT))); |
| 851 | theta = MULT16_32_P15(prior_ratio, QCONST32(1.f,EXPIN_SHIFT)+SHL32(EXTEND32(st->post[i]),EXPIN_SHIFT-SNR_SHIFT)); |
| 852 | |
| 853 | MM = hypergeom_gain(theta); |
| 854 | /* Gain with bound */ |
| 855 | st->gain[i] = EXTRACT16(MIN32(Q15_ONE, MULT16_32_Q15(prior_ratio, MM))); |
| 856 | /* Save old Bark power spectrum */ |
| 857 | st->old_ps[i] = MULT16_32_P15(QCONST16(.2f,15),st->old_ps[i]) + MULT16_32_P15(MULT16_16_P15(QCONST16(.8f,15),SQR16_Q15(st->gain[i])),ps[i]); |
| 858 | |
| 859 | P1 = QCONST16(.199f,15)+MULT16_16_Q15(QCONST16(.8f,15),qcurve (st->zeta[i])); |
| 860 | q = Q15_ONE-MULT16_16_Q15(Pframe,P1); |
| 861 | #ifdef FIXED_POINT |
| 862 | theta = MIN32(theta, EXTEND32(32767)); |
| 863 | /*Q8*/tmp = MULT16_16_Q15((SHL32(1,SNR_SHIFT)+st->prior[i]),EXTRACT16(MIN32(Q15ONE,SHR32(spx_exp(-EXTRACT16(theta)),1)))); |
| 864 | tmp = MIN16(QCONST16(3.,SNR_SHIFT), tmp); /* Prevent overflows in the next line*/ |
| 865 | /*Q8*/tmp = EXTRACT16(PSHR32(MULT16_16(PDIV32_16(SHL32(EXTEND32(q),8),(Q15_ONE-q)),tmp),8)); |
| 866 | st->gain2[i]=DIV32_16(SHL32(EXTEND32(32767),SNR_SHIFT), ADD16(256,tmp)); |
| 867 | #else |
| 868 | st->gain2[i]=1/(1.f + (q/(1.f-q))*(1+st->prior[i])*exp(-theta)); |
| 869 | #endif |
| 870 | } |
| 871 | /* Convert the EM gains and speech prob to linear frequency */ |
| 872 | filterbank_compute_psd16(st->bank,st->gain2+N, st->gain2); |
| 873 | filterbank_compute_psd16(st->bank,st->gain+N, st->gain); |
| 874 | |
| 875 | /* Use 1 for linear gain resolution (best) or 0 for Bark gain resolution (faster) */ |
| 876 | if (1) |
| 877 | { |
| 878 | filterbank_compute_psd16(st->bank,st->gain_floor+N, st->gain_floor); |
| 879 | |
| 880 | /* Compute gain according to the Ephraim-Malah algorithm -- linear frequency */ |
| 881 | for (i=0;i<N;i++) |
| 882 | { |
| 883 | spx_word32_t MM; |
| 884 | spx_word32_t theta; |
| 885 | spx_word16_t prior_ratio; |
| 886 | spx_word16_t tmp; |
| 887 | spx_word16_t p; |
| 888 | spx_word16_t g; |
| 889 | |
| 890 | /* Wiener filter gain */ |
| 891 | prior_ratio = PDIV32_16(SHL32(EXTEND32(st->prior[i]), 15), ADD16(st->prior[i], SHL32(1,SNR_SHIFT))); |
| 892 | theta = MULT16_32_P15(prior_ratio, QCONST32(1.f,EXPIN_SHIFT)+SHL32(EXTEND32(st->post[i]),EXPIN_SHIFT-SNR_SHIFT)); |
| 893 | |
| 894 | /* Optimal estimator for loudness domain */ |
| 895 | MM = hypergeom_gain(theta); |
| 896 | /* EM gain with bound */ |
| 897 | g = EXTRACT16(MIN32(Q15_ONE, MULT16_32_Q15(prior_ratio, MM))); |
| 898 | /* Interpolated speech probability of presence */ |
