diff options
| author | erdgeist <erdgeist@erdgeist.org> | 2025-08-15 12:42:40 +0200 |
|---|---|---|
| committer | erdgeist <erdgeist@erdgeist.org> | 2025-08-15 12:42:40 +0200 |
| commit | 30325d24d107dbf133da39f7c96d1510fd1c9449 (patch) | |
| tree | 932baa5b2a4475821f16dccf9e3e05011daa6d92 /quantise.c | |
| parent | 9022d768021bbe15c7815cc6f8b64218b46f0e10 (diff) | |
Bump to codec2 version 1.2.0erdgeist-bump-to-1.2.0
Diffstat (limited to 'quantise.c')
| -rw-r--r-- | quantise.c | 1992 |
1 files changed, 536 insertions, 1456 deletions
| @@ -24,26 +24,27 @@ | |||
| 24 | 24 | ||
| 25 | */ | 25 | */ |
| 26 | 26 | ||
| 27 | #include "quantise.h" | ||
| 28 | |||
| 27 | #include <assert.h> | 29 | #include <assert.h> |
| 28 | #include <ctype.h> | 30 | #include <ctype.h> |
| 31 | #include <math.h> | ||
| 29 | #include <stdio.h> | 32 | #include <stdio.h> |
| 30 | #include <stdlib.h> | 33 | #include <stdlib.h> |
| 31 | #include <string.h> | 34 | #include <string.h> |
| 32 | #include <math.h> | ||
| 33 | 35 | ||
| 36 | #include "codec2_fft.h" | ||
| 34 | #include "defines.h" | 37 | #include "defines.h" |
| 35 | #include "dump.h" | 38 | #include "dump.h" |
| 36 | #include "quantise.h" | ||
| 37 | #include "lpc.h" | 39 | #include "lpc.h" |
| 38 | #include "lsp.h" | 40 | #include "lsp.h" |
| 39 | #include "codec2_fft.h" | ||
| 40 | #include "phase.h" | ||
| 41 | #include "mbest.h" | 41 | #include "mbest.h" |
| 42 | #include "phase.h" | ||
| 42 | 43 | ||
| 43 | #undef PROFILE | 44 | #undef PROFILE |
| 44 | #include "machdep.h" | 45 | #include "machdep.h" |
| 45 | 46 | ||
| 46 | #define LSP_DELTA1 0.01 /* grid spacing for LSP root searches */ | 47 | #define LSP_DELTA1 0.01 /* grid spacing for LSP root searches */ |
| 47 | 48 | ||
| 48 | /*---------------------------------------------------------------------------*\ | 49 | /*---------------------------------------------------------------------------*\ |
| 49 | 50 | ||
| @@ -52,7 +53,7 @@ | |||
| 52 | \*---------------------------------------------------------------------------*/ | 53 | \*---------------------------------------------------------------------------*/ |
| 53 | 54 | ||
| 54 | float speech_to_uq_lsps(float lsp[], float ak[], float Sn[], float w[], | 55 | float speech_to_uq_lsps(float lsp[], float ak[], float Sn[], float w[], |
| 55 | int m_pitch, int order); | 56 | int m_pitch, int order); |
| 56 | 57 | ||
| 57 | /*---------------------------------------------------------------------------*\ | 58 | /*---------------------------------------------------------------------------*\ |
| 58 | 59 | ||
| @@ -60,46 +61,11 @@ float speech_to_uq_lsps(float lsp[], float ak[], float Sn[], float w[], | |||
| 60 | 61 | ||
| 61 | \*---------------------------------------------------------------------------*/ | 62 | \*---------------------------------------------------------------------------*/ |
| 62 | 63 | ||
| 63 | int lsp_bits(int i) { | 64 | int lsp_bits(int i) { return lsp_cb[i].log2m; } |
| 64 | return lsp_cb[i].log2m; | ||
| 65 | } | ||
| 66 | 65 | ||
| 67 | int lspd_bits(int i) { | 66 | int lspd_bits(int i) { return lsp_cbd[i].log2m; } |
| 68 | return lsp_cbd[i].log2m; | ||
| 69 | } | ||
| 70 | 67 | ||
| 71 | #ifndef CORTEX_M4 | 68 | int lsp_pred_vq_bits(int i) { return lsp_cbjmv[i].log2m; } |
| 72 | int mel_bits(int i) { | ||
| 73 | return mel_cb[i].log2m; | ||
| 74 | } | ||
| 75 | |||
| 76 | int lspmelvq_cb_bits(int i) { | ||
| 77 | return lspmelvq_cb[i].log2m; | ||
| 78 | } | ||
| 79 | #endif | ||
| 80 | |||
| 81 | #ifdef __EXPERIMENTAL__ | ||
| 82 | int lspdt_bits(int i) { | ||
| 83 | return lsp_cbdt[i].log2m; | ||
| 84 | } | ||
| 85 | #endif | ||
| 86 | |||
| 87 | int lsp_pred_vq_bits(int i) { | ||
| 88 | return lsp_cbjvm[i].log2m; | ||
| 89 | } | ||
| 90 | |||
| 91 | /*---------------------------------------------------------------------------*\ | ||
| 92 | |||
| 93 | quantise_init | ||
| 94 | |||
| 95 | Loads the entire LSP quantiser comprised of several vector quantisers | ||
| 96 | (codebooks). | ||
| 97 | |||
| 98 | \*---------------------------------------------------------------------------*/ | ||
| 99 | |||
| 100 | void quantise_init() | ||
| 101 | { | ||
| 102 | } | ||
| 103 | 69 | ||
| 104 | /*---------------------------------------------------------------------------*\ | 70 | /*---------------------------------------------------------------------------*\ |
| 105 | 71 | ||
| @@ -111,7 +77,7 @@ void quantise_init() | |||
| 111 | 77 | ||
| 112 | \*---------------------------------------------------------------------------*/ | 78 | \*---------------------------------------------------------------------------*/ |
| 113 | 79 | ||
| 114 | long quantise(const float * cb, float vec[], float w[], int k, int m, float *se) | 80 | long quantise(const float *cb, float vec[], float w[], int k, int m, float *se) |
| 115 | /* float cb[][K]; current VQ codebook */ | 81 | /* float cb[][K]; current VQ codebook */ |
| 116 | /* float vec[]; vector to quantise */ | 82 | /* float vec[]; vector to quantise */ |
| 117 | /* float w[]; weighting vector */ | 83 | /* float w[]; weighting vector */ |
| @@ -119,363 +85,123 @@ long quantise(const float * cb, float vec[], float w[], int k, int m, float *se) | |||
| 119 | /* int m; size of codebook */ | 85 | /* int m; size of codebook */ |
| 120 | /* float *se; accumulated squared error */ | 86 | /* float *se; accumulated squared error */ |
| 121 | { | 87 | { |
| 122 | float e; /* current error */ | 88 | float e; /* current error */ |
| 123 | long besti; /* best index so far */ | 89 | long besti; /* best index so far */ |
| 124 | float beste; /* best error so far */ | 90 | float beste; /* best error so far */ |
| 125 | long j; | 91 | long j; |
| 126 | int i; | 92 | int i; |
| 127 | float diff; | 93 | float diff; |
| 128 | |||
| 129 | besti = 0; | ||
| 130 | beste = 1E32; | ||
| 131 | for(j=0; j<m; j++) { | ||
| 132 | e = 0.0; | ||
| 133 | for(i=0; i<k; i++) { | ||
| 134 | diff = cb[j*k+i]-vec[i]; | ||
| 135 | e += (diff*w[i] * diff*w[i]); | ||
| 136 | } | ||
| 137 | if (e < beste) { | ||
| 138 | beste = e; | ||
| 139 | besti = j; | ||
| 140 | } | ||
| 141 | } | ||
| 142 | |||
| 143 | *se += beste; | ||
| 144 | |||
| 145 | return(besti); | ||
| 146 | } | ||
| 147 | |||
| 148 | |||
| 149 | |||
| 150 | /*---------------------------------------------------------------------------*\ | ||
| 151 | |||
| 152 | encode_lspds_scalar() | ||
| 153 | |||
| 154 | Scalar/VQ LSP difference quantiser. | ||
| 155 | |||
| 156 | \*---------------------------------------------------------------------------*/ | ||
| 157 | 94 | ||
| 158 | void encode_lspds_scalar( | 95 | besti = 0; |
| 159 | int indexes[], | 96 | beste = 1E32; |
| 160 | float lsp[], | 97 | for (j = 0; j < m; j++) { |
| 161 | int order | 98 | e = 0.0; |
| 162 | ) | 99 | for (i = 0; i < k; i++) { |
| 163 | { | 100 | diff = cb[j * k + i] - vec[i]; |
| 164 | int i,k,m; | 101 | e += (diff * w[i] * diff * w[i]); |
| 165 | float lsp_hz[order]; | ||
| 166 | float lsp__hz[order]; | ||
| 167 | float dlsp[order]; | ||
| 168 | float dlsp_[order]; | ||
| 169 | float wt[order]; | ||
| 170 | const float *cb; | ||
| 171 | float se; | ||
| 172 | |||
| 173 | for(i=0; i<order; i++) { | ||
| 174 | wt[i] = 1.0; | ||
| 175 | } | 102 | } |
| 176 | 103 | if (e < beste) { | |
| 177 | /* convert from radians to Hz so we can use human readable | 104 | beste = e; |
| 178 | frequencies */ | 105 | besti = j; |
| 179 | |||
| 180 | for(i=0; i<order; i++) | ||
| 181 | lsp_hz[i] = (4000.0/PI)*lsp[i]; | ||
| 182 | |||
| 183 | //printf("\n"); | ||
| 184 | |||
| 185 | wt[0] = 1.0; | ||
| 186 | for(i=0; i<order; i++) { | ||
| 187 | |||
| 188 | /* find difference from previous qunatised lsp */ | ||
| 189 | |||
| 190 | if (i) | ||
| 191 | dlsp[i] = lsp_hz[i] - lsp__hz[i-1]; | ||
| 192 | else | ||
| 193 | dlsp[0] = lsp_hz[0]; | ||
| 194 | |||
| 195 | k = lsp_cbd[i].k; | ||
| 196 | m = lsp_cbd[i].m; | ||
| 197 | cb = lsp_cbd[i].cb; | ||
| 198 | indexes[i] = quantise(cb, &dlsp[i], wt, k, m, &se); | ||
| 199 | dlsp_[i] = cb[indexes[i]*k]; | ||
| 200 | |||
| 201 | |||
| 202 | if (i) | ||
| 203 | lsp__hz[i] = lsp__hz[i-1] + dlsp_[i]; | ||
| 204 | else | ||
| 205 | lsp__hz[0] = dlsp_[0]; | ||
| 206 | |||
| 207 | //printf("%d lsp %3.2f dlsp %3.2f dlsp_ %3.2f lsp_ %3.2f\n", i, lsp_hz[i], dlsp[i], dlsp_[i], lsp__hz[i]); | ||
| 208 | } | 106 | } |
| 107 | } | ||
| 209 | 108 | ||
| 210 | } | 109 | *se += beste; |
| 211 | |||
| 212 | |||
| 213 | void decode_lspds_scalar( | ||
| 214 | float lsp_[], | ||
| 215 | int indexes[], | ||
| 216 | int order | ||
| 217 | ) | ||
| 218 | { | ||
| 219 | int i,k; | ||
| 220 | float lsp__hz[order]; | ||
| 221 | float dlsp_[order]; | ||
| 222 | const float *cb; | ||
| 223 | |||
| 224 | for(i=0; i<order; i++) { | ||
| 225 | |||
| 226 | k = lsp_cbd[i].k; | ||
| 227 | cb = lsp_cbd[i].cb; | ||
| 228 | dlsp_[i] = cb[indexes[i]*k]; | ||
| 229 | |||
| 230 | if (i) | ||
| 231 | lsp__hz[i] = lsp__hz[i-1] + dlsp_[i]; | ||
| 232 | else | ||
| 233 | lsp__hz[0] = dlsp_[0]; | ||
| 234 | |||
| 235 | lsp_[i] = (PI/4000.0)*lsp__hz[i]; | ||
| 236 | |||
| 237 | //printf("%d dlsp_ %3.2f lsp_ %3.2f\n", i, dlsp_[i], lsp__hz[i]); | ||
| 238 | } | ||
| 239 | 110 | ||
| 111 | return (besti); | ||
| 240 | } | 112 | } |
| 241 | 113 | ||
| 242 | #ifdef __EXPERIMENTAL__ | ||
| 243 | /*---------------------------------------------------------------------------*\ | 114 | /*---------------------------------------------------------------------------*\ |
| 244 | 115 | ||
| 245 | lspvq_quantise | 116 | encode_lspds_scalar() |
| 246 | 117 | ||
| 247 | Vector LSP quantiser. | 118 | Scalar/VQ LSP difference-in-frequency quantiser. |
| 248 | 119 | ||
| 249 | \*---------------------------------------------------------------------------*/ | 120 | \*---------------------------------------------------------------------------*/ |
| 250 | 121 | ||
| 251 | void lspvq_quantise( | 122 | void encode_lspds_scalar(int indexes[], float lsp[], int order) { |
| 252 | float lsp[], | 123 | int i, k, m; |
| 253 | float lsp_[], | 124 | float lsp_hz[order]; |
| 254 | int order | 125 | float lsp__hz[order]; |
| 255 | ) | 126 | float dlsp[order]; |
| 256 | { | 127 | float dlsp_[order]; |
| 257 | int i,k,m,ncb, nlsp; | 128 | float wt[order]; |
| 258 | float wt[order], lsp_hz[order]; | 129 | const float *cb; |
| 259 | const float *cb; | 130 | float se; |
| 260 | float se; | 131 | |
| 261 | int index; | 132 | for (i = 0; i < order; i++) { |
| 262 | 133 | wt[i] = 1.0; | |
| 263 | for(i=0; i<order; i++) { | 134 | } |
| 264 | wt[i] = 1.0; | ||
| 265 | lsp_hz[i] = 4000.0*lsp[i]/PI; | ||
| 266 | } | ||
| 267 | |||
| 268 | /* scalar quantise LSPs 1,2,3,4 */ | ||
| 269 | |||
| 270 | /* simple uniform scalar quantisers */ | ||
| 271 | |||
| 272 | for(i=0; i<4; i++) { | ||
| 273 | k = lsp_cb[i].k; | ||
| 274 | m = lsp_cb[i].m; | ||
| 275 | cb = lsp_cb[i].cb; | ||
| 276 | index = quantise(cb, &lsp_hz[i], wt, k, m, &se); | ||
| 277 | lsp_[i] = cb[index*k]*PI/4000.0; | ||
| 278 | } | ||
| 279 | |||
| 280 | //#define WGHT | ||
| 281 | #ifdef WGHT | ||
| 282 | for(i=4; i<9; i++) { | ||
| 283 | wt[i] = 1.0/(lsp[i]-lsp[i-1]) + 1.0/(lsp[i+1]-lsp[i]); | ||
| 284 | //printf("wt[%d] = %f\n", i, wt[i]); | ||
| 285 | } | ||
| 286 | wt[9] = 1.0/(lsp[i]-lsp[i-1]); | ||
| 287 | #endif | ||
| 288 | |||
| 289 | /* VQ LSPs 5,6,7,8,9,10 */ | ||
| 290 | |||
| 291 | ncb = 4; | ||
| 292 | nlsp = 4; | ||
| 293 | k = lsp_cbjnd[ncb].k; | ||
| 294 | m = lsp_cbjnd[ncb].m; | ||
| 295 | cb = lsp_cbjnd[ncb].cb; | ||
| 296 | index = quantise(cb, &lsp_hz[nlsp], &wt[nlsp], k, m, &se); | ||
| 297 | for(i=4; i<order; i++) { | ||
| 298 | lsp_[i] = cb[index*k+i-4]*(PI/4000.0); | ||
| 299 | //printf("%4.f (%4.f) ", lsp_hz[i], cb[index*k+i-4]); | ||
| 300 | } | ||
| 301 | } | ||
| 302 | |||
| 303 | /*---------------------------------------------------------------------------*\ | ||
| 304 | |||
| 305 | lspjnd_quantise | ||
| 306 | 135 | ||
| 307 | Experimental JND LSP quantiser. | 136 | /* convert from radians to Hz so we can use human readable |
| 137 | frequencies */ | ||
| 308 | 138 | ||
| 309 | \*---------------------------------------------------------------------------*/ | 139 | for (i = 0; i < order; i++) lsp_hz[i] = (4000.0 / PI) * lsp[i]; |
| 310 | 140 | ||
| 311 | void lspjnd_quantise(float lsps[], float lsps_[], int order) | 141 | wt[0] = 1.0; |
| 312 | { | 142 | for (i = 0; i < order; i++) { |
| 313 | int i,k,m; | 143 | /* find difference from previous quantised lsp */ |
| 314 | float wt[order], lsps_hz[order]; | ||
| 315 | const float *cb; | ||
| 316 | float se = 0.0; | ||
| 317 | int index; | ||
| 318 | |||
| 319 | for(i=0; i<order; i++) { | ||
| 320 | wt[i] = 1.0; | ||
| 321 | } | ||
| 322 | |||
| 323 | /* convert to Hz */ | ||
| 324 | 144 | ||
| 325 | for(i=0; i<order; i++) { | 145 | if (i) |
| 326 | lsps_hz[i] = lsps[i]*(4000.0/PI); | 146 | dlsp[i] = lsp_hz[i] - lsp__hz[i - 1]; |
| 327 | lsps_[i] = lsps[i]; | 147 | else |
| 328 | } | 148 | dlsp[0] = lsp_hz[0]; |
| 329 | 149 | ||
| 330 | /* simple uniform scalar quantisers */ | 150 | k = lsp_cbd[i].k; |
| 151 | m = lsp_cbd[i].m; | ||
| 152 | cb = lsp_cbd[i].cb; | ||
| 153 | indexes[i] = quantise(cb, &dlsp[i], wt, k, m, &se); | ||
| 154 | dlsp_[i] = cb[indexes[i] * k]; | ||
| 331 | 155 | ||
| 332 | for(i=0; i<4; i++) { | 156 | if (i) |
| 333 | k = lsp_cbjnd[i].k; | 157 | lsp__hz[i] = lsp__hz[i - 1] + dlsp_[i]; |
| 334 | m = lsp_cbjnd[i].m; | 158 | else |
| 335 | cb = lsp_cbjnd[i].cb; | 159 | lsp__hz[0] = dlsp_[0]; |
| 336 | index = quantise(cb, &lsps_hz[i], wt, k, m, &se); | 160 | } |
| 337 | lsps_[i] = cb[index*k]*(PI/4000.0); | ||
| 338 | } | ||
| 339 | |||
| 340 | /* VQ LSPs 5,6,7,8,9,10 */ | ||
| 341 | |||
| 342 | k = lsp_cbjnd[4].k; | ||
| 343 | m = lsp_cbjnd[4].m; | ||
| 344 | cb = lsp_cbjnd[4].cb; | ||
| 345 | index = quantise(cb, &lsps_hz[4], &wt[4], k, m, &se); | ||
| 346 | //printf("k = %d m = %d c[0] %f cb[k] %f\n", k,m,cb[0],cb[k]); | ||
| 347 | //printf("index = %4d: ", index); | ||
| 348 | for(i=4; i<order; i++) { | ||
| 349 | lsps_[i] = cb[index*k+i-4]*(PI/4000.0); | ||
| 350 | //printf("%4.f (%4.f) ", lsps_hz[i], cb[index*k+i-4]); | ||
| 351 | } | ||
| 352 | //printf("\n"); | ||
| 353 | } | 161 | } |
| 354 | 162 | ||
| 355 | void compute_weights(const float *x, float *w, int ndim); | 163 | void decode_lspds_scalar(float lsp_[], int indexes[], int order) { |
| 356 | 164 | int i, k; | |
| 357 | /*---------------------------------------------------------------------------*\ | 165 | float lsp__hz[order]; |
| 358 | 166 | float dlsp_[order]; | |
| 359 | lspdt_quantise | 167 | const float *cb; |
| 360 | |||
| 361 | LSP difference in time quantiser. Split VQ, encoding LSPs 1-4 with | ||
| 362 | one VQ, and LSPs 5-10 with a second. Update of previous lsp memory | ||
| 363 | is done outside of this function to handle dT between 10 or 20ms | ||
| 364 | frames. | ||
| 365 | |||
| 366 | mode action | ||
| 367 | ------------------ | ||
| 368 | |||
| 369 | LSPDT_ALL VQ LSPs 1-4 and 5-10 | ||
| 370 | LSPDT_LOW Just VQ LSPs 1-4, for LSPs 5-10 just copy previous | ||
| 371 | LSPDT_HIGH Just VQ LSPs 5-10, for LSPs 1-4 just copy previous | ||
| 372 | |||
| 373 | \*---------------------------------------------------------------------------*/ | ||
| 374 | |||
| 375 | void lspdt_quantise(float lsps[], float lsps_[], float lsps__prev[], int mode) | ||
| 376 | { | ||
| 377 | int i; | ||
| 378 | float wt[LPC_ORD]; | ||
| 379 | float lsps_dt[LPC_ORD]; | ||
| 380 | #ifdef TRY_LSPDT_VQ | ||
| 381 | int k,m; | ||
| 382 | int index; | ||
| 383 | const float *cb; | ||
| 384 | float se = 0.0; | ||
| 385 | #endif // TRY_LSPDT_VQ | ||
| 386 | |||
| 387 | //compute_weights(lsps, wt, LPC_ORD); | ||
| 388 | for(i=0; i<LPC_ORD; i++) { | ||
| 389 | wt[i] = 1.0; | ||
| 390 | } | ||
| 391 | |||
| 392 | //compute_weights(lsps, wt, LPC_ORD ); | ||
| 393 | 168 | ||
| 394 | for(i=0; i<LPC_ORD; i++) { | 169 | for (i = 0; i < order; i++) { |
| 395 | lsps_dt[i] = lsps[i] - lsps__prev[i]; | 170 | k = lsp_cbd[i].k; |
| 396 | lsps_[i] = lsps__prev[i]; | 171 | cb = lsp_cbd[i].cb; |
| 397 | } | 172 | dlsp_[i] = cb[indexes[i] * k]; |
| 398 | 173 | ||
| 399 | //#define TRY_LSPDT_VQ | 174 | if (i) |
| 400 | #ifdef TRY_LSPDT_VQ | 175 | lsp__hz[i] = lsp__hz[i - 1] + dlsp_[i]; |
| 401 | /* this actually improves speech a bit, but 40ms updates works surprsingly well.... */ | 176 | else |
| 402 | k = lsp_cbdt[0].k; | 177 | lsp__hz[0] = dlsp_[0]; |
| 403 | m = lsp_cbdt[0].m; | ||
| 404 | cb = lsp_cbdt[0].cb; | ||
| 405 | index = quantise(cb, lsps_dt, wt, k, m, &se); | ||
| 406 | for(i=0; i<LPC_ORD; i++) { | ||
| 407 | lsps_[i] += cb[index*k + i]; | ||
| 408 | } | ||
| 409 | #endif | ||
| 410 | 178 | ||
| 179 | lsp_[i] = (PI / 4000.0) * lsp__hz[i]; | ||
| 180 | } | ||
| 411 | } | 181 | } |
| 412 | #endif | ||
| 413 | 182 | ||
| 414 | #define MIN(a,b) ((a)<(b)?(a):(b)) | 183 | #define MIN(a, b) ((a) < (b) ? (a) : (b)) |
| 415 | #define MAX_ENTRIES 16384 | 184 | #define MAX_ENTRIES 16384 |
| 416 | 185 | ||
| 417 | void compute_weights(const float *x, float *w, int ndim) | 186 | void compute_weights(const float *x, float *w, int ndim) { |
| 418 | { | ||
| 419 | int i; | 187 | int i; |
| 420 | w[0] = MIN(x[0], x[1]-x[0]); | 188 | w[0] = MIN(x[0], x[1] - x[0]); |
| 421 | for (i=1;i<ndim-1;i++) | 189 | for (i = 1; i < ndim - 1; i++) w[i] = MIN(x[i] - x[i - 1], x[i + 1] - x[i]); |
| 422 | w[i] = MIN(x[i]-x[i-1], x[i+1]-x[i]); | 190 | w[ndim - 1] = MIN(x[ndim - 1] - x[ndim - 2], PI - x[ndim - 1]); |
| 423 | w[ndim-1] = MIN(x[ndim-1]-x[ndim-2], PI-x[ndim-1]); | ||
| 424 | |||
| 425 | for (i=0;i<ndim;i++) | ||
| 426 | w[i] = 1./(.01+w[i]); | ||
| 427 | //w[0]*=3; | ||
| 428 | //w[1]*=2; | ||
| 429 | } | ||
| 430 | |||
| 431 | #ifdef __EXPERIMENTAL__ | ||
| 432 | /* LSP weight calculation ported from m-file function kindly submitted | ||
| 433 | by Anssi, OH3GDD */ | ||
| 434 | 191 | ||
| 435 | void compute_weights_anssi_mode2(const float *x, float *w, int ndim) | 192 | for (i = 0; i < ndim; i++) w[i] = 1. / (.01 + w[i]); |
| 436 | { | ||
| 437 | int i; | ||
| 438 | float d[LPC_ORD]; | ||
| 439 | |||
| 440 | assert(ndim == LPC_ORD); | ||
| 441 | |||
| 442 | for(i=0; i<LPC_ORD; i++) | ||
| 443 | d[i] = 1.0; | ||
| 444 | |||
| 445 | d[0] = x[1]; | ||
| 446 | for (i=1; i<LPC_ORD-1; i++) | ||
| 447 | d[i] = x[i+1] - x[i-1]; | ||
| 448 | d[LPC_ORD-1] = PI - x[8]; | ||
| 449 | for (i=0; i<LPC_ORD; i++) { | ||
| 450 | if (x[i]<((400.0/4000.0)*PI)) | ||
| 451 | w[i]=5.0/(0.01+d[i]); | ||
| 452 | else if (x[i]<((700.0/4000.0)*PI)) | ||
| 453 | w[i]=4.0/(0.01+d[i]); | ||
| 454 | else if (x[i]<((1200.0/4000.0)*PI)) | ||
| 455 | w[i]=3.0/(0.01+d[i]); | ||
| 456 | else if (x[i]<((2000.0/4000.0)*PI)) | ||
| 457 | w[i]=2.0/(0.01+d[i]); | ||
| 458 | else | ||
| 459 | w[i]=1.0/(0.01+d[i]); | ||
| 460 | |||
| 461 | w[i]=powf(w[i]+0.3, 0.66); | ||
| 462 | } | ||
| 463 | } | 193 | } |
| 464 | #endif | ||
| 465 | 194 | ||
| 466 | int find_nearest(const float *codebook, int nb_entries, float *x, int ndim) | 195 | int find_nearest(const float *codebook, int nb_entries, float *x, int ndim) { |
| 467 | { | ||
| 468 | int i, j; | 196 | int i, j; |
| 469 | float min_dist = 1e15; | 197 | float min_dist = 1e15; |
| 470 | int nearest = 0; | 198 | int nearest = 0; |
| 471 | 199 | ||
| 472 | for (i=0;i<nb_entries;i++) | 200 | for (i = 0; i < nb_entries; i++) { |
| 473 | { | 201 | float dist = 0; |
| 474 | float dist=0; | 202 | for (j = 0; j < ndim; j++) |
| 475 | for (j=0;j<ndim;j++) | 203 | dist += (x[j] - codebook[i * ndim + j]) * (x[j] - codebook[i * ndim + j]); |
| 476 | dist += (x[j]-codebook[i*ndim+j])*(x[j]-codebook[i*ndim+j]); | 204 | if (dist < min_dist) { |
| 477 | if (dist<min_dist) | ||
| 478 | { | ||
| 479 | min_dist = dist; | 205 | min_dist = dist; |
| 480 | nearest = i; | 206 | nearest = i; |
| 481 | } | 207 | } |
| @@ -483,19 +209,18 @@ int find_nearest(const float *codebook, int nb_entries, float *x, int ndim) | |||
| 483 | return nearest; | 209 | return nearest; |
| 484 | } | 210 | } |
| 485 | 211 | ||
| 486 | int find_nearest_weighted(const float *codebook, int nb_entries, float *x, const float *w, int ndim) | 212 | int find_nearest_weighted(const float *codebook, int nb_entries, float *x, |
| 487 | { | 213 | const float *w, int ndim) { |
| 488 | int i, j; | 214 | int i, j; |
| 489 | float min_dist = 1e15; | 215 | float min_dist = 1e15; |
| 490 | int nearest = 0; | 216 | int nearest = 0; |
| 491 | 217 | ||
| 492 | for (i=0;i<nb_entries;i++) | 218 | for (i = 0; i < nb_entries; i++) { |
| 493 | { | 219 | float dist = 0; |
| 494 | float dist=0; | 220 | for (j = 0; j < ndim; j++) |
| 495 | for (j=0;j<ndim;j++) | 221 | dist += w[j] * (x[j] - codebook[i * ndim + j]) * |
| 496 | dist += w[j]*(x[j]-codebook[i*ndim+j])*(x[j]-codebook[i*ndim+j]); | 222 | (x[j] - codebook[i * ndim + j]); |
| 497 | if (dist<min_dist) | 223 | if (dist < min_dist) { |
| 498 | { | ||
| 499 | min_dist = dist; | 224 | min_dist = dist; |
| 500 | nearest = i; | 225 | nearest = i; |
| 501 | } | 226 | } |
| @@ -503,212 +228,74 @@ int find_nearest_weighted(const float *codebook, int nb_entries, float *x, const | |||
| 503 | return nearest; | 228 | return nearest; |
| 504 | } | 229 | } |
| 505 | 230 | ||
| 506 | void lspjvm_quantise(float *x, float *xq, int order) | 231 | void lspjmv_quantise(float *x, float *xq, int order) { |
| 507 | { | ||
| 508 | int i, n1, n2, n3; | 232 | int i, n1, n2, n3; |
| 509 | float err[order], err2[order], err3[order]; | 233 | float err[order], err2[order], err3[order]; |
| 510 | float w[order], w2[order], w3[order]; | 234 | float w[order], w2[order], w3[order]; |
| 511 | const float *codebook1 = lsp_cbjvm[0].cb; | 235 | const float *codebook1 = lsp_cbjmv[0].cb; |
| 512 | const float *codebook2 = lsp_cbjvm[1].cb; | 236 | const float *codebook2 = lsp_cbjmv[1].cb; |
| 513 | const float *codebook3 = lsp_cbjvm[2].cb; | 237 | const float *codebook3 = lsp_cbjmv[2].cb; |
| 514 | 238 | ||
| 515 | w[0] = MIN(x[0], x[1]-x[0]); | 239 | w[0] = MIN(x[0], x[1] - x[0]); |
| 516 | for (i=1;i<order-1;i++) | 240 | for (i = 1; i < order - 1; i++) w[i] = MIN(x[i] - x[i - 1], x[i + 1] - x[i]); |
| 517 | w[i] = MIN(x[i]-x[i-1], x[i+1]-x[i]); | 241 | w[order - 1] = MIN(x[order - 1] - x[order - 2], PI - x[order - 1]); |
| 518 | w[order-1] = MIN(x[order-1]-x[order-2], PI-x[order-1]); | ||
| 519 | 242 | ||
| 520 | compute_weights(x, w, order); | 243 | compute_weights(x, w, order); |
| 521 | 244 | ||
| 522 | n1 = find_nearest(codebook1, lsp_cbjvm[0].m, x, order); | 245 | n1 = find_nearest(codebook1, lsp_cbjmv[0].m, x, order); |
| 523 | 246 | ||
| 524 | for (i=0;i<order;i++) | 247 | for (i = 0; i < order; i++) { |
| 525 | { | 248 | xq[i] = codebook1[order * n1 + i]; |
| 526 | xq[i] = codebook1[order*n1+i]; | ||
| 527 | err[i] = x[i] - xq[i]; | 249 | err[i] = x[i] - xq[i]; |
| 528 | } | 250 | } |
| 529 | for (i=0;i<order/2;i++) | 251 | for (i = 0; i < order / 2; i++) { |
| 530 | { | 252 | err2[i] = err[2 * i]; |
| 531 | err2[i] = err[2*i]; | 253 | err3[i] = err[2 * i + 1]; |
| 532 | err3[i] = err[2*i+1]; | 254 | w2[i] = w[2 * i]; |
| 533 | w2[i] = w[2*i]; | 255 | w3[i] = w[2 * i + 1]; |
| 534 | w3[i] = w[2*i+1]; | ||
| 535 | } | 256 | } |
| 536 | n2 = find_nearest_weighted(codebook2, lsp_cbjvm[1].m, err2, w2, order/2); | 257 | n2 = find_nearest_weighted(codebook2, lsp_cbjmv[1].m, err2, w2, order / 2); |
| 537 | n3 = find_nearest_weighted(codebook3, lsp_cbjvm[2].m, err3, w3, order/2); | 258 | n3 = find_nearest_weighted(codebook3, lsp_cbjmv[2].m, err3, w3, order / 2); |
| 538 | 259 | ||
| 539 | for (i=0;i<order/2;i++) | 260 | for (i = 0; i < order / 2; i++) { |
| 540 | { | 261 | xq[2 * i] += codebook2[order * n2 / 2 + i]; |
| 541 | xq[2*i] += codebook2[order*n2/2+i]; | 262 | xq[2 * i + 1] += codebook3[order * n3 / 2 + i]; |
| 542 | xq[2*i+1] += codebook3[order*n3/2+i]; | ||
| 543 | } | 263 | } |
| 544 | } | 264 | } |
| 545 | 265 | ||
| 266 | int check_lsp_order(float lsp[], int order) { | ||
| 267 | int i; | ||
| 268 | float tmp; | ||
| 269 | int swaps = 0; | ||
| 270 | |||
| 271 | for (i = 1; i < order; i++) | ||
| 272 | if (lsp[i] < lsp[i - 1]) { | ||
| 273 | // fprintf(stderr, "swap %d\n",i); | ||
| 274 | swaps++; | ||
| 275 | tmp = lsp[i - 1]; | ||
| 276 | lsp[i - 1] = lsp[i] - 0.1; | ||
| 277 | lsp[i] = tmp + 0.1; | ||
| 278 | i = 1; /* start check again, as swap may have caused out of order */ | ||
| 279 | } | ||
| 546 | 280 | ||
| 547 | #ifndef CORTEX_M4 | 281 | return swaps; |
| 548 | /* simple (non mbest) 6th order LSP MEL VQ quantiser. Returns MSE of result */ | ||
| 549 | |||
| 550 | float lspmelvq_quantise(float *x, float *xq, int order) | ||
| 551 | { | ||
| 552 | int i, n1, n2, n3; | ||
| 553 | float err[order]; | ||
| 554 | const float *codebook1 = lspmelvq_cb[0].cb; | ||
| 555 | const float *codebook2 = lspmelvq_cb[1].cb; | ||
| 556 | const float *codebook3 = lspmelvq_cb[2].cb; | ||
| 557 | float tmp[order]; | ||
| 558 | float mse; | ||
| 559 | |||