| 899 | p = st->gain2[i]; |
| 900 | |
| 901 | /* Constrain the gain to be close to the Bark scale gain */ |
| 902 | if (MULT16_16_Q15(QCONST16(.333f,15),g) > st->gain[i]) |
| 903 | g = MULT16_16(3,st->gain[i]); |
| 904 | st->gain[i] = g; |
| 905 | |
| 906 | /* Save old power spectrum */ |
| 907 | st->old_ps[i] = MULT16_32_P15(QCONST16(.2f,15),st->old_ps[i]) + MULT16_32_P15(MULT16_16_P15(QCONST16(.8f,15),SQR16_Q15(st->gain[i])),ps[i]); |
| 908 | |
| 909 | /* Apply gain floor */ |
| 910 | if (st->gain[i] < st->gain_floor[i]) |
| 911 | st->gain[i] = st->gain_floor[i]; |
| 912 | |
| 913 | /* Exponential decay model for reverberation (unused) */ |
| 914 | /*st->reverb_estimate[i] = st->reverb_decay*st->reverb_estimate[i] + st->reverb_decay*st->reverb_level*st->gain[i]*st->gain[i]*st->ps[i];*/ |
| 915 | |
| 916 | /* Take into account speech probability of presence (loudness domain MMSE estimator) */ |
| 917 | /* gain2 = [p*sqrt(gain)+(1-p)*sqrt(gain _floor) ]^2 */ |
| 918 | tmp = MULT16_16_P15(p,spx_sqrt(SHL32(EXTEND32(st->gain[i]),15))) + MULT16_16_P15(SUB16(Q15_ONE,p),spx_sqrt(SHL32(EXTEND32(st->gain_floor[i]),15))); |
| 919 | st->gain2[i]=SQR16_Q15(tmp); |
| 920 | |
| 921 | /* Use this if you want a log-domain MMSE estimator instead */ |
| 922 | /*st->gain2[i] = pow(st->gain[i], p) * pow(st->gain_floor[i],1.f-p);*/ |
| 923 | } |
| 924 | } else { |
| 925 | for (i=N;i<N+M;i++) |
| 926 | { |
| 927 | spx_word16_t tmp; |
| 928 | spx_word16_t p = st->gain2[i]; |
| 929 | st->gain[i] = MAX16(st->gain[i], st->gain_floor[i]); |
| 930 | tmp = MULT16_16_P15(p,spx_sqrt(SHL32(EXTEND32(st->gain[i]),15))) + MULT16_16_P15(SUB16(Q15_ONE,p),spx_sqrt(SHL32(EXTEND32(st->gain_floor[i]),15))); |
| 931 | st->gain2[i]=SQR16_Q15(tmp); |
| 932 | } |
| 933 | filterbank_compute_psd16(st->bank,st->gain2+N, st->gain2); |
| 934 | } |
| 935 | |
| 936 | /* If noise suppression is off, don't apply the gain (but then why call this in the first place!) */ |
| 937 | if (!st->denoise_enabled) |
| 938 | { |
| 939 | for (i=0;i<N+M;i++) |
| 940 | st->gain2[i]=Q15_ONE; |
| 941 | } |
| 942 | |
| 943 | /* Apply computed gain */ |
| 944 | for (i=1;i<N;i++) |
| 945 | { |
| 946 | st->ft[2*i-1] = MULT16_16_P15(st->gain2[i],st->ft[2*i-1]); |
| 947 | st->ft[2*i] = MULT16_16_P15(st->gain2[i],st->ft[2*i]); |
| 948 | } |
| 949 | st->ft[0] = MULT16_16_P15(st->gain2[0],st->ft[0]); |
| 950 | st->ft[2*N-1] = MULT16_16_P15(st->gain2[N-1],st->ft[2*N-1]); |
| 951 | |
| 952 | /*FIXME: This *will* not work for fixed-point */ |