| 560 | assert(order == lspmelvq_cb[0].k); | ||
| 561 | |||
| 562 | n1 = find_nearest(codebook1, lspmelvq_cb[0].m, x, order); | ||
| 563 | |||
| 564 | for (i=0; i<order; i++) { | ||
| 565 | tmp[i] = codebook1[order*n1+i]; | ||
| 566 | err[i] = x[i] - tmp[i]; | ||
| 567 | } | ||
| 568 | |||
| 569 | n2 = find_nearest(codebook2, lspmelvq_cb[1].m, err, order); | ||
| 570 | |||
| 571 | for (i=0; i<order; i++) { | ||
| 572 | tmp[i] += codebook2[order*n2+i]; | ||
| 573 | err[i] = x[i] - tmp[i]; | ||
| 574 | } | ||
| 575 | |||
| 576 | n3 = find_nearest(codebook3, lspmelvq_cb[2].m, err, order); | ||
| 577 | |||
| 578 | mse = 0.0; | ||
| 579 | for (i=0; i<order; i++) { | ||
| 580 | tmp[i] += codebook3[order*n3+i]; | ||
| 581 | err[i] = x[i] - tmp[i]; | ||
| 582 | mse += err[i]*err[i]; | ||
| 583 | } | ||
| 584 | |||
| 585 | for (i=0; i<order; i++) { | ||
| 586 | xq[i] = tmp[i]; | ||
| 587 | } | ||
| 588 | |||
| 589 | return mse; | ||
| 590 | } | ||
| 591 | |||
| 592 | /* 3 stage VQ LSP quantiser useing mbest search. Design and guidance kindly submitted by Anssi, OH3GDD */ | ||
| 593 | |||
| 594 | float lspmelvq_mbest_encode(int *indexes, float *x, float *xq, int ndim, int mbest_entries) | ||
| 595 | { | ||
| 596 | int i, j, n1, n2, n3; | ||
| 597 | const float *codebook1 = lspmelvq_cb[0].cb; | ||
| 598 | const float *codebook2 = lspmelvq_cb[1].cb; | ||
| 599 | const float *codebook3 = lspmelvq_cb[2].cb; | ||
| 600 | struct MBEST *mbest_stage1, *mbest_stage2, *mbest_stage3; | ||
| 601 | float target[ndim]; | ||
| 602 | float w[ndim]; | ||
| 603 | int index[MBEST_STAGES]; | ||
| 604 | float mse, tmp; | ||
| 605 | |||
| 606 | for(i=0; i<ndim; i++) | ||
| 607 | w[i] = 1.0; | ||
| 608 | |||
| 609 | mbest_stage1 = mbest_create(mbest_entries); | ||
| 610 | mbest_stage2 = mbest_create(mbest_entries); | ||
| 611 | mbest_stage3 = mbest_create(mbest_entries); | ||
| 612 | for(i=0; i<MBEST_STAGES; i++) | ||
| 613 | index[i] = 0; | ||
| 614 | |||
| 615 | /* Stage 1 */ | ||
| 616 | |||
| 617 | mbest_search(codebook1, x, w, ndim, lspmelvq_cb[0].m, mbest_stage1, index); | ||
| 618 | MBEST_PRINT("Stage 1:", mbest_stage1); | ||
| 619 | |||
| 620 | /* Stage 2 */ | ||
| 621 | |||
| 622 | for (j=0; j<mbest_entries; j++) { | ||
| 623 | index[1] = n1 = mbest_stage1->list[j].index[0]; | ||
| 624 | for(i=0; i<ndim; i++) | ||
| 625 | target[i] = x[i] - codebook1[ndim*n1+i]; | ||
| 626 | mbest_search(codebook2, target, w, ndim, lspmelvq_cb[1].m, mbest_stage2, index); | ||
| 627 | } | ||
| 628 | MBEST_PRINT("Stage 2:", mbest_stage2); | ||
| 629 | |||
| 630 | /* Stage 3 */ | ||
| 631 | |||
| 632 | for (j=0; j<mbest_entries; j++) { | ||
| 633 | index[2] = n1 = mbest_stage2->list[j].index[1]; | ||
| 634 | index[1] = n2 = mbest_stage2->list[j].index[0]; | ||
| 635 | for(i=0; i<ndim; i++) | ||
| 636 | target[i] = x[i] - codebook1[ndim*n1+i] - codebook2[ndim*n2+i]; | ||
| 637 | mbest_search(codebook3, target, w, ndim, lspmelvq_cb[2].m, mbest_stage3, index); | ||
| 638 | } | ||
| 639 | MBEST_PRINT("Stage 3:", mbest_stage3); | ||
| 640 | |||
| 641 | n1 = mbest_stage3->list[0].index[2]; | ||
| 642 | n2 = mbest_stage3->list[0].index[1]; | ||
| 643 | n3 = mbest_stage3->list[0].index[0]; | ||
| 644 | mse = 0.0; | ||
| 645 | for (i=0;i<ndim;i++) { | ||
| 646 | tmp = codebook1[ndim*n1+i] + codebook2[ndim*n2+i] + codebook3[ndim*n3+i]; | ||
| 647 | mse += (x[i]-tmp)*(x[i]-tmp); | ||
| 648 | xq[i] = tmp; | ||
| 649 | } | ||
| 650 | |||
| 651 | mbest_destroy(mbest_stage1); | ||
| 652 | mbest_destroy(mbest_stage2); | ||
| 653 | mbest_destroy(mbest_stage3); | ||
| 654 | |||
| 655 | indexes[0] = n1; indexes[1] = n2; indexes[2] = n3; | ||
| 656 | |||
| 657 | return mse; | ||
| 658 | } | ||
| 659 | |||
| 660 | |||
| 661 | void lspmelvq_decode(int *indexes, float *xq, int ndim) | ||
| 662 | { | ||
| 663 | int i, n1, n2, n3; | ||
| 664 | const float *codebook1 = lspmelvq_cb[0].cb; | ||
| 665 | const float *codebook2 = lspmelvq_cb[1].cb; | ||
| 666 | const float *codebook3 = lspmelvq_cb[2].cb; | ||
| 667 | |||
| 668 | n1 = indexes[0]; n2 = indexes[1]; n3 = indexes[2]; | ||
| 669 | for (i=0;i<ndim;i++) { | ||
| 670 | xq[i] = codebook1[ndim*n1+i] + codebook2[ndim*n2+i] + codebook3[ndim*n3+i]; | ||
| 671 | } | ||
| 672 | } | ||
| 673 | #endif | ||
| 674 | |||
| 675 | |||
| 676 | int check_lsp_order(float lsp[], int order) | ||
| 677 | { | ||
| 678 | int i; | ||
| 679 | float tmp; | ||
| 680 | int swaps = 0; | ||
| 681 | |||
| 682 | for(i=1; i<order; i++) | ||
| 683 | if (lsp[i] < lsp[i-1]) { | ||
| 684 | //fprintf(stderr, "swap %d\n",i); | ||
| 685 | swaps++; | ||
| 686 | tmp = lsp[i-1]; | ||
| 687 | lsp[i-1] = lsp[i]-0.1; | ||
| 688 | lsp[i] = tmp+0.1; | ||
| 689 | i = 1; /* start check again, as swap may have caused out of order */ | ||
| 690 | } | ||
| 691 | |||
| 692 | return swaps; | ||
| 693 | } | 282 | } |
| 694 | 283 | ||
| 695 | void force_min_lsp_dist(float lsp[], int order) | 284 | void force_min_lsp_dist(float lsp[], int order) { |
| 696 | { | 285 | int i; |
| 697 | int i; | ||
| 698 | 286 | ||
| 699 | for(i=1; i<order; i++) | 287 | for (i = 1; i < order; i++) |
| 700 | if ((lsp[i]-lsp[i-1]) < 0.01) { | 288 | if ((lsp[i] - lsp[i - 1]) < 0.01) { |
| 701 | lsp[i] += 0.01; | 289 | lsp[i] += 0.01; |
| 702 | } | 290 | } |
| 703 | } | 291 | } |
| 704 | 292 | ||
| 705 | |||
| 706 | /*---------------------------------------------------------------------------*\ | 293 | /*---------------------------------------------------------------------------*\ |
| 707 | 294 | ||
| 708 | lpc_post_filter() | 295 | lpc_post_filter() |
| 709 | 296 | ||
| 710 | Applies a post filter to the LPC synthesis filter power spectrum | 297 | Applies a post filter to the LPC synthesis filter power spectrum |
| 711 | Pw, which supresses the inter-formant energy. | 298 | Pw, which suppresses the inter-formant energy. |
| 712 | 299 | ||
| 713 | The algorithm is from p267 (Section 8.6) of "Digital Speech", | 300 | The algorithm is from p267 (Section 8.6) of "Digital Speech", |
| 714 | edited by A.M. Kondoz, 1994 published by Wiley and Sons. Chapter 8 | 301 | edited by A.M. Kondoz, 1994 published by Wiley and Sons. Chapter 8 |
| @@ -722,7 +309,7 @@ void force_min_lsp_dist(float lsp[], int order) | |||
| 722 | it should be possible to implement this more efficiently in the | 309 | it should be possible to implement this more efficiently in the |
| 723 | time domain. Just not sure how to handle relative time delays | 310 | time domain. Just not sure how to handle relative time delays |
| 724 | between the synthesis stage and updating these coeffs. A smaller | 311 | between the synthesis stage and updating these coeffs. A smaller |
| 725 | FFT size might also be accetable to save CPU. | 312 | FFT size might also be acceptable to save CPU. |
| 726 | 313 | ||
| 727 | TODO: | 314 | TODO: |
| 728 | [ ] sync var names between Octave and C version | 315 | [ ] sync var names between Octave and C version |
| @@ -733,104 +320,97 @@ void force_min_lsp_dist(float lsp[], int order) | |||
| 733 | \*---------------------------------------------------------------------------*/ | 320 | \*---------------------------------------------------------------------------*/ |
| 734 | 321 | ||
| 735 | void lpc_post_filter(codec2_fftr_cfg fftr_fwd_cfg, float Pw[], float ak[], | 322 | void lpc_post_filter(codec2_fftr_cfg fftr_fwd_cfg, float Pw[], float ak[], |
| 736 | int order, int dump, float beta, float gamma, int bass_boost, float E) | 323 | int order, int dump, float beta, float gamma, |
| 737 | { | 324 | int bass_boost, float E) { |
| 738 | int i; | 325 | int i; |
| 739 | float x[FFT_ENC]; /* input to FFTs */ | 326 | float x[FFT_ENC]; /* input to FFTs */ |
| 740 | COMP Ww[FFT_ENC/2+1]; /* weighting spectrum */ | 327 | COMP Ww[FFT_ENC / 2 + 1]; /* weighting spectrum */ |
| 741 | float Rw[FFT_ENC/2+1]; /* R = WA */ | 328 | float Rw[FFT_ENC / 2 + 1]; /* R = WA */ |
| 742 | float e_before, e_after, gain; | 329 | float e_before, e_after, gain; |
| 743 | float Pfw; | 330 | float Pfw; |
| 744 | float max_Rw, min_Rw; | 331 | float max_Rw, min_Rw; |
| 745 | float coeff; | 332 | float coeff; |
| 746 | PROFILE_VAR(tstart, tfft1, taw, tfft2, tww, tr); | 333 | PROFILE_VAR(tstart, tfft1, taw, tfft2, tww, tr); |
| 747 | |||
| 748 | PROFILE_SAMPLE(tstart); | ||
| 749 | |||
| 750 | /* Determine weighting filter spectrum W(exp(jw)) ---------------*/ | ||
| 751 | |||
| 752 | for(i=0; i<FFT_ENC; i++) { | ||
| 753 | x[i] = 0.0; | ||
| 754 | } | ||
| 755 | 334 | ||
| 756 | x[0] = ak[0]; | 335 | PROFILE_SAMPLE(tstart); |
| 757 | coeff = gamma; | ||
| 758 | for(i=1; i<=order; i++) { | ||
| 759 | x[i] = ak[i] * coeff; | ||
| 760 | coeff *= gamma; | ||
| 761 | } | ||
| 762 | codec2_fftr(fftr_fwd_cfg, x, Ww); | ||
| 763 | 336 | ||
| 764 | PROFILE_SAMPLE_AND_LOG(tfft2, taw, " fft2"); | 337 | /* Determine weighting filter spectrum W(exp(jw)) ---------------*/ |
| 765 | 338 | ||
| 766 | for(i=0; i<FFT_ENC/2; i++) { | 339 | for (i = 0; i < FFT_ENC; i++) { |
| 767 | Ww[i].real = Ww[i].real*Ww[i].real + Ww[i].imag*Ww[i].imag; | 340 | x[i] = 0.0; |
| 768 | } | 341 | } |
| 769 | 342 | ||
| 770 | PROFILE_SAMPLE_AND_LOG(tww, tfft2, " Ww"); | 343 | x[0] = ak[0]; |
| 344 | coeff = gamma; | ||
| 345 | for (i = 1; i <= order; i++) { | ||
| 346 | x[i] = ak[i] * coeff; | ||
| 347 | coeff *= gamma; | ||
| 348 | } | ||
| 349 | codec2_fftr(fftr_fwd_cfg, x, Ww); | ||
| 771 | 350 | ||
| 772 | /* Determined combined filter R = WA ---------------------------*/ | 351 | PROFILE_SAMPLE_AND_LOG(tfft2, taw, " fft2"); |
| 773 | 352 | ||
| 774 | max_Rw = 0.0; min_Rw = 1E32; | 353 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 775 | for(i=0; i<FFT_ENC/2; i++) { | 354 | Ww[i].real = Ww[i].real * Ww[i].real + Ww[i].imag * Ww[i].imag; |
| 776 | Rw[i] = sqrtf(Ww[i].real * Pw[i]); | 355 | } |
| 777 | if (Rw[i] > max_Rw) | ||
| 778 | max_Rw = Rw[i]; | ||
| 779 | if (Rw[i] < min_Rw) | ||
| 780 | min_Rw = Rw[i]; | ||
| 781 | 356 | ||
| 782 | } | 357 | PROFILE_SAMPLE_AND_LOG(tww, tfft2, " Ww"); |
| 783 | 358 | ||
| 784 | PROFILE_SAMPLE_AND_LOG(tr, tww, " R"); | 359 | /* Determined combined filter R = WA ---------------------------*/ |
| 785 | 360 | ||
| 786 | #ifdef DUMP | 361 | max_Rw = 0.0; |
| 787 | if (dump) | 362 | min_Rw = 1E32; |
| 788 | dump_Rw(Rw); | 363 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 789 | #endif | 364 | Rw[i] = sqrtf(Ww[i].real * Pw[i]); |
| 365 | if (Rw[i] > max_Rw) max_Rw = Rw[i]; | ||
| 366 | if (Rw[i] < min_Rw) min_Rw = Rw[i]; | ||
| 367 | } | ||
| 790 | 368 | ||
| 791 | /* create post filter mag spectrum and apply ------------------*/ | 369 | PROFILE_SAMPLE_AND_LOG(tr, tww, " R"); |
| 792 | 370 | ||
| 793 | /* measure energy before post filtering */ | 371 | #ifdef DUMP |
| 372 | if (dump) dump_Rw(Rw); | ||
| 373 | #endif | ||
| 794 | 374 | ||
| 795 | e_before = 1E-4; | 375 | /* create post filter mag spectrum and apply ------------------*/ |
| 796 | for(i=0; i<FFT_ENC/2; i++) | ||
| 797 | e_before += Pw[i]; | ||
| 798 | 376 | ||
| 799 | /* apply post filter and measure energy */ | 377 | /* measure energy before post filtering */ |
| 800 | 378 | ||
| 801 | #ifdef DUMP | 379 | e_before = 1E-4; |
| 802 | if (dump) | 380 | for (i = 0; i < FFT_ENC / 2; i++) e_before += Pw[i]; |
| 803 | dump_Pwb(Pw); | ||
| 804 | #endif | ||
| 805 | 381 | ||
| 382 | /* apply post filter and measure energy */ | ||
| 806 | 383 | ||
| 807 | e_after = 1E-4; | 384 | #ifdef DUMP |
| 808 | for(i=0; i<FFT_ENC/2; i++) { | 385 | if (dump) dump_Pwb(Pw); |
| 809 | Pfw = powf(Rw[i], beta); | 386 | #endif |
| 810 | Pw[i] *= Pfw * Pfw; | ||
| 811 | e_after += Pw[i]; | ||
| 812 | } | ||
| 813 | gain = e_before/e_after; | ||
| 814 | 387 | ||
| 815 | /* apply gain factor to normalise energy, and LPC Energy */ | 388 | e_after = 1E-4; |
| 389 | for (i = 0; i < FFT_ENC / 2; i++) { | ||
| 390 | Pfw = powf(Rw[i], beta); | ||
| 391 | Pw[i] *= Pfw * Pfw; | ||
| 392 | e_after += Pw[i]; | ||
| 393 | } | ||
| 394 | gain = e_before / e_after; | ||
| 816 | 395 | ||
| 817 | gain *= E; | 396 | /* apply gain factor to normalise energy, and LPC Energy */ |
| 818 | for(i=0; i<FFT_ENC/2; i++) { | 397 | |
| 819 | Pw[i] *= gain; | 398 | gain *= E; |
| 820 | } | 399 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 400 | Pw[i] *= gain; | ||
| 401 | } | ||
| 821 | 402 | ||
| 822 | if (bass_boost) { | 403 | if (bass_boost) { |
| 823 | /* add 3dB to first 1 kHz to account for LP effect of PF */ | 404 | /* add 3dB to first 1 kHz to account for LP effect of PF */ |
| 824 | 405 | ||
| 825 | for(i=0; i<FFT_ENC/8; i++) { | 406 | for (i = 0; i < FFT_ENC / 8; i++) { |
| 826 | Pw[i] *= 1.4*1.4; | 407 | Pw[i] *= 1.4 * 1.4; |
| 827 | } | ||
| 828 | } | 408 | } |
| 409 | } | ||
| 829 | 410 | ||
| 830 | PROFILE_SAMPLE_AND_LOG2(tr, " filt"); | 411 | PROFILE_SAMPLE_AND_LOG2(tr, " filt"); |
| 831 | } | 412 | } |
| 832 | 413 | ||
| 833 | |||
| 834 | /*---------------------------------------------------------------------------*\ | 414 | /*---------------------------------------------------------------------------*\ |
| 835 | 415 | ||
| 836 | aks_to_M2() | 416 | aks_to_M2() |
| @@ -841,134 +421,125 @@ void lpc_post_filter(codec2_fftr_cfg fftr_fwd_cfg, float Pw[], float ak[], | |||
| 841 | 421 | ||
| 842 | \*---------------------------------------------------------------------------*/ | 422 | \*---------------------------------------------------------------------------*/ |
| 843 | 423 | ||
| 844 | void aks_to_M2( | 424 | void aks_to_M2(codec2_fftr_cfg fftr_fwd_cfg, float ak[], /* LPC's */ |
| 845 | codec2_fftr_cfg fftr_fwd_cfg, | 425 | int order, |
| 846 | float ak[], /* LPC's */ | 426 | MODEL *model, /* sinusoidal model parameters for this frame */ |
| 847 | int order, | 427 | float E, /* energy term */ |
| 848 | MODEL *model, /* sinusoidal model parameters for this frame */ | 428 | float *snr, /* signal to noise ratio for this frame in dB */ |
| 849 | float E, /* energy term */ | 429 | int dump, /* true to dump sample to dump file */ |
| 850 | float *snr, /* signal to noise ratio for this frame in dB */ | 430 | int sim_pf, /* true to simulate a post filter */ |
| 851 | int dump, /* true to dump sample to dump file */ | 431 | int pf, /* true to enable actual LPC post filter */ |
| 852 | int sim_pf, /* true to simulate a post filter */ | 432 | int bass_boost, /* enable LPC filter 0-1kHz 3dB boost */ |
| 853 | int pf, /* true to enable actual LPC post filter */ | 433 | float beta, float gamma, /* LPC post filter parameters */ |
| 854 | int bass_boost, /* enable LPC filter 0-1kHz 3dB boost */ | 434 | COMP Aw[] /* output power spectrum */ |
| 855 | float beta, | 435 | ) { |
| 856 | float gamma, /* LPC post filter parameters */ | 436 | int i, m; /* loop variables */ |
| 857 | COMP Aw[] /* output power spectrum */ | 437 | int am, bm; /* limits of current band */ |
| 858 | ) | 438 | float r; /* no. rads/bin */ |
| 859 | { | 439 | float Em; /* energy in band */ |
| 860 | int i,m; /* loop variables */ | 440 | float Am; /* spectral amplitude sample */ |
| 861 | int am,bm; /* limits of current band */ | ||
| 862 | float r; /* no. rads/bin */ | ||
| 863 | float Em; /* energy in band */ | ||
| 864 | float Am; /* spectral amplitude sample */ | ||
| 865 | float signal, noise; | 441 | float signal, noise; |
| 866 | PROFILE_VAR(tstart, tfft, tpw, tpf); | 442 | PROFILE_VAR(tstart, tfft, tpw, tpf); |
| 867 | 443 | ||
| 868 | PROFILE_SAMPLE(tstart); | 444 | PROFILE_SAMPLE(tstart); |
| 869 | 445 | ||
| 870 | r = TWO_PI/(FFT_ENC); | 446 | r = TWO_PI / (FFT_ENC); |
| 871 | 447 | ||
| 872 | /* Determine DFT of A(exp(jw)) --------------------------------------------*/ | 448 | /* Determine DFT of A(exp(jw)) --------------------------------------------*/ |
| 873 | { | 449 | { |
| 874 | float a[FFT_ENC]; /* input to FFT for power spectrum */ | 450 | float a[FFT_ENC]; /* input to FFT for power spectrum */ |
| 875 | 451 | ||
| 876 | for(i=0; i<FFT_ENC; i++) { | 452 | for (i = 0; i < FFT_ENC; i++) { |
| 877 | a[i] = 0.0; | 453 | a[i] = 0.0; |
| 878 | } | 454 | } |
| 879 | 455 | ||
| 880 | for(i=0; i<=order; i++) | 456 | for (i = 0; i <= order; i++) a[i] = ak[i]; |
| 881 | a[i] = ak[i]; | 457 | codec2_fftr(fftr_fwd_cfg, a, Aw); |
| 882 | codec2_fftr(fftr_fwd_cfg, a, Aw); | ||
| 883 | } | 458 | } |
| 884 | PROFILE_SAMPLE_AND_LOG(tfft, tstart, " fft"); | 459 | PROFILE_SAMPLE_AND_LOG(tfft, tstart, " fft"); |
| 885 | 460 | ||
| 886 | /* Determine power spectrum P(w) = E/(A(exp(jw))^2 ------------------------*/ | 461 | /* Determine power spectrum P(w) = E/(A(exp(jw))^2 ------------------------*/ |
| 887 | 462 | ||
| 888 | float Pw[FFT_ENC/2]; | 463 | float Pw[FFT_ENC / 2]; |
| 889 | 464 | ||
| 890 | #ifndef FDV_ARM_MATH | 465 | #ifndef FDV_ARM_MATH |
| 891 | for(i=0; i<FFT_ENC/2; i++) { | 466 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 892 | Pw[i] = 1.0/(Aw[i].real*Aw[i].real + Aw[i].imag*Aw[i].imag + 1E-6); | 467 | Pw[i] = 1.0 / (Aw[i].real * Aw[i].real + Aw[i].imag * Aw[i].imag + 1E-6); |
| 893 | } | 468 | } |
| 894 | #else | 469 | #else |
| 895 | // this difference may seem strange, but the gcc for STM32F4 generates almost 5 times | 470 | // this difference may seem strange, but the gcc for STM32F4 generates almost |
| 896 | // faster code with the two loops: 1120 ms -> 242 ms | 471 | // 5 times faster code with the two loops: 1120 ms -> 242 ms so please leave |
| 897 | // so please leave it as is or improve further | 472 | // it as is or improve further since this code is called 4 times it results in |
| 898 | // since this code is called 4 times it results in almost 4ms gain (21ms -> 17ms per audio frame decode @ 1300 ) | 473 | // almost 4ms gain (21ms -> 17ms per audio frame decode @ 1300 ) |
| 899 | 474 | ||
| 900 | for(i=0; i<FFT_ENC/2; i++) | 475 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 901 | { | 476 | Pw[i] = Aw[i].real * Aw[i].real + Aw[i].imag * Aw[i].imag + 1E-6; |
| 902 | Pw[i] = Aw[i].real * Aw[i].real + Aw[i].imag * Aw[i].imag + 1E-6; | ||
| 903 | } | 477 | } |
| 904 | for(i=0; i<FFT_ENC/2; i++) { | 478 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 905 | Pw[i] = 1.0/(Pw[i]); | 479 | Pw[i] = 1.0 / (Pw[i]); |
| 906 | } | 480 | } |
| 907 | #endif | 481 | #endif |
| 908 | 482 | ||
| 909 | PROFILE_SAMPLE_AND_LOG(tpw, tfft, " Pw"); | 483 | PROFILE_SAMPLE_AND_LOG(tpw, tfft, " Pw"); |
| 910 | 484 | ||
| 911 | if (pf) | 485 | if (pf) |
| 912 | lpc_post_filter(fftr_fwd_cfg, Pw, ak, order, dump, beta, gamma, bass_boost, E); | 486 | lpc_post_filter(fftr_fwd_cfg, Pw, ak, order, dump, beta, gamma, bass_boost, |
| 487 | E); | ||
| 913 | else { | 488 | else { |
| 914 | for(i=0; i<FFT_ENC/2; i++) { | 489 | for (i = 0; i < FFT_ENC / 2; i++) { |
| 915 | Pw[i] *= E; | 490 | Pw[i] *= E; |
| 916 | } | 491 | } |
| 917 | } | 492 | } |
| 918 | 493 | ||
| 919 | PROFILE_SAMPLE_AND_LOG(tpf, tpw, " LPC post filter"); | 494 | PROFILE_SAMPLE_AND_LOG(tpf, tpw, " LPC post filter"); |
| 920 | 495 | ||
| 921 | #ifdef DUMP | 496 | #ifdef DUMP |
| 922 | if (dump) | 497 | if (dump) dump_Pw(Pw); |
| 923 | dump_Pw(Pw); | 498 | #endif |
| 924 | #endif | ||
| 925 | 499 | ||
| 926 | /* Determine magnitudes from P(w) ----------------------------------------*/ | 500 | /* Determine magnitudes from P(w) ----------------------------------------*/ |
| 927 | 501 | ||
| 928 | /* when used just by decoder {A} might be all zeroes so init signal | 502 | /* when used just by decoder {A} might be all zeroes so init signal |
| 929 | and noise to prevent log(0) errors */ | 503 | and noise to prevent log(0) errors */ |
| 930 | 504 | ||
| 931 | signal = 1E-30; noise = 1E-32; | 505 | signal = 1E-30; |
| 932 | 506 | noise = 1E-32; | |
| 933 | for(m=1; m<=model->L; m++) { | 507 | |
| 934 | am = (int)((m - 0.5)*model->Wo/r + 0.5); | 508 | for (m = 1; m <= model->L; m++) { |
| 935 | bm = (int)((m + 0.5)*model->Wo/r + 0.5); | 509 | am = (int)((m - 0.5) * model->Wo / r + 0.5); |
| 936 | 510 | bm = (int)((m + 0.5) * model->Wo / r + 0.5); | |
| 937 | // FIXME: With arm_rfft_fast_f32 we have to use this | 511 | |
| 938 | // otherwise sometimes a to high bm is calculated | 512 | // FIXME: With arm_rfft_fast_f32 we have to use this |
| 939 | // which causes trouble later in the calculation | 513 | // otherwise sometimes a to high bm is calculated |
| 940 | // chain | 514 | // which causes trouble later in the calculation |
| 941 | // it seems for some reason model->Wo is calculated somewhat too high | 515 | // chain |
| 942 | if (bm>FFT_ENC/2) | 516 | // it seems for some reason model->Wo is calculated somewhat too high |
| 943 | { | 517 | if (bm > FFT_ENC / 2) { |
| 944 | bm = FFT_ENC/2; | 518 | bm = FFT_ENC / 2; |
| 945 | } | 519 | } |
| 946 | Em = 0.0; | 520 | Em = 0.0; |
| 947 | 521 | ||
| 948 | for(i=am; i<bm; i++) | 522 | for (i = am; i < bm; i++) Em += Pw[i]; |
| 949 | Em += Pw[i]; | 523 | Am = sqrtf(Em); |
| 950 | Am = sqrtf(Em); | 524 | |
| 951 | 525 | signal += model->A[m] * model->A[m]; | |
| 952 | signal += model->A[m]*model->A[m]; | 526 | noise += (model->A[m] - Am) * (model->A[m] - Am); |
| 953 | noise += (model->A[m] - Am)*(model->A[m] - Am); | 527 | |
| 954 | 528 | /* This code significantly improves perf of LPC model, in | |
| 955 | /* This code significantly improves perf of LPC model, in | 529 | particular when combined with phase0. The LPC spectrum tends |
| 956 | particular when combined with phase0. The LPC spectrum tends | 530 | to track just under the peaks of the spectral envelope, and |
| 957 | to track just under the peaks of the spectral envelope, and | 531 | just above nulls. This algorithm does the reverse to |
| 958 | just above nulls. This algorithm does the reverse to | 532 | compensate - raising the amplitudes of spectral peaks, while |
| 959 | compensate - raising the amplitudes of spectral peaks, while | 533 | attenuating the null. This enhances the formants, and |
| 960 | attenuating the null. This enhances the formants, and | 534 | suppresses the energy between formants. */ |
| 961 | supresses the energy between formants. */ | 535 | |
| 962 | 536 | if (sim_pf) { | |
| 963 | if (sim_pf) { | 537 | if (Am > model->A[m]) Am *= 0.7; |
| 964 | if (Am > model->A[m]) | 538 | if (Am < model->A[m]) Am *= 1.4; |
| 965 | Am *= 0.7; | 539 | } |
| 966 | if (Am < model->A[m]) | 540 | model->A[m] = Am; |
| 967 | Am *= 1.4; | ||
| 968 | } | ||
| 969 | model->A[m] = Am; | ||
| 970 | } | 541 | } |
| 971 | *snr = 10.0*log10f(signal/noise); | 542 | *snr = 10.0 * log10f(signal / noise); |
| 972 | 543 | ||
| 973 | PROFILE_SAMPLE_AND_LOG2(tpf, " rec"); | 544 | PROFILE_SAMPLE_AND_LOG2(tpf, " rec"); |
| 974 | } | 545 | } |
| @@ -983,19 +554,18 @@ void aks_to_M2( | |||
| 983 | 554 | ||
| 984 | \*---------------------------------------------------------------------------*/ | 555 | \*---------------------------------------------------------------------------*/ |
| 985 | 556 | ||
| 986 | int encode_Wo(C2CONST *c2const, float Wo, int bits) | 557 | int encode_Wo(C2CONST *c2const, float Wo, int bits) { |
| 987 | { | 558 | int index, Wo_levels = 1 << bits; |
| 988 | int index, Wo_levels = 1<<bits; | 559 | float Wo_min = c2const->Wo_min; |
| 989 | float Wo_min = c2const->Wo_min; | 560 | float Wo_max = c2const->Wo_max; |
| 990 | float Wo_max = c2const->Wo_max; | 561 | float norm; |
| 991 | float norm; | ||
| 992 | 562 | ||
| 993 | norm = (Wo - Wo_min)/(Wo_max - Wo_min); | 563 | norm = (Wo - Wo_min) / (Wo_max - Wo_min); |
| 994 | index = floorf(Wo_levels * norm + 0.5); | 564 | index = floorf(Wo_levels * norm + 0.5); |
| 995 | if (index < 0 ) index = 0; | 565 | if (index < 0) index = 0; |
| 996 | if (index > (Wo_levels-1)) index = Wo_levels-1; | 566 | if (index > (Wo_levels - 1)) index = Wo_levels - 1; |
| 997 | 567 | ||
| 998 | return index; | 568 | return index; |
| 999 | } | 569 | } |
| 1000 | 570 | ||
| 1001 | /*---------------------------------------------------------------------------*\ | 571 | /*---------------------------------------------------------------------------*\ |
| @@ -1008,18 +578,17 @@ int encode_Wo(C2CONST *c2const, float Wo, int bits) | |||
| 1008 | 578 | ||
| 1009 | \*---------------------------------------------------------------------------*/ | 579 | \*---------------------------------------------------------------------------*/ |
| 1010 | 580 | ||
| 1011 | float decode_Wo(C2CONST *c2const, int index, int bits) | 581 | float decode_Wo(C2CONST *c2const, int index, int bits) { |
| 1012 | { | 582 | float Wo_min = c2const->Wo_min; |
| 1013 | float Wo_min = c2const->Wo_min; | 583 | float Wo_max = c2const->Wo_max; |
| 1014 | float Wo_max = c2const->Wo_max; | 584 | float step; |
| 1015 | float step; | 585 | float Wo; |
| 1016 | float Wo; | 586 | int Wo_levels = 1 << bits; |
| 1017 | int Wo_levels = 1<<bits; | ||
| 1018 | 587 | ||
| 1019 | step = (Wo_max - Wo_min)/Wo_levels; | 588 | step = (Wo_max - Wo_min) / Wo_levels; |
| 1020 | Wo = Wo_min + step*(index); | 589 | Wo = Wo_min + step * (index); |
| 1021 | 590 | ||
| 1022 | return Wo; | 591 | return Wo; |