| 953 | #ifndef FIXED_POINT |
| 954 | if (st->agc_enabled) |
| 955 | speex_compute_agc(st, Pframe, st->ft); |
| 956 | #endif |
| 957 | |
| 958 | /* Inverse FFT with 1/N scaling */ |
| 959 | spx_ifft(st->fft_lookup, st->ft, st->frame); |
| 960 | /* Scale back to original (lower) amplitude */ |
| 961 | for (i=0;i<2*N;i++) |
| 962 | st->frame[i] = PSHR16(st->frame[i], st->frame_shift); |
| 963 | |
| 964 | /*FIXME: This *will* not work for fixed-point */ |
| 965 | #ifndef FIXED_POINT |
| 966 | if (st->agc_enabled) |
| 967 | { |
| 968 | float max_sample=0; |
| 969 | for (i=0;i<2*N;i++) |
| 970 | if (fabs(st->frame[i])>max_sample) |
| 971 | max_sample = fabs(st->frame[i]); |
| 972 | if (max_sample>28000.f) |
| 973 | { |
| 974 | float damp = 28000.f/max_sample; |
| 975 | for (i=0;i<2*N;i++) |
| 976 | st->frame[i] *= damp; |
| 977 | } |
| 978 | } |
| 979 | #endif |
| 980 | |
| 981 | /* Synthesis window (for WOLA) */ |
| 982 | for (i=0;i<2*N;i++) |
| 983 | st->frame[i] = MULT16_16_Q15(st->frame[i], st->window[i]); |
| 984 | |
| 985 | /* Perform overlap and add */ |
| 986 | for (i=0;i<N3;i++) |
| 987 | x[i] = st->outbuf[i] + st->frame[i]; |
| 988 | for (i=0;i<N4;i++) |
| 989 | x[N3+i] = st->frame[N3+i]; |
| 990 | |
| 991 | /* Update outbuf */ |
| 992 | for (i=0;i<N3;i++) |
| 993 | st->outbuf[i] = st->frame[st->frame_size+i]; |
| 994 | |
| 995 | /* FIXME: This VAD is a kludge */ |
| 996 | st->speech_prob = Pframe; |
| 997 | if (st->vad_enabled) |
| 998 | { |
| 999 | if (st->speech_prob > st->speech_prob_start || (st->was_speech && st->speech_prob > st->speech_prob_continue)) |
| 1000 | { |
| 1001 | st->was_speech=1; |
| 1002 | return 1; |
| 1003 | } else |
| 1004 | { |
| 1005 | st->was_speech=0; |
| 1006 | return 0; |
| 1007 | } |
| 1008 | } else { |
| 1009 | return 1; |
| 1010 | } |
| 1011 | } |
| 1012 | |
| 1013 | EXPORT void speex_preprocess_estimate_update(SpeexPreprocessState *st, spx_int16_t *x) |
| 1014 | { |
| 1015 | int i; |
| 1016 | int N = st->ps_size; |
| 1017 | int N3 = 2*N - st->frame_size; |
| 1018 | int M; |
| 1019 | spx_word32_t *ps=st->ps; |
| 1020 | |
| 1021 | M = st->nbands; |
| 1022 | st->min_count++; |
| 1023 | |
| 1024 | preprocess_analysis(st, x); |
| 1025 | |
| 1026 | update_noise_prob(st); |
| 1027 | |
| 1028 | for (i=1;i<N-1;i++) |
| 1029 | { |
| 1030 | if (!st->update_prob[i] || st->ps[i] < PSHR32(st->noise[i],NOISE_SHIFT)) |
| 1031 | { |
| 1032 | st->noise[i] = MULT16_32_Q15(QCONST16(.95f,15),st->noise[i]) + MULT16_32_Q15(QCONST16(.05f,15),SHL32(st->ps[i],NOISE_SHIFT)); |