| 1023 | } | 592 | } |
| 1024 | 593 | ||
| 1025 | /*---------------------------------------------------------------------------*\ | 594 | /*---------------------------------------------------------------------------*\ |
| @@ -1032,19 +601,18 @@ float decode_Wo(C2CONST *c2const, int index, int bits) | |||
| 1032 | 601 | ||
| 1033 | \*---------------------------------------------------------------------------*/ | 602 | \*---------------------------------------------------------------------------*/ |
| 1034 | 603 | ||
| 1035 | int encode_log_Wo(C2CONST *c2const, float Wo, int bits) | 604 | int encode_log_Wo(C2CONST *c2const, float Wo, int bits) { |
| 1036 | { | 605 | int index, Wo_levels = 1 << bits; |
| 1037 | int index, Wo_levels = 1<<bits; | 606 | float Wo_min = c2const->Wo_min; |
| 1038 | float Wo_min = c2const->Wo_min; | 607 | float Wo_max = c2const->Wo_max; |
| 1039 | float Wo_max = c2const->Wo_max; | 608 | float norm; |
| 1040 | float norm; | ||
| 1041 | 609 | ||
| 1042 | norm = (log10f(Wo) - log10f(Wo_min))/(log10f(Wo_max) - log10f(Wo_min)); | 610 | norm = (log10f(Wo) - log10f(Wo_min)) / (log10f(Wo_max) - log10f(Wo_min)); |
| 1043 | index = floorf(Wo_levels * norm + 0.5); | 611 | index = floorf(Wo_levels * norm + 0.5); |
| 1044 | if (index < 0 ) index = 0; | 612 | if (index < 0) index = 0; |
| 1045 | if (index > (Wo_levels-1)) index = Wo_levels-1; | 613 | if (index > (Wo_levels - 1)) index = Wo_levels - 1; |
| 1046 | 614 | ||
| 1047 | return index; | 615 | return index; |
| 1048 | } | 616 | } |
| 1049 | 617 | ||
| 1050 | /*---------------------------------------------------------------------------*\ | 618 | /*---------------------------------------------------------------------------*\ |
| @@ -1057,99 +625,18 @@ int encode_log_Wo(C2CONST *c2const, float Wo, int bits) | |||
| 1057 | 625 | ||
| 1058 | \*---------------------------------------------------------------------------*/ | 626 | \*---------------------------------------------------------------------------*/ |
| 1059 | 627 | ||
| 1060 | float decode_log_Wo(C2CONST *c2const, int index, int bits) | 628 | float decode_log_Wo(C2CONST *c2const, int index, int bits) { |
| 1061 | { | 629 | float Wo_min = c2const->Wo_min; |
| 1062 | float Wo_min = c2const->Wo_min; | 630 | float Wo_max = c2const->Wo_max; |
| 1063 | float Wo_max = c2const->Wo_max; | 631 | float step; |
| 1064 | float step; | 632 | float Wo; |
| 1065 | float Wo; | 633 | int Wo_levels = 1 << bits; |
| 1066 | int Wo_levels = 1<<bits; | ||
| 1067 | |||
| 1068 | step = (log10f(Wo_max) - log10f(Wo_min))/Wo_levels; | ||
| 1069 | Wo = log10f(Wo_min) + step*(index); | ||
| 1070 | |||
| 1071 | return POW10F(Wo); | ||
| 1072 | } | ||
| 1073 | |||
| 1074 | #if 0 | ||
| 1075 | /*---------------------------------------------------------------------------*\ | ||
| 1076 | |||
| 1077 | FUNCTION....: encode_Wo_dt() | ||
| 1078 | AUTHOR......: David Rowe | ||
| 1079 | DATE CREATED: 6 Nov 2011 | ||
| 1080 | |||
| 1081 | Encodes Wo difference from last frame. | ||
| 1082 | |||
| 1083 | \*---------------------------------------------------------------------------*/ | ||
| 1084 | |||
| 1085 | int encode_Wo_dt(C2CONST *c2const, float Wo, float prev_Wo) | ||
| 1086 | { | ||
| 1087 | int index, mask, max_index, min_index; | ||
| 1088 | float Wo_min = c2const->Wo_min; | ||
| 1089 | float Wo_max = c2const->Wo_max; | ||
| 1090 | float norm; | ||
| 1091 | |||
| 1092 | norm = (Wo - prev_Wo)/(Wo_max - Wo_min); | ||
| 1093 | index = floorf(WO_LEVELS * norm + 0.5); | ||
| 1094 | //printf("ENC index: %d ", index); | ||
| 1095 | |||
| 1096 | /* hard limit */ | ||
| 1097 | |||
| 1098 | max_index = (1 << (WO_DT_BITS-1)) - 1; | ||
| 1099 | min_index = - (max_index+1); | ||
| 1100 | if (index > max_index) index = max_index; | ||
| 1101 | if (index < min_index) index = min_index; | ||
| 1102 | //printf("max_index: %d min_index: %d hard index: %d ", | ||
| 1103 | // max_index, min_index, index); | ||
| 1104 | |||
| 1105 | /* mask so that only LSB WO_DT_BITS remain, bit WO_DT_BITS is the sign bit */ | ||
| 1106 | |||
| 1107 | mask = ((1 << WO_DT_BITS) - 1); | ||
| 1108 | index &= mask; | ||
| 1109 | //printf("mask: 0x%x index: 0x%x\n", mask, index); | ||
| 1110 | |||
| 1111 | return index; | ||
| 1112 | } | ||
| 1113 | |||
| 1114 | /*---------------------------------------------------------------------------*\ | ||
| 1115 | |||
| 1116 | FUNCTION....: decode_Wo_dt() | ||
| 1117 | AUTHOR......: David Rowe | ||
| 1118 | DATE CREATED: 6 Nov 2011 | ||
| 1119 | |||
| 1120 | Decodes Wo using WO_DT_BITS difference from last frame. | ||
| 1121 | |||
| 1122 | \*---------------------------------------------------------------------------*/ | ||
| 1123 | |||
| 1124 | float decode_Wo_dt(C2CONST *c2const, int index, float prev_Wo) | ||
| 1125 | { | ||
| 1126 | float Wo_min = c2const->Wo_min; | ||
| 1127 | float Wo_max = c2const->Wo_max; | ||
| 1128 | float step; | ||
| 1129 | float Wo; | ||
| 1130 | int mask; | ||
| 1131 | |||
| 1132 | /* sign extend index */ | ||
| 1133 | |||
| 1134 | //printf("DEC index: %d "); | ||
| 1135 | if (index & (1 << (WO_DT_BITS-1))) { | ||
| 1136 | mask = ~((1 << WO_DT_BITS) - 1); | ||
| 1137 | index |= mask; | ||
| 1138 | } | ||
| 1139 | //printf("DEC mask: 0x%x index: %d \n", mask, index); | ||
| 1140 | |||
| 1141 | step = (Wo_max - Wo_min)/WO_LEVELS; | ||
| 1142 | Wo = prev_Wo + step*(index); | ||
| 1143 | |||
| 1144 | /* bit errors can make us go out of range leading to all sorts of | ||
| 1145 | probs like seg faults */ | ||
| 1146 | 634 | ||
| 1147 | if (Wo > Wo_max) Wo = Wo_max; | 635 | step = (log10f(Wo_max) - log10f(Wo_min)) / Wo_levels; |
| 1148 | if (Wo < Wo_min) Wo = Wo_min; | 636 | Wo = log10f(Wo_min) + step * (index); |
| 1149 | 637 | ||
| 1150 | return Wo; | 638 | return POW10F(Wo); |
| 1151 | } | 639 | } |
| 1152 | #endif | ||
| 1153 | 640 | ||
| 1154 | /*---------------------------------------------------------------------------*\ | 641 | /*---------------------------------------------------------------------------*\ |
| 1155 | 642 | ||
| @@ -1163,56 +650,46 @@ float decode_Wo_dt(C2CONST *c2const, int index, float prev_Wo) | |||
| 1163 | 650 | ||
| 1164 | \*---------------------------------------------------------------------------*/ | 651 | \*---------------------------------------------------------------------------*/ |
| 1165 | 652 | ||
| 1166 | float speech_to_uq_lsps(float lsp[], | 653 | float speech_to_uq_lsps(float lsp[], float ak[], float Sn[], float w[], |
| 1167 | float ak[], | 654 | int m_pitch, int order) { |
| 1168 | float Sn[], | 655 | int i, roots; |
| 1169 | float w[], | 656 | float Wn[m_pitch]; |
| 1170 | int m_pitch, | 657 | float R[order + 1]; |
| 1171 | int order | 658 | float e, E; |
| 1172 | ) | 659 | |
| 1173 | { | 660 | e = 0.0; |
| 1174 | int i, roots; | 661 | for (i = 0; i < m_pitch; i++) { |
| 1175 | float Wn[m_pitch]; | 662 | Wn[i] = Sn[i] * w[i]; |
| 1176 | float R[order+1]; | 663 | e += Wn[i] * Wn[i]; |
| 1177 | float e, E; | 664 | } |
| 1178 | |||
| 1179 | e = 0.0; | ||
| 1180 | for(i=0; i<m_pitch; i++) { | ||
| 1181 | Wn[i] = Sn[i]*w[i]; | ||
| 1182 | e += Wn[i]*Wn[i]; | ||
| 1183 | } | ||
| 1184 | 665 | ||
| 1185 | /* trap 0 energy case as LPC analysis will fail */ | 666 | /* trap 0 energy case as LPC analysis will fail */ |
| 1186 | 667 | ||
| 1187 | if (e == 0.0) { | 668 | if (e == 0.0) { |
| 1188 | for(i=0; i<order; i++) | 669 | for (i = 0; i < order; i++) lsp[i] = (PI / order) * (float)i; |
| 1189 | lsp[i] = (PI/order)*(float)i; | 670 | return 0.0; |
| 1190 | return 0.0; | 671 | } |
| 1191 | } | ||
| 1192 | 672 | ||
| 1193 | autocorrelate(Wn, R, m_pitch, order); | 673 | autocorrelate(Wn, R, m_pitch, order); |
| 1194 | levinson_durbin(R, ak, order); | 674 | levinson_durbin(R, ak, order); |
| 1195 | 675 | ||
| 1196 | E = 0.0; | 676 | E = 0.0; |
| 1197 | for(i=0; i<=order; i++) | 677 | for (i = 0; i <= order; i++) E += ak[i] * R[i]; |
| 1198 | E += ak[i]*R[i]; | ||
| 1199 | 678 | ||
| 1200 | /* 15 Hz BW expansion as I can't hear the difference and it may help | 679 | /* 15 Hz BW expansion as I can't hear the difference and it may help |
| 1201 | help occasional fails in the LSP root finding. Important to do this | 680 | help occasional fails in the LSP root finding. Important to do this |
| 1202 | after energy calculation to avoid -ve energy values. | 681 | after energy calculation to avoid -ve energy values. |
| 1203 | */ | 682 | */ |
| 1204 | 683 | ||
| 1205 | for(i=0; i<=order; i++) | 684 | for (i = 0; i <= order; i++) ak[i] *= powf(0.994, (float)i); |
| 1206 | ak[i] *= powf(0.994,(float)i); | ||
| 1207 | 685 | ||
| 1208 | roots = lpc_to_lsp(ak, order, lsp, 5, LSP_DELTA1); | 686 | roots = lpc_to_lsp(ak, order, lsp, 5, LSP_DELTA1); |
| 1209 | if (roots != order) { | 687 | if (roots != order) { |
| 1210 | /* if root finding fails use some benign LSP values instead */ | 688 | /* if root finding fails use some benign LSP values instead */ |
| 1211 | for(i=0; i<order; i++) | 689 | for (i = 0; i < order; i++) lsp[i] = (PI / order) * (float)i; |
| 1212 | lsp[i] = (PI/order)*(float)i; | 690 | } |
| 1213 | } | ||
| 1214 | 691 | ||
| 1215 | return E; | 692 | return E; |
| 1216 | } | 693 | } |
| 1217 | 694 | ||
| 1218 | /*---------------------------------------------------------------------------*\ | 695 | /*---------------------------------------------------------------------------*\ |
| @@ -1221,317 +698,60 @@ float speech_to_uq_lsps(float lsp[], | |||
| 1221 | AUTHOR......: David Rowe | 698 | AUTHOR......: David Rowe |
| 1222 | DATE CREATED: 22/8/2010 | 699 | DATE CREATED: 22/8/2010 |
| 1223 | 700 | ||
| 1224 | Thirty-six bit sclar LSP quantiser. From a vector of unquantised | 701 | Scalar LSP quantiser. From a vector of unquantised (floating point) |
| 1225 | (floating point) LSPs finds the quantised LSP indexes. | 702 | LSPs finds the quantised LSP indexes. |
| 1226 | |||
| 1227 | \*---------------------------------------------------------------------------*/ | ||
| 1228 | |||
| 1229 | void encode_lsps_scalar(int indexes[], float lsp[], int order) | ||
| 1230 | { | ||
| 1231 | int i,k,m; | ||
| 1232 | float wt[1]; | ||
| 1233 | float lsp_hz[order]; | ||
| 1234 | const float * cb; | ||
| 1235 | float se; | ||
| 1236 | |||
| 1237 | /* convert from radians to Hz so we can use human readable | ||
| 1238 | frequencies */ | ||
| 1239 | |||
| 1240 | for(i=0; i<order; i++) | ||
| 1241 | lsp_hz[i] = (4000.0/PI)*lsp[i]; | ||
| 1242 | |||
| 1243 | /* scalar quantisers */ | ||
| 1244 | |||
| 1245 | wt[0] = 1.0; | ||
| 1246 | for(i=0; i<order; i++) { | ||
| 1247 | k = lsp_cb[i].k; | ||
| 1248 | m = lsp_cb[i].m; | ||
| 1249 | cb = lsp_cb[i].cb; | ||
| 1250 | indexes[i] = quantise(cb, &lsp_hz[i], wt, k, m, &se); | ||
| 1251 | } | ||
| 1252 | } | ||
| 1253 | |||
| 1254 | /*---------------------------------------------------------------------------*\ | ||
| 1255 | |||
| 1256 | FUNCTION....: decode_lsps_scalar() | ||
| 1257 | AUTHOR......: David Rowe | ||
| 1258 | DATE CREATED: 22/8/2010 | ||
| 1259 | |||
| 1260 | From a vector of quantised LSP indexes, returns the quantised | ||
| 1261 | (floating point) LSPs. | ||
| 1262 | |||
| 1263 | \*---------------------------------------------------------------------------*/ | ||
| 1264 | |||
| 1265 | void decode_lsps_scalar(float lsp[], int indexes[], int order) | ||
| 1266 | { | ||
| 1267 | int i,k; | ||
| 1268 | float lsp_hz[order]; | ||
| 1269 | const float * cb; | ||
| 1270 | |||
| 1271 | for(i=0; i<order; i++) { | ||
| 1272 | k = lsp_cb[i].k; | ||
| 1273 | cb = lsp_cb[i].cb; | ||
| 1274 | lsp_hz[i] = cb[indexes[i]*k]; | ||
| 1275 | } | ||
| 1276 | |||
| 1277 | /* convert back to radians */ | ||
| 1278 | |||
| 1279 | for(i=0; i<order; i++) | ||
| 1280 | lsp[i] = (PI/4000.0)*lsp_hz[i]; | ||
| 1281 | } | ||
| 1282 | |||
| 1283 | |||
| 1284 | /*---------------------------------------------------------------------------*\ | ||
| 1285 | |||
| 1286 | FUNCTION....: encode_mels_scalar() | ||
| 1287 | AUTHOR......: David Rowe | ||
| 1288 | DATE CREATED: April 2015 | ||
| 1289 | |||
| 1290 | Low bit rate mel coeff encoder. | ||
| 1291 | 703 | ||
| 1292 | \*---------------------------------------------------------------------------*/ | 704 | \*---------------------------------------------------------------------------*/ |
| 1293 | 705 | ||
| 1294 | void encode_mels_scalar(int indexes[], float mels[], int order) | 706 | void encode_lsps_scalar(int indexes[], float lsp[], int order) { |
| 1295 | { | 707 | int i, k, m; |
| 1296 | int i,m; | 708 | float wt[1]; |
| 1297 | float wt[1]; | 709 | float lsp_hz[order]; |
| 1298 | const float * cb; | 710 | const float *cb; |
| 1299 | float se, mel_, dmel; | 711 | float se; |
| 1300 | |||
| 1301 | /* scalar quantisers */ | ||
| 1302 | |||
| 1303 | wt[0] = 1.0; | ||
| 1304 | for(i=0; i<order; i++) { | ||
| 1305 | m = mel_cb[i].m; | ||
| 1306 | cb = mel_cb[i].cb; | ||
| 1307 | if (i%2) { | ||
| 1308 | /* on odd mels quantise difference */ | ||
| 1309 | mel_ = mel_cb[i-1].cb[indexes[i-1]]; | ||
| 1310 | dmel = mels[i] - mel_; | ||
| 1311 | indexes[i] = quantise(cb, &dmel, wt, 1, m, &se); | ||
| 1312 | //printf("%d mel: %f mel_: %f dmel: %f index: %d\n", i, mels[i], mel_, dmel, indexes[i]); | ||
| 1313 | } | ||
| 1314 | else { | ||
| 1315 | indexes[i] = quantise(cb, &mels[i], wt, 1, m, &se); | ||
| 1316 | //printf("%d mel: %f dmel: %f index: %d\n", i, mels[i], 0.0, indexes[i]); | ||
| 1317 | } | ||
| 1318 | 712 | ||
| 1319 | } | 713 | /* convert from radians to Hz so we can use human readable |
| 1320 | } | 714 | frequencies */ |
| 1321 | 715 | ||
| 716 | for (i = 0; i < order; i++) lsp_hz[i] = (4000.0 / PI) * lsp[i]; | ||
| 1322 | 717 | ||
| 1323 | /*---------------------------------------------------------------------------*\ | 718 | /* scalar quantisers */ |
| 1324 | 719 | ||
| 1325 | FUNCTION....: decode_mels_scalar() | 720 | wt[0] = 1.0; |
| 1326 | AUTHOR......: David Rowe | 721 | for (i = 0; i < order; i++) { |
| 1327 | DATE CREATED: April 2015 | 722 | k = lsp_cb[i].k; |
| 1328 | 723 | m = lsp_cb[i].m; | |
| 1329 | From a vector of quantised mel indexes, returns the quantised | 724 | cb = lsp_cb[i].cb; |
| 1330 | (floating point) mels. | 725 | indexes[i] = quantise(cb, &lsp_hz[i], wt, k, m, &se); |
| 1331 | 726 | } | |
| 1332 | \*---------------------------------------------------------------------------*/ | ||
| 1333 | |||
| 1334 | void decode_mels_scalar(float mels[], int indexes[], int order) | ||
| 1335 | { | ||
| 1336 | int i; | ||
| 1337 | const float * cb; | ||
| 1338 | |||
| 1339 | for(i=0; i<order; i++) { | ||
| 1340 | cb = mel_cb[i].cb; | ||
| 1341 | if (i%2) { | ||
| 1342 | /* on odd mels quantise difference */ | ||
| 1343 | mels[i] = mels[i-1] + cb[indexes[i]]; | ||
| 1344 | } | ||
| 1345 | else | ||
| 1346 | mels[i] = cb[indexes[i]]; | ||
| 1347 | } | ||
| 1348 | |||
| 1349 | } | ||
| 1350 | |||
| 1351 | |||
| 1352 | #ifdef __EXPERIMENTAL__ | ||
| 1353 | |||
| 1354 | /*---------------------------------------------------------------------------*\ | ||
| 1355 | |||
| 1356 | FUNCTION....: encode_lsps_diff_freq_vq() | ||
| 1357 | AUTHOR......: David Rowe | ||
| 1358 | DATE CREATED: 15 November 2011 | ||
| 1359 | |||
| 1360 | Twenty-five bit LSP quantiser. LSPs 1-4 are quantised with scalar | ||
| 1361 | LSP differences (in frequency, i.e difference from the previous | ||
| 1362 | LSP). LSPs 5-10 are quantised with a VQ trained generated using | ||
| 1363 | vqtrainjnd.c | ||
| 1364 | |||
| 1365 | \*---------------------------------------------------------------------------*/ | ||
| 1366 | |||
| 1367 | void encode_lsps_diff_freq_vq(int indexes[], float lsp[], int order) | ||
| 1368 | { | ||
| 1369 | int i,k,m; | ||
| 1370 | float lsp_hz[order]; | ||
| 1371 | float lsp__hz[order]; | ||
| 1372 | float dlsp[order]; | ||
| 1373 | float dlsp_[order]; | ||
| 1374 | float wt[order]; | ||
| 1375 | const float * cb; | ||
| 1376 | float se; | ||
| 1377 | |||
| 1378 | for(i=0; i<order; i++) { | ||
| 1379 | wt[i] = 1.0; | ||
| 1380 | } | ||
| 1381 | |||
| 1382 | /* convert from radians to Hz so we can use human readable | ||
| 1383 | frequencies */ | ||
| 1384 | |||
| 1385 | for(i=0; i<order; i++) | ||
| 1386 | lsp_hz[i] = (4000.0/PI)*lsp[i]; | ||
| 1387 | |||
| 1388 | /* scalar quantisers for LSP differences 1..4 */ | ||
| 1389 | |||
| 1390 | wt[0] = 1.0; | ||
| 1391 | for(i=0; i<4; i++) { | ||
| 1392 | if (i) | ||
| 1393 | dlsp[i] = lsp_hz[i] - lsp__hz[i-1]; | ||
| 1394 | else | ||
| 1395 | dlsp[0] = lsp_hz[0]; | ||
| 1396 | |||
| 1397 | k = lsp_cbd[i].k; | ||
| 1398 | m = lsp_cbd[i].m; | ||
| 1399 | cb = lsp_cbd[i].cb; | ||
| 1400 | indexes[i] = quantise(cb, &dlsp[i], wt, k, m, &se); | ||
| 1401 | dlsp_[i] = cb[indexes[i]*k]; | ||
| 1402 | |||
| 1403 | if (i) | ||
| 1404 | lsp__hz[i] = lsp__hz[i-1] + dlsp_[i]; | ||
| 1405 | else | ||
| 1406 | lsp__hz[0] = dlsp_[0]; | ||
| 1407 | } | ||
| 1408 | |||
| 1409 | /* VQ LSPs 5,6,7,8,9,10 */ | ||
| 1410 | |||
| 1411 | k = lsp_cbjnd[4].k; | ||
| 1412 | m = lsp_cbjnd[4].m; | ||
| 1413 | cb = lsp_cbjnd[4].cb; | ||
| 1414 | indexes[4] = quantise(cb, &lsp_hz[4], &wt[4], k, m, &se); | ||
| 1415 | } | 727 | } |
| 1416 | 728 | ||
| 1417 | |||
| 1418 | /*---------------------------------------------------------------------------*\ | 729 | /*---------------------------------------------------------------------------*\ |
| 1419 | 730 | ||
| 1420 | FUNCTION....: decode_lsps_diff_freq_vq() | 731 | FUNCTION....: decode_lsps_scalar() |
| 1421 | AUTHOR......: David Rowe | 732 | AUTHOR......: David Rowe |
| 1422 | DATE CREATED: 15 Nov 2011 | 733 | DATE CREATED: 22/8/2010 |
| 1423 | 734 | ||
| 1424 | From a vector of quantised LSP indexes, returns the quantised | 735 | From a vector of quantised LSP indexes, returns the quantised |
| 1425 | (floating point) LSPs. | 736 | (floating point) LSPs. |
| 1426 | 737 | ||
| 1427 | \*---------------------------------------------------------------------------*/ | 738 | \*---------------------------------------------------------------------------*/ |
| 1428 | 739 | ||
| 1429 | void decode_lsps_diff_freq_vq(float lsp_[], int indexes[], int order) | 740 | void decode_lsps_scalar(float lsp[], int indexes[], int order) { |
| 1430 | { | 741 | int i, k; |
| 1431 | int i,k,m; | 742 | float lsp_hz[order]; |
| 1432 | float dlsp_[order]; | 743 | const float *cb; |
| 1433 | float lsp__hz[order]; | ||
| 1434 | const float * cb; | ||
| 1435 | |||
| 1436 | /* scalar LSP differences */ | ||
| 1437 | |||
| 1438 | for(i=0; i<4; i++) { | ||
| 1439 | cb = lsp_cbd[i].cb; | ||
| 1440 | dlsp_[i] = cb[indexes[i]]; | ||
| 1441 | if (i) | ||
| 1442 | lsp__hz[i] = lsp__hz[i-1] + dlsp_[i]; | ||
| 1443 | else | ||
| 1444 | lsp__hz[0] = dlsp_[0]; | ||
| 1445 | } | ||
| 1446 | |||
| 1447 | /* VQ */ | ||
| 1448 | |||
| 1449 | k = lsp_cbjnd[4].k; | ||
| 1450 | m = lsp_cbjnd[4].m; | ||
| 1451 | cb = lsp_cbjnd[4].cb; | ||
| 1452 | for(i=4; i<order; i++) | ||
| 1453 | lsp__hz[i] = cb[indexes[4]*k+i-4]; | ||
| 1454 | |||
| 1455 | /* convert back to radians */ | ||
| 1456 | |||
| 1457 | for(i=0; i<order; i++) | ||
| 1458 | lsp_[i] = (PI/4000.0)*lsp__hz[i]; | ||
| 1459 | } | ||
| 1460 | |||
| 1461 | |||
| 1462 | /*---------------------------------------------------------------------------*\ | ||
| 1463 | |||
| 1464 | FUNCTION....: encode_lsps_diff_time() | ||
| 1465 | AUTHOR......: David Rowe | ||
| 1466 | DATE CREATED: 12 Sep 2012 | ||
| 1467 | |||
| 1468 | Encode difference from preious frames's LSPs using | ||
| 1469 | 3,3,2,2,2,2,1,1,1,1 scalar quantisers (18 bits total). | ||
| 1470 | |||
| 1471 | \*---------------------------------------------------------------------------*/ | ||
| 1472 | 744 | ||
| 1473 | void encode_lsps_diff_time(int indexes[], | 745 | for (i = 0; i < order; i++) { |
| 1474 | float lsps[], | 746 | k = lsp_cb[i].k; |
| 1475 | float lsps__prev[], | 747 | cb = lsp_cb[i].cb; |
| 1476 | int order) | 748 | lsp_hz[i] = cb[indexes[i] * k]; |
| 1477 | { | 749 | } |
| 1478 | int i,k,m; | ||
| 1479 | float lsps_dt[order]; | ||
| 1480 | float wt[LPC_MAX]; | ||
| 1481 | const float * cb; | ||
| 1482 | float se; | ||
| 1483 | |||
| 1484 | /* Determine difference in time and convert from radians to Hz so | ||
| 1485 | we can use human readable frequencies */ | ||
| 1486 | |||
| 1487 | for(i=0; i<order; i++) { | ||
| 1488 | lsps_dt[i] = (4000/PI)*(lsps[i] - lsps__prev[i]); | ||
| 1489 | } | ||
| 1490 | |||
| 1491 | /* scalar quantisers */ | ||
| 1492 | |||
| 1493 | wt[0] = 1.0; | ||
| 1494 | for(i=0; i<order; i++) { | ||
| 1495 | k = lsp_cbdt[i].k; | ||
| 1496 | m = lsp_cbdt[i].m; | ||
| 1497 | cb = lsp_cbdt[i].cb; | ||
| 1498 | indexes[i] = quantise(cb, &lsps_dt[i], wt, k, m, &se); | ||
| 1499 | } | ||
| 1500 | |||
| 1501 | } | ||
| 1502 | |||
| 1503 | |||
| 1504 | /*---------------------------------------------------------------------------*\ | ||
| 1505 | |||
| 1506 | FUNCTION....: decode_lsps_diff_time() | ||
| 1507 | AUTHOR......: David Rowe | ||
| 1508 | DATE CREATED: 15 Nov 2011 | ||
| 1509 | |||
| 1510 | From a quantised LSP indexes, returns the quantised | ||
| 1511 | (floating point) LSPs. | ||
| 1512 | |||
| 1513 | \*---------------------------------------------------------------------------*/ | ||
| 1514 | |||
| 1515 | void decode_lsps_diff_time( | ||
| 1516 | float lsps_[], | ||
| 1517 | int indexes[], | ||
| 1518 | float lsps__prev[], | ||
| 1519 | int order) | ||
| 1520 | { | ||
| 1521 | int i,k,m; | ||
| 1522 | const float * cb; | ||
| 1523 | |||
| 1524 | for(i=0; i<order; i++) | ||
| 1525 | lsps_[i] = lsps__prev[i]; | ||
| 1526 | 750 | ||
| 1527 | for(i=0; i<order; i++) { | 751 | /* convert back to radians */ |
| 1528 | k = lsp_cbdt[i].k; | ||
| 1529 | cb = lsp_cbdt[i].cb; | ||
| 1530 | lsps_[i] += (PI/4000.0)*cb[indexes[i]*k]; | ||
| 1531 | } | ||
| 1532 | 752 | ||
| 753 | for (i = 0; i < order; i++) lsp[i] = (PI / 4000.0) * lsp_hz[i]; | ||
| 1533 | } | 754 | } |
| 1534 | #endif | ||
| 1535 | 755 | ||
| 1536 | /*---------------------------------------------------------------------------*\ | 756 | /*---------------------------------------------------------------------------*\ |
| 1537 | 757 | ||
| @@ -1543,45 +763,40 @@ void decode_lsps_diff_time( | |||
| 1543 | 763 | ||
| 1544 | \*---------------------------------------------------------------------------*/ | 764 | \*---------------------------------------------------------------------------*/ |
| 1545 | 765 | ||
| 1546 | void encode_lsps_vq(int *indexes, float *x, float *xq, int order) | 766 | void encode_lsps_vq(int *indexes, float *x, float *xq, int order) { |
| 1547 | { | ||
| 1548 | int i, n1, n2, n3; | 767 | int i, n1, n2, n3; |
| 1549 | float err[order], err2[order], err3[order]; | 768 | float err[order], err2[order], err3[order]; |
| 1550 | float w[order], w2[order], w3[order]; | 769 | float w[order], w2[order], w3[order]; |
| 1551 | const float *codebook1 = lsp_cbjvm[0].cb; | 770 | const float *codebook1 = lsp_cbjmv[0].cb; |
| 1552 | const float *codebook2 = lsp_cbjvm[1].cb; | 771 | const float *codebook2 = lsp_cbjmv[1].cb; |
| 1553 | const float *codebook3 = lsp_cbjvm[2].cb; | 772 | const float *codebook3 = lsp_cbjmv[2].cb; |
| 1554 | 773 | ||
| 1555 | w[0] = MIN(x[0], x[1]-x[0]); | 774 | w[0] = MIN(x[0], x[1] - x[0]); |
| 1556 | for (i=1;i<order-1;i++) | 775 | for (i = 1; i < order - 1; i++) w[i] = MIN(x[i] - x[i - 1], x[i + 1] - x[i]); |
| 1557 | w[i] = MIN(x[i]-x[i-1], x[i+1]-x[i]); | 776 | w[order - 1] = MIN(x[order - 1] - x[order - 2], PI - x[order - 1]); |
| 1558 | w[order-1] = MIN(x[order-1]-x[order-2], PI-x[order-1]); | ||
| 1559 | 777 | ||
| 1560 | compute_weights(x, w, order); | 778 | compute_weights(x, w, order); |
| 1561 | 779 | ||
| 1562 | n1 = find_nearest(codebook1, lsp_cbjvm[0].m, x, order); | 780 | n1 = find_nearest(codebook1, lsp_cbjmv[0].m, x, order); |
| 1563 | 781 | ||
| 1564 | for (i=0;i<order;i++) | 782 | for (i = 0; i < order; i++) { |
| 1565 | { | 783 | xq[i] = codebook1[order * n1 + i]; |
| 1566 | xq[i] = codebook1[order*n1+i]; | ||
| 1567 | err[i] = x[i] - xq[i]; | 784 | err[i] = x[i] - xq[i]; |
| 1568 | } | 785 | } |
| 1569 | for (i=0;i<order/2;i++) | 786 | for (i = 0; i < order / 2; i++) { |
| 1570 | { | 787 | err2[i] = err[2 * i]; |
| 1571 | err2[i] = err[2*i]; | 788 | err3[i] = err[2 * i + 1]; |
| 1572 | err3[i] = err[2*i+1]; | 789 | w2[i] = w[2 * i]; |
| 1573 | w2[i] = w[2*i]; | 790 | w3[i] = w[2 * i + 1]; |
| 1574 | w3[i] = w[2*i+1]; | ||
| 1575 | } | 791 | } |
| 1576 | n2 = find_nearest_weighted(codebook2, lsp_cbjvm[1].m, err2, w2, order/2); | 792 | n2 = find_nearest_weighted(codebook2, lsp_cbjmv[1].m, err2, w2, order / 2); |
| 1577 | n3 = find_nearest_weighted(codebook3, lsp_cbjvm[2].m, err3, w3, order/2); | 793 | n3 = find_nearest_weighted(codebook3, lsp_cbjmv[2].m, err3, w3, order / 2); |
| 1578 | 794 | ||
| 1579 | indexes[0] = n1; | 795 | indexes[0] = n1; |
| 1580 | indexes[1] = n2; | 796 | indexes[1] = n2; |
| 1581 | indexes[2] = n3; | 797 | indexes[2] = n3; |