| 1033 | } |
| 1034 | } |
| 1035 | |
| 1036 | for (i=0;i<N3;i++) |
| 1037 | st->outbuf[i] = MULT16_16_Q15(x[st->frame_size-N3+i],st->window[st->frame_size+i]); |
| 1038 | |
| 1039 | /* Save old power spectrum */ |
| 1040 | for (i=0;i<N+M;i++) |
| 1041 | st->old_ps[i] = ps[i]; |
| 1042 | |
| 1043 | for (i=0;i<N;i++) |
| 1044 | st->reverb_estimate[i] = MULT16_32_Q15(st->reverb_decay, st->reverb_estimate[i]); |
| 1045 | } |
| 1046 | |
| 1047 | |
| 1048 | EXPORT int speex_preprocess_ctl(SpeexPreprocessState *state, int request, void *ptr) |
| 1049 | { |
| 1050 | int i; |
| 1051 | SpeexPreprocessState *st; |
| 1052 | st=(SpeexPreprocessState*)state; |
| 1053 | switch(request) |
| 1054 | { |
| 1055 | case SPEEX_PREPROCESS_SET_DENOISE: |
| 1056 | st->denoise_enabled = (*(spx_int32_t*)ptr); |
| 1057 | break; |
| 1058 | case SPEEX_PREPROCESS_GET_DENOISE: |
| 1059 | (*(spx_int32_t*)ptr) = st->denoise_enabled; |
| 1060 | break; |
| 1061 | #ifndef FIXED_POINT |
| 1062 | case SPEEX_PREPROCESS_SET_AGC: |
| 1063 | st->agc_enabled = (*(spx_int32_t*)ptr); |
| 1064 | break; |
| 1065 | case SPEEX_PREPROCESS_GET_AGC: |
| 1066 | (*(spx_int32_t*)ptr) = st->agc_enabled; |
| 1067 | break; |
| 1068 | #ifndef DISABLE_FLOAT_API |
| 1069 | case SPEEX_PREPROCESS_SET_AGC_LEVEL: |
| 1070 | st->agc_level = (*(float*)ptr); |
| 1071 | if (st->agc_level<1) |
| 1072 | st->agc_level=1; |
| 1073 | if (st->agc_level>32768) |
| 1074 | st->agc_level=32768; |
| 1075 | break; |
| 1076 | case SPEEX_PREPROCESS_GET_AGC_LEVEL: |
| 1077 | (*(float*)ptr) = st->agc_level; |
| 1078 | break; |
| 1079 | #endif /* #ifndef DISABLE_FLOAT_API */ |
| 1080 | case SPEEX_PREPROCESS_SET_AGC_INCREMENT: |
| 1081 | st->max_increase_step = exp(0.11513f * (*(spx_int32_t*)ptr)*st->frame_size / st->sampling_rate); |
| 1082 | break; |
| 1083 | case SPEEX_PREPROCESS_GET_AGC_INCREMENT: |
| 1084 | (*(spx_int32_t*)ptr) = floor(.5+8.6858*log(st->max_increase_step)*st->sampling_rate/st->frame_size); |
| 1085 | break; |
| 1086 | case SPEEX_PREPROCESS_SET_AGC_DECREMENT: |
| 1087 | st->max_decrease_step = exp(0.11513f * (*(spx_int32_t*)ptr)*st->frame_size / st->sampling_rate); |
| 1088 | break; |
| 1089 | case SPEEX_PREPROCESS_GET_AGC_DECREMENT: |
| 1090 | (*(spx_int32_t*)ptr) = floor(.5+8.6858*log(st->max_decrease_step)*st->sampling_rate/st->frame_size); |
| 1091 | break; |
| 1092 | case SPEEX_PREPROCESS_SET_AGC_MAX_GAIN: |
| 1093 | st->max_gain = exp(0.11513f * (*(spx_int32_t*)ptr)); |
| 1094 | break; |
| 1095 | case SPEEX_PREPROCESS_GET_AGC_MAX_GAIN: |
| 1096 | (*(spx_int32_t*)ptr) = floor(.5+8.6858*log(st->max_gain)); |
| 1097 | break; |
| 1098 | #endif |
| 1099 | case SPEEX_PREPROCESS_SET_VAD: |
| 1100 | speex_warning("The VAD has been replaced by a hack pending a complete rewrite"); |
| 1101 | st->vad_enabled = (*(spx_int32_t*)ptr); |
| 1102 | break; |
| 1103 | case SPEEX_PREPROCESS_GET_VAD: |
| 1104 | (*(spx_int32_t*)ptr) = st->vad_enabled; |
| 1105 | break; |
| 1106 | |
| 1107 | case SPEEX_PREPROCESS_SET_DEREVERB: |
| 1108 | st->dereverb_enabled = (*(spx_int32_t*)ptr); |
| 1109 | for (i=0;i<st->ps_size;i++) |
| 1110 | st->reverb_estimate[i]=0; |
| 1111 | break; |
| 1112 | case SPEEX_PREPROCESS_GET_DEREVERB: |
| 1113 | (*(spx_int32_t*)ptr) = st->dereverb_enabled; |
| 1114 | break; |
| 1115 | |
| 1116 | case SPEEX_PREPROCESS_SET_DEREVERB_LEVEL: |
| 1117 | /* FIXME: Re-enable when de-reverberation is actually enabled again */ |
| 1118 | /*st->reverb_level = (*(float*)ptr);*/ |
| 1119 | break; |
| 1120 | case SPEEX_PREPROCESS_GET_DEREVERB_LEVEL: |
| 1121 | /* FIXME: Re-enable when de-reverberation is actually enabled again */ |
| 1122 | /*(*(float*)ptr) = st->reverb_level;*/ |
| 1123 | break; |
| 1124 | |
| 1125 | case SPEEX_PREPROCESS_SET_DEREVERB_DECAY: |
| 1126 | /* FIXME: Re-enable when de-reverberation is actually enabled again */ |
| 1127 | /*st->reverb_decay = (*(float*)ptr);*/ |
| 1128 | break; |
| 1129 | case SPEEX_PREPROCESS_GET_DEREVERB_DECAY: |
| 1130 | /* FIXME: Re-enable when de-reverberation is actually enabled again */ |
| 1131 | /*(*(float*)ptr) = st->reverb_decay;*/ |
| 1132 | break; |
| 1133 | |
| 1134 | case SPEEX_PREPROCESS_SET_PROB_START: |
| 1135 | *(spx_int32_t*)ptr = MIN32(100,MAX32(0, *(spx_int32_t*)ptr)); |
| 1136 | st->speech_prob_start = DIV32_16(MULT16_16(Q15ONE,*(spx_int32_t*)ptr), 100); |
| 1137 | break; |
| 1138 | case SPEEX_PREPROCESS_GET_PROB_START: |
| 1139 | (*(spx_int32_t*)ptr) = MULT16_16_Q15(st->speech_prob_start, 100); |
| 1140 | break; |
| 1141 | |
| 1142 | case SPEEX_PREPROCESS_SET_PROB_CONTINUE: |
| 1143 | *(spx_int32_t*)ptr = MIN32(100,MAX32(0, *(spx_int32_t*)ptr)); |
| 1144 | st->speech_prob_continue = DIV32_16(MULT16_16(Q15ONE,*(spx_int32_t*)ptr), 100); |
| 1145 | break; |
| 1146 | case SPEEX_PREPROCESS_GET_PROB_CONTINUE: |
| 1147 | (*(spx_int32_t*)ptr) = MULT16_16_Q15(st->speech_prob_continue, 100); |
| 1148 | break; |
| 1149 | |
| 1150 | case SPEEX_PREPROCESS_SET_NOISE_SUPPRESS: |
| 1151 | st->noise_suppress = -ABS(*(spx_int32_t*)ptr); |
| 1152 | break; |
| 1153 | case SPEEX_PREPROCESS_GET_NOISE_SUPPRESS: |