| 1582 | } | 798 | } |
| 1583 | 799 | ||
| 1584 | |||
| 1585 | /*---------------------------------------------------------------------------*\ | 800 | /*---------------------------------------------------------------------------*\ |
| 1586 | 801 | ||
| 1587 | FUNCTION....: decode_lsps_vq() | 802 | FUNCTION....: decode_lsps_vq() |
| @@ -1590,31 +805,28 @@ void encode_lsps_vq(int *indexes, float *x, float *xq, int order) | |||
| 1590 | 805 | ||
| 1591 | \*---------------------------------------------------------------------------*/ | 806 | \*---------------------------------------------------------------------------*/ |
| 1592 | 807 | ||
| 1593 | void decode_lsps_vq(int *indexes, float *xq, int order, int stages) | 808 | void decode_lsps_vq(int *indexes, float *xq, int order, int stages) { |
| 1594 | { | ||
| 1595 | int i, n1, n2, n3; | 809 | int i, n1, n2, n3; |
| 1596 | const float *codebook1 = lsp_cbjvm[0].cb; | 810 | const float *codebook1 = lsp_cbjmv[0].cb; |
| 1597 | const float *codebook2 = lsp_cbjvm[1].cb; | 811 | const float *codebook2 = lsp_cbjmv[1].cb; |
| 1598 | const float *codebook3 = lsp_cbjvm[2].cb; | 812 | const float *codebook3 = lsp_cbjmv[2].cb; |
| 1599 | 813 | ||
| 1600 | n1 = indexes[0]; | 814 | n1 = indexes[0]; |
| 1601 | n2 = indexes[1]; | 815 | n2 = indexes[1]; |
| 1602 | n3 = indexes[2]; | 816 | n3 = indexes[2]; |
| 1603 | 817 | ||
| 1604 | for (i=0;i<order;i++) { | 818 | for (i = 0; i < order; i++) { |
| 1605 | xq[i] = codebook1[order*n1+i]; | 819 | xq[i] = codebook1[order * n1 + i]; |
| 1606 | } | 820 | } |
| 1607 | 821 | ||
| 1608 | if (stages != 1) { | 822 | if (stages != 1) { |
| 1609 | for (i=0;i<order/2;i++) { | 823 | for (i = 0; i < order / 2; i++) { |
| 1610 | xq[2*i] += codebook2[order*n2/2+i]; | 824 | xq[2 * i] += codebook2[order * n2 / 2 + i]; |
| 1611 | xq[2*i+1] += codebook3[order*n3/2+i]; | 825 | xq[2 * i + 1] += codebook3[order * n3 / 2 + i]; |
| 1612 | } | 826 | } |
| 1613 | } | 827 | } |
| 1614 | |||
| 1615 | } | 828 | } |
| 1616 | 829 | ||
| 1617 | |||
| 1618 | /*---------------------------------------------------------------------------*\ | 830 | /*---------------------------------------------------------------------------*\ |
| 1619 | 831 | ||
| 1620 | FUNCTION....: bw_expand_lsps() | 832 | FUNCTION....: bw_expand_lsps() |
| @@ -1628,124 +840,45 @@ void decode_lsps_vq(int *indexes, float *xq, int order, int stages) | |||
| 1628 | 840 | ||
| 1629 | \*---------------------------------------------------------------------------*/ | 841 | \*---------------------------------------------------------------------------*/ |
| 1630 | 842 | ||
| 1631 | void bw_expand_lsps(float lsp[], int order, float min_sep_low, float min_sep_high) | 843 | void bw_expand_lsps(float lsp[], int order, float min_sep_low, |
| 1632 | { | 844 | float min_sep_high) { |
| 1633 | int i; | 845 | int i; |
| 1634 | |||
| 1635 | for(i=1; i<4; i++) { | ||
| 1636 | |||
| 1637 | if ((lsp[i] - lsp[i-1]) < min_sep_low*(PI/4000.0)) | ||
| 1638 | lsp[i] = lsp[i-1] + min_sep_low*(PI/4000.0); | ||
| 1639 | |||
| 1640 | } | ||
| 1641 | |||
| 1642 | /* As quantiser gaps increased, larger BW expansion was required | ||
| 1643 | to prevent twinkly noises. This may need more experiment for | ||
| 1644 | different quanstisers. | ||
| 1645 | */ | ||
| 1646 | |||
| 1647 | for(i=4; i<order; i++) { | ||
| 1648 | if (lsp[i] - lsp[i-1] < min_sep_high*(PI/4000.0)) | ||
| 1649 | lsp[i] = lsp[i-1] + min_sep_high*(PI/4000.0); | ||
| 1650 | } | ||
| 1651 | } | ||
| 1652 | |||
| 1653 | void bw_expand_lsps2(float lsp[], | ||
| 1654 | int order | ||
| 1655 | ) | ||
| 1656 | { | ||
| 1657 | int i; | ||
| 1658 | |||
| 1659 | for(i=1; i<4; i++) { | ||
| 1660 | |||
| 1661 | if ((lsp[i] - lsp[i-1]) < 100.0*(PI/4000.0)) | ||
| 1662 | lsp[i] = lsp[i-1] + 100.0*(PI/4000.0); | ||
| 1663 | 846 | ||
| 1664 | } | 847 | for (i = 1; i < 4; i++) { |
| 848 | if ((lsp[i] - lsp[i - 1]) < min_sep_low * (PI / 4000.0)) | ||
| 849 | lsp[i] = lsp[i - 1] + min_sep_low * (PI / 4000.0); | ||
| 850 | } | ||
| 1665 | 851 | ||
| 1666 | /* As quantiser gaps increased, larger BW expansion was required | 852 | /* As quantiser gaps increased, larger BW expansion was required |
| 1667 | to prevent twinkly noises. This may need more experiment for | 853 | to prevent twinkly noises. This may need more experiment for |
| 1668 | different quanstisers. | 854 | different quanstisers. |
| 1669 | */ | 855 | */ |
| 1670 | 856 | ||
| 1671 | for(i=4; i<order; i++) { | 857 | for (i = 4; i < order; i++) { |
| 1672 | if (lsp[i] - lsp[i-1] < 200.0*(PI/4000.0)) | 858 | if (lsp[i] - lsp[i - 1] < min_sep_high * (PI / 4000.0)) |
| 1673 | lsp[i] = lsp[i-1] + 200.0*(PI/4000.0); | 859 | lsp[i] = lsp[i - 1] + min_sep_high * (PI / 4000.0); |
| 1674 | } | 860 | } |
| 1675 | } | 861 | } |
| 1676 | 862 | ||
| 1677 | /*---------------------------------------------------------------------------*\ | 863 | void bw_expand_lsps2(float lsp[], int order) { |
| 1678 | 864 | int i; | |
| 1679 | FUNCTION....: locate_lsps_jnd_steps() | ||
| 1680 | AUTHOR......: David Rowe | ||
| 1681 | DATE CREATED: 27/10/2011 | ||
| 1682 | |||
| 1683 | Applies a form of Bandwidth Expansion (BW) to a vector of LSPs. | ||
| 1684 | Listening tests have determined that "quantising" the position of | ||
| 1685 | each LSP to the non-linear steps below introduces a "just noticable | ||
| 1686 | difference" in the synthesised speech. | ||
| 1687 | |||
| 1688 | This operation can be used before quantisation to limit the input | ||
| 1689 | data to the quantiser to a number of discrete steps. | ||
| 1690 | |||
| 1691 | This operation can also be used during quantisation as a form of | ||
| 1692 | hysteresis in the calculation of quantiser error. For example if | ||
| 1693 | the quantiser target of lsp1 is 500 Hz, candidate vectors with lsp1 | ||
| 1694 | of 515 and 495 Hz sound effectively the same. | ||
| 1695 | |||
| 1696 | \*---------------------------------------------------------------------------*/ | ||
| 1697 | |||
| 1698 | void locate_lsps_jnd_steps(float lsps[], int order) | ||
| 1699 | { | ||
| 1700 | int i; | ||
| 1701 | float lsp_hz, step; | ||
| 1702 | |||
| 1703 | assert(order == 10); | ||
| 1704 | |||
| 1705 | /* quantise to 25Hz steps */ | ||
| 1706 | |||
| 1707 | step = 25; | ||
| 1708 | for(i=0; i<2; i++) { | ||
| 1709 | lsp_hz = lsps[i]*4000.0/PI; | ||
| 1710 | lsp_hz = floorf(lsp_hz/step + 0.5)*step; | ||
| 1711 | lsps[i] = lsp_hz*PI/4000.0; | ||
| 1712 | if (i) { | ||
| 1713 | if (lsps[i] == lsps[i-1]) | ||
| 1714 | lsps[i] += step*PI/4000.0; | ||
| 1715 | |||
| 1716 | } | ||
| 1717 | } | ||
| 1718 | |||
| 1719 | /* quantise to 50Hz steps */ | ||
| 1720 | |||
| 1721 | step = 50; | ||
| 1722 | for(i=2; i<4; i++) { | ||
| 1723 | lsp_hz = lsps[i]*4000.0/PI; | ||
| 1724 | lsp_hz = floorf(lsp_hz/step + 0.5)*step; | ||
| 1725 | lsps[i] = lsp_hz*PI/4000.0; | ||
| 1726 | if (i) { | ||
| 1727 | if (lsps[i] == lsps[i-1]) | ||
| 1728 | lsps[i] += step*PI/4000.0; | ||
| 1729 | |||
| 1730 | } | ||
| 1731 | } | ||
| 1732 | 865 | ||
| 1733 | /* quantise to 100Hz steps */ | 866 | for (i = 1; i < 4; i++) { |
| 867 | if ((lsp[i] - lsp[i - 1]) < 100.0 * (PI / 4000.0)) | ||
| 868 | lsp[i] = lsp[i - 1] + 100.0 * (PI / 4000.0); | ||
| 869 | } | ||
| 1734 | 870 | ||
| 1735 | step = 100; | 871 | /* As quantiser gaps increased, larger BW expansion was required |
| 1736 | for(i=4; i<10; i++) { | 872 | to prevent twinkly noises. This may need more experiment for |
| 1737 | lsp_hz = lsps[i]*4000.0/PI; | 873 | different quanstisers. |
| 1738 | lsp_hz = floorf(lsp_hz/step + 0.5)*step; | 874 | */ |
| 1739 | lsps[i] = lsp_hz*PI/4000.0; | ||
| 1740 | if (i) { | ||
| 1741 | if (lsps[i] == lsps[i-1]) | ||
| 1742 | lsps[i] += step*PI/4000.0; | ||
| 1743 | 875 | ||
| 1744 | } | 876 | for (i = 4; i < order; i++) { |
| 1745 | } | 877 | if (lsp[i] - lsp[i - 1] < 200.0 * (PI / 4000.0)) |
| 878 | lsp[i] = lsp[i - 1] + 200.0 * (PI / 4000.0); | ||
| 879 | } | ||
| 1746 | } | 880 | } |
| 1747 | 881 | ||
| 1748 | |||
| 1749 | /*---------------------------------------------------------------------------*\ | 882 | /*---------------------------------------------------------------------------*\ |
| 1750 | 883 | ||
| 1751 | FUNCTION....: apply_lpc_correction() | 884 | FUNCTION....: apply_lpc_correction() |
| @@ -1757,11 +890,10 @@ void locate_lsps_jnd_steps(float lsps[], int order) | |||
| 1757 | 890 | ||
| 1758 | \*---------------------------------------------------------------------------*/ | 891 | \*---------------------------------------------------------------------------*/ |
| 1759 | 892 | ||
| 1760 | void apply_lpc_correction(MODEL *model) | 893 | void apply_lpc_correction(MODEL *model) { |
| 1761 | { | 894 | if (model->Wo < (PI * 150.0 / 4000)) { |
| 1762 | if (model->Wo < (PI*150.0/4000)) { | 895 | model->A[1] *= 0.032; |
| 1763 | model->A[1] *= 0.032; | 896 | } |
| 1764 | } | ||
| 1765 | } | 897 | } |
| 1766 | 898 | ||
| 1767 | /*---------------------------------------------------------------------------*\ | 899 | /*---------------------------------------------------------------------------*\ |
| @@ -1774,124 +906,81 @@ void apply_lpc_correction(MODEL *model) | |||
| 1774 | 906 | ||
| 1775 | \*---------------------------------------------------------------------------*/ | 907 | \*---------------------------------------------------------------------------*/ |
| 1776 | 908 | ||
| 1777 | int encode_energy(float e, int bits) | 909 | int encode_energy(float e, int bits) { |
| 1778 | { | 910 | int index, e_levels = 1 << bits; |
| 1779 | int index, e_levels = 1<<bits; | 911 | float e_min = E_MIN_DB; |
| 1780 | float e_min = E_MIN_DB; | 912 | float e_max = E_MAX_DB; |
| 1781 | float e_max = E_MAX_DB; | 913 | float norm; |
| 1782 | float norm; | ||
| 1783 | |||
| 1784 | e = 10.0*log10f(e); | ||
| 1785 | norm = (e - e_min)/(e_max - e_min); | ||
| 1786 | index = floorf(e_levels * norm + 0.5); | ||
| 1787 | if (index < 0 ) index = 0; | ||
| 1788 | if (index > (e_levels-1)) index = e_levels-1; | ||
| 1789 | |||
| 1790 | return index; | ||
| 1791 | } | ||
| 1792 | |||
| 1793 | /*---------------------------------------------------------------------------*\ | ||
| 1794 | |||
| 1795 | FUNCTION....: decode_energy() | ||
| 1796 | AUTHOR......: David Rowe | ||
| 1797 | DATE CREATED: 22/8/2010 | ||
| 1798 | |||
| 1799 | Decodes energy using a E_LEVELS quantiser. | ||
| 1800 | |||
| 1801 | \*---------------------------------------------------------------------------*/ | ||
| 1802 | |||
| 1803 | float decode_energy(int index, int bits) | ||
| 1804 | { | ||
| 1805 | float e_min = E_MIN_DB; | ||
| 1806 | float e_max = E_MAX_DB; | ||
| 1807 | float step; | ||
| 1808 | float e; | ||
| 1809 | int e_levels = 1<<bits; | ||
| 1810 | 914 | ||
| 1811 | step = (e_max - e_min)/e_levels; | 915 | e = 10.0 * log10f(e); |
| 1812 | e = e_min + step*(index); | 916 | norm = (e - e_min) / (e_max - e_min); |
| 1813 | e = POW10F(e/10.0); | 917 | index = floorf(e_levels * norm + 0.5); |
| 918 | if (index < 0) index = 0; | ||
| 919 | if (index > (e_levels - 1)) index = e_levels - 1; | ||
| 1814 | 920 | ||
| 1815 | return e; | 921 | return index; |
| 1816 | } | 922 | } |
| 1817 | 923 | ||
| 1818 | #ifdef NOT_USED | ||
| 1819 | /*---------------------------------------------------------------------------*\ | 924 | /*---------------------------------------------------------------------------*\ |
| 1820 | 925 | ||
| 1821 | FUNCTION....: decode_amplitudes() | 926 | FUNCTION....: decode_energy() |
| 1822 | AUTHOR......: David Rowe | 927 | AUTHOR......: David Rowe |
| 1823 | DATE CREATED: 22/8/2010 | 928 | DATE CREATED: 22/8/2010 |
| 1824 | 929 | ||
| 1825 | Given the amplitude quantiser indexes recovers the harmonic | 930 | Decodes energy using a E_LEVELS quantiser. |
| 1826 | amplitudes. | ||
| 1827 | 931 | ||
| 1828 | \*---------------------------------------------------------------------------*/ | 932 | \*---------------------------------------------------------------------------*/ |
| 1829 | 933 | ||
| 1830 | float decode_amplitudes(codec2_fft_cfg fft_fwd_cfg, | 934 | float decode_energy(int index, int bits) { |
| 1831 | MODEL *model, | 935 | float e_min = E_MIN_DB; |
| 1832 | float ak[], | 936 | float e_max = E_MAX_DB; |
| 1833 | int lsp_indexes[], | 937 | float step; |
| 1834 | int energy_index, | 938 | float e; |
| 1835 | float lsps[], | 939 | int e_levels = 1 << bits; |
| 1836 | float *e | ||
| 1837 | ) | ||
| 1838 | { | ||
| 1839 | float snr; | ||
| 1840 | 940 | ||
| 1841 | decode_lsps_scalar(lsps, lsp_indexes, LPC_ORD); | 941 | step = (e_max - e_min) / e_levels; |
| 1842 | bw_expand_lsps(lsps, LPC_ORD); | 942 | e = e_min + step * (index); |
| 1843 | lsp_to_lpc(lsps, ak, LPC_ORD); | 943 | e = POW10F(e / 10.0); |
| 1844 | *e = decode_energy(energy_index); | ||
| 1845 | aks_to_M2(ak, LPC_ORD, model, *e, &snr, 1, 0, 0, 1); | ||
| 1846 | apply_lpc_correction(model); | ||
| 1847 | 944 | ||
| 1848 | return snr; | 945 | return e; |
| 1849 | } | 946 | } |
| 1850 | #endif | ||
| 1851 | 947 | ||
| 1852 | static float ge_coeff[2] = {0.8, 0.9}; | 948 | static float ge_coeff[2] = {0.8, 0.9}; |
| 1853 | 949 | ||
| 1854 | void compute_weights2(const float *x, const float *xp, float *w) | 950 | void compute_weights2(const float *x, const float *xp, float *w) { |
| 1855 | { | ||
| 1856 | w[0] = 30; | 951 | w[0] = 30; |
| 1857 | w[1] = 1; | 952 | w[1] = 1; |
| 1858 | if (x[1]<0) | 953 | if (x[1] < 0) { |
| 1859 | { | 954 | w[0] *= .6; |
| 1860 | w[0] *= .6; | 955 | w[1] *= .3; |
| 1861 | w[1] *= .3; | ||
| 1862 | } | 956 | } |
| 1863 | if (x[1]<-10) | 957 | if (x[1] < -10) { |
| 1864 | { | 958 | w[0] *= .3; |
| 1865 | w[0] *= .3; | 959 | w[1] *= .3; |
| 1866 | w[1] *= .3; | ||
| 1867 | } | 960 | } |
| 1868 | /* Higher weight if pitch is stable */ | 961 | /* Higher weight if pitch is stable */ |
| 1869 | if (fabsf(x[0]-xp[0])<.2) | 962 | if (fabsf(x[0] - xp[0]) < .2) { |
| 1870 | { | 963 | w[0] *= 2; |
| 1871 | w[0] *= 2; | 964 | w[1] *= 1.5; |
| 1872 | w[1] *= 1.5; | 965 | } else if (fabsf(x[0] - xp[0]) > .5) /* Lower if not stable */ |
| 1873 | } else if (fabsf(x[0]-xp[0])>.5) /* Lower if not stable */ | ||
| 1874 | { | 966 | { |
| 1875 | w[0] *= .5; | 967 | w[0] *= .5; |
| 1876 | } | 968 | } |
| 1877 | 969 | ||
| 1878 | /* Lower weight for low energy */ | 970 | /* Lower weight for low energy */ |
| 1879 | if (x[1] < xp[1]-10) | 971 | if (x[1] < xp[1] - 10) { |
| 1880 | { | 972 | w[1] *= .5; |
| 1881 | w[1] *= .5; | ||
| 1882 | } | 973 | } |
| 1883 | if (x[1] < xp[1]-20) | 974 | if (x[1] < xp[1] - 20) { |
| 1884 | { | 975 | w[1] *= .5; |
| 1885 | w[1] *= .5; | ||
| 1886 | } | 976 | } |
| 1887 | 977 | ||
| 1888 | //w[0] = 30; | 978 | // w[0] = 30; |
| 1889 | //w[1] = 1; | 979 | // w[1] = 1; |
| 1890 | 980 | ||
| 1891 | /* Square the weights because it's applied on the squared error */ | 981 | /* Square the weights because it's applied on the squared error */ |
| 1892 | w[0] *= w[0]; | 982 | w[0] *= w[0]; |
| 1893 | w[1] *= w[1]; | 983 | w[1] *= w[1]; |
| 1894 | |||
| 1895 | } | 984 | } |
| 1896 | 985 | ||
| 1897 | /*---------------------------------------------------------------------------*\ | 986 | /*---------------------------------------------------------------------------*\ |
| @@ -1906,7 +995,7 @@ void compute_weights2(const float *x, const float *xp, float *w) | |||
| 1906 | both the pitch and energy tend to only change by small amounts | 995 | both the pitch and energy tend to only change by small amounts |
| 1907 | during voiced speech, however it is important that these changes be | 996 | during voiced speech, however it is important that these changes be |
| 1908 | coded carefully. During unvoiced speech they both change a lot but | 997 | coded carefully. During unvoiced speech they both change a lot but |
| 1909 | the ear is less sensitve to errors so coarser quantisation is OK. | 998 | the ear is less sensitive to errors so coarser quantisation is OK. |
| 1910 | 999 | ||
| 1911 | The ear is sensitive to log energy and loq pitch so we quantise in | 1000 | The ear is sensitive to log energy and loq pitch so we quantise in |
| 1912 | these domains. That way the error measure used to quantise the | 1001 | these domains. That way the error measure used to quantise the |
| @@ -1916,15 +1005,14 @@ void compute_weights2(const float *x, const float *xp, float *w) | |||
| 1916 | 1005 | ||
| 1917 | \*---------------------------------------------------------------------------*/ | 1006 | \*---------------------------------------------------------------------------*/ |
| 1918 | 1007 | ||
| 1919 | void quantise_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[]) | 1008 | void quantise_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[]) { |
| 1920 | { | 1009 | int i, n1; |
| 1921 | int i, n1; | 1010 | float x[2]; |
| 1922 | float x[2]; | 1011 | float err[2]; |
| 1923 | float err[2]; | 1012 | float w[2]; |
| 1924 | float w[2]; | ||
| 1925 | const float *codebook1 = ge_cb[0].cb; | 1013 | const float *codebook1 = ge_cb[0].cb; |
| 1926 | int nb_entries = ge_cb[0].m; | 1014 | int nb_entries = ge_cb[0].m; |
| 1927 | int ndim = ge_cb[0].k; | 1015 | int ndim = ge_cb[0].k; |
| 1928 | float Wo_min = c2const->Wo_min; | 1016 | float Wo_min = c2const->Wo_min; |
| 1929 | float Wo_max = c2const->Wo_max; | 1017 | float Wo_max = c2const->Wo_max; |
| 1930 | float Fs = c2const->Fs; | 1018 | float Fs = c2const->Fs; |
| @@ -1933,18 +1021,16 @@ void quantise_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[]) | |||
| 1933 | 1021 | ||
| 1934 | assert(Fs == 8000); | 1022 | assert(Fs == 8000); |
| 1935 | 1023 | ||
| 1936 | x[0] = log10f((model->Wo/PI)*4000.0/50.0)/log10f(2); | 1024 | x[0] = log10f((model->Wo / PI) * 4000.0 / 50.0) / log10f(2); |
| 1937 | x[1] = 10.0*log10f(1e-4 + *e); | 1025 | x[1] = 10.0 * log10f(1e-4 + *e); |
| 1938 | 1026 | ||
| 1939 | compute_weights2(x, xq, w); | 1027 | compute_weights2(x, xq, w); |
| 1940 | for (i=0;i<ndim;i++) | 1028 | for (i = 0; i < ndim; i++) err[i] = x[i] - ge_coeff[i] * xq[i]; |
| 1941 | err[i] = x[i]-ge_coeff[i]*xq[i]; | ||
| 1942 | n1 = find_nearest_weighted(codebook1, nb_entries, err, w, ndim); | 1029 | n1 = find_nearest_weighted(codebook1, nb_entries, err, w, ndim); |
| 1943 | 1030 | ||
| 1944 | for (i=0;i<ndim;i++) | 1031 | for (i = 0; i < ndim; i++) { |
| 1945 | { | 1032 | xq[i] = ge_coeff[i] * xq[i] + codebook1[ndim * n1 + i]; |
| 1946 | xq[i] = ge_coeff[i]*xq[i] + codebook1[ndim*n1+i]; | 1033 | err[i] -= codebook1[ndim * n1 + i]; |
| 1947 | err[i] -= codebook1[ndim*n1+i]; | ||
| 1948 | } | 1034 | } |
| 1949 | 1035 | ||
| 1950 | /* | 1036 | /* |
| @@ -1953,7 +1039,7 @@ void quantise_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[]) | |||
| 1953 | Wo = (2^x)*(PI*50)/4000; | 1039 | Wo = (2^x)*(PI*50)/4000; |
| 1954 | */ | 1040 | */ |
| 1955 | 1041 | ||
| 1956 | model->Wo = powf(2.0, xq[0])*(PI*50.0)/4000.0; | 1042 | model->Wo = powf(2.0, xq[0]) * (PI * 50.0) / 4000.0; |
| 1957 | 1043 | ||
| 1958 | /* bit errors can make us go out of range leading to all sorts of | 1044 | /* bit errors can make us go out of range leading to all sorts of |
| 1959 | probs like seg faults */ | 1045 | probs like seg faults */ |
| @@ -1961,9 +1047,9 @@ void quantise_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[]) | |||
| 1961 | if (model->Wo > Wo_max) model->Wo = Wo_max; | 1047 | if (model->Wo > Wo_max) model->Wo = Wo_max; |
| 1962 | if (model->Wo < Wo_min) model->Wo = Wo_min; | 1048 | if (model->Wo < Wo_min) model->Wo = Wo_min; |
| 1963 | 1049 | ||
| 1964 | model->L = PI/model->Wo; /* if we quantise Wo re-compute L */ | 1050 | model->L = PI / model->Wo; /* if we quantise Wo re-compute L */ |
| 1965 | 1051 | ||
| 1966 | *e = POW10F(xq[1]/10.0); | 1052 | *e = POW10F(xq[1] / 10.0); |
| 1967 | } | 1053 | } |
| 1968 | 1054 | ||
| 1969 | /*---------------------------------------------------------------------------*\ | 1055 | /*---------------------------------------------------------------------------*\ |
| @@ -1977,39 +1063,36 @@ void quantise_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[]) | |||
| 1977 | 1063 | ||
| 1978 | \*---------------------------------------------------------------------------*/ | 1064 | \*---------------------------------------------------------------------------*/ |
| 1979 | 1065 | ||
| 1980 | int encode_WoE(MODEL *model, float e, float xq[]) | 1066 | int encode_WoE(MODEL *model, float e, float xq[]) { |
| 1981 | { | 1067 | int i, n1; |
| 1982 | int i, n1; | 1068 | float x[2]; |
| 1983 | float x[2]; | 1069 | float err[2]; |
| 1984 | float err[2]; | 1070 | float w[2]; |
| 1985 | float w[2]; | ||
| 1986 | const float *codebook1 = ge_cb[0].cb; | 1071 | const float *codebook1 = ge_cb[0].cb; |
| 1987 | int nb_entries = ge_cb[0].m; | 1072 | int nb_entries = ge_cb[0].m; |
| 1988 | int ndim = ge_cb[0].k; | 1073 | int ndim = ge_cb[0].k; |
| 1989 | 1074 | ||
| 1990 | assert((1<<WO_E_BITS) == nb_entries); | 1075 | assert((1 << WO_E_BITS) == nb_entries); |
| 1991 | 1076 | ||
| 1992 | if (e < 0.0) e = 0; /* occasional small negative energies due LPC round off I guess */ | 1077 | if (e < 0.0) |
| 1078 | e = 0; /* occasional small negative energies due LPC round off I guess */ | ||
| 1993 | 1079 | ||
| 1994 | x[0] = log10f((model->Wo/PI)*4000.0/50.0)/log10f(2); | 1080 | x[0] = log10f((model->Wo / PI) * 4000.0 / 50.0) / log10f(2); |
| 1995 | x[1] = 10.0*log10f(1e-4 + e); | 1081 | x[1] = 10.0 * log10f(1e-4 + e); |
| 1996 | 1082 | ||
| 1997 | compute_weights2(x, xq, w); | 1083 | compute_weights2(x, xq, w); |
| 1998 | for (i=0;i<ndim;i++) | 1084 | for (i = 0; i < ndim; i++) err[i] = x[i] - ge_coeff[i] * xq[i]; |
| 1999 | err[i] = x[i]-ge_coeff[i]*xq[i]; | ||
| 2000 | n1 = find_nearest_weighted(codebook1, nb_entries, err, w, ndim); | 1085 | n1 = find_nearest_weighted(codebook1, nb_entries, err, w, ndim); |
| 2001 | 1086 | ||
| 2002 | for (i=0;i<ndim;i++) | 1087 | for (i = 0; i < ndim; i++) { |
| 2003 | { | 1088 | xq[i] = ge_coeff[i] * xq[i] + codebook1[ndim * n1 + i]; |
| 2004 | xq[i] = ge_coeff[i]*xq[i] + codebook1[ndim*n1+i]; | 1089 | err[i] -= codebook1[ndim * n1 + i]; |
| 2005 | err[i] -= codebook1[ndim*n1+i]; | ||
| 2006 | } | 1090 | } |
| 2007 | 1091 | ||
| 2008 | //printf("enc: %f %f (%f)(%f) \n", xq[0], xq[1], e, 10.0*log10(1e-4 + e)); | 1092 | // printf("enc: %f %f (%f)(%f) \n", xq[0], xq[1], e, 10.0*log10(1e-4 + e)); |
| 2009 | return n1; | 1093 | return n1; |
| 2010 | } | 1094 | } |
| 2011 | 1095 | ||
| 2012 | |||
| 2013 | /*---------------------------------------------------------------------------*\ | 1096 | /*---------------------------------------------------------------------------*\ |
| 2014 | 1097 | ||
| 2015 | FUNCTION....: decode_WoE() | 1098 | FUNCTION....: decode_WoE() |
| @@ -2022,21 +1105,19 @@ int encode_WoE(MODEL *model, float e, float xq[]) | |||
| 2022 | 1105 | ||
| 2023 | \*---------------------------------------------------------------------------*/ | 1106 | \*---------------------------------------------------------------------------*/ |
| 2024 | 1107 | ||
| 2025 | void decode_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[], int n1) | 1108 | void decode_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[], int n1) { |
| 2026 | { | 1109 | int i; |
| 2027 | int i; | ||
| 2028 | const float *codebook1 = ge_cb[0].cb; | 1110 | const float *codebook1 = ge_cb[0].cb; |
| 2029 | int ndim = ge_cb[0].k; | 1111 | int ndim = ge_cb[0].k; |
| 2030 | float Wo_min = c2const->Wo_min; | 1112 | float Wo_min = c2const->Wo_min; |
| 2031 | float Wo_max = c2const->Wo_max; | 1113 | float Wo_max = c2const->Wo_max; |
| 2032 | 1114 | ||
| 2033 | for (i=0;i<ndim;i++) | 1115 | for (i = 0; i < ndim; i++) { |
| 2034 | { | 1116 | xq[i] = ge_coeff[i] * xq[i] + codebook1[ndim * n1 + i]; |
| 2035 | xq[i] = ge_coeff[i]*xq[i] + codebook1[ndim*n1+i]; | ||
| 2036 | } | 1117 | } |
| 2037 | 1118 | ||
| 2038 | //printf("dec: %f %f\n", xq[0], xq[1]); | 1119 | // printf("dec: %f %f\n", xq[0], xq[1]); |
| 2039 | model->Wo = powf(2.0, xq[0])*(PI*50.0)/4000.0; | 1120 | model->Wo = powf(2.0, xq[0]) * (PI * 50.0) / 4000.0; |
| 2040 | 1121 | ||
| 2041 | /* bit errors can make us go out of range leading to all sorts of | 1122 | /* bit errors can make us go out of range leading to all sorts of |
| 2042 | probs like seg faults */ | 1123 | probs like seg faults */ |
| @@ -2044,8 +1125,7 @@ void decode_WoE(C2CONST *c2const, MODEL *model, float *e, float xq[], int n1) | |||
| 2044 | if (model->Wo > Wo_max) model->Wo = Wo_max; | 1125 | if (model->Wo > Wo_max) model->Wo = Wo_max; |
| 2045 | if (model->Wo < Wo_min) model->Wo = Wo_min; | 1126 | if (model->Wo < Wo_min) model->Wo = Wo_min; |
| 2046 | 1127 | ||
| 2047 | model->L = PI/model->Wo; /* if we quantise Wo re-compute L */ | 1128 | model->L = PI / model->Wo; /* if we quantise Wo re-compute L */ |
| 2048 | 1129 | ||
| 2049 | *e = POW10F(xq[1]/10.0); | 1130 | *e = POW10F(xq[1] / 10.0); |
| 2050 | } | 1131 | } |
| 2051 | |||