| 1154 | (*(spx_int32_t*)ptr) = st->noise_suppress; |
| 1155 | break; |
| 1156 | case SPEEX_PREPROCESS_SET_ECHO_SUPPRESS: |
| 1157 | st->echo_suppress = -ABS(*(spx_int32_t*)ptr); |
| 1158 | break; |
| 1159 | case SPEEX_PREPROCESS_GET_ECHO_SUPPRESS: |
| 1160 | (*(spx_int32_t*)ptr) = st->echo_suppress; |
| 1161 | break; |
| 1162 | case SPEEX_PREPROCESS_SET_ECHO_SUPPRESS_ACTIVE: |
| 1163 | st->echo_suppress_active = -ABS(*(spx_int32_t*)ptr); |
| 1164 | break; |
| 1165 | case SPEEX_PREPROCESS_GET_ECHO_SUPPRESS_ACTIVE: |
| 1166 | (*(spx_int32_t*)ptr) = st->echo_suppress_active; |
| 1167 | break; |
| 1168 | case SPEEX_PREPROCESS_SET_ECHO_STATE: |
| 1169 | st->echo_state = (SpeexEchoState*)ptr; |
| 1170 | break; |
| 1171 | case SPEEX_PREPROCESS_GET_ECHO_STATE: |
| 1172 | (*(SpeexEchoState**)ptr) = (SpeexEchoState*)st->echo_state; |
| 1173 | break; |
| 1174 | #ifndef FIXED_POINT |
| 1175 | case SPEEX_PREPROCESS_GET_AGC_LOUDNESS: |
| 1176 | (*(spx_int32_t*)ptr) = pow(st->loudness, 1.0/LOUDNESS_EXP); |
| 1177 | break; |
| 1178 | case SPEEX_PREPROCESS_GET_AGC_GAIN: |
| 1179 | (*(spx_int32_t*)ptr) = floor(.5+8.6858*log(st->agc_gain)); |
| 1180 | break; |
| 1181 | #endif |
| 1182 | case SPEEX_PREPROCESS_GET_PSD_SIZE: |
| 1183 | case SPEEX_PREPROCESS_GET_NOISE_PSD_SIZE: |
| 1184 | (*(spx_int32_t*)ptr) = st->ps_size; |
| 1185 | break; |
| 1186 | case SPEEX_PREPROCESS_GET_PSD: |
| 1187 | for(i=0;i<st->ps_size;i++) |
| 1188 | ((spx_int32_t *)ptr)[i] = (spx_int32_t) st->ps[i]; |
| 1189 | break; |
| 1190 | case SPEEX_PREPROCESS_GET_NOISE_PSD: |
| 1191 | for(i=0;i<st->ps_size;i++) |
| 1192 | ((spx_int32_t *)ptr)[i] = (spx_int32_t) PSHR32(st->noise[i], NOISE_SHIFT); |
| 1193 | break; |
| 1194 | case SPEEX_PREPROCESS_GET_PROB: |
| 1195 | (*(spx_int32_t*)ptr) = MULT16_16_Q15(st->speech_prob, 100); |
| 1196 | break; |
| 1197 | #ifndef FIXED_POINT |
| 1198 | case SPEEX_PREPROCESS_SET_AGC_TARGET: |
| 1199 | st->agc_level = (*(spx_int32_t*)ptr); |
| 1200 | if (st->agc_level<1) |
| 1201 | st->agc_level=1; |
| 1202 | if (st->agc_level>32768) |
| 1203 | st->agc_level=32768; |
| 1204 | break; |
| 1205 | case SPEEX_PREPROCESS_GET_AGC_TARGET: |
| 1206 | (*(spx_int32_t*)ptr) = st->agc_level; |
| 1207 | break; |
| 1208 | #endif |
| 1209 | default: |
| 1210 | speex_warning_int("Unknown speex_preprocess_ctl request: ", request); |
| 1211 | return -1; |
| 1212 | } |
| 1213 | return 0; |
| 1214 | } |
| 1215 | |
| 1216 | #ifdef FIXED_DEBUG |
| 1217 | long long spx_mips=0; |
| 1218 | #endif |
| 1219 | |