File size: 60,640 Bytes
6fa4bc9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 |
{
"paper_id": "O12-2002",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T08:03:04.368130Z"
},
"title": "Variable Speech Rate Mandarin Chinese Text-to-Speech System",
"authors": [
{
"first": "",
"middle": [],
"last": "\u6c5f\u632f\u5b87",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "National Chiao Tung University",
"location": {
"settlement": "Hsinchu",
"country": "Taiwan"
}
},
"email": ""
},
{
"first": "#",
"middle": [],
"last": "\u3001\u9ec3\u555f\u5168",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "National Chiao Tung University",
"location": {
"settlement": "Hsinchu",
"country": "Taiwan"
}
},
"email": ""
},
{
"first": "\u3001\u4f59\u79c0\u654f",
"middle": [
"+"
],
"last": "\u3001\u9673\u4fe1\u5b8f",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "National Chiao Tung University",
"location": {
"settlement": "Hsinchu",
"country": "Taiwan"
}
},
"email": ""
},
{
"first": "Chen-Yu",
"middle": [],
"last": "Chiang",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "National Chiao Tung University",
"location": {
"settlement": "Hsinchu",
"country": "Taiwan"
}
},
"email": "cychiang@mail.ntpu.edu.tw"
},
{
"first": "Qi-Quan",
"middle": [],
"last": "Huang",
"suffix": "",
"affiliation": {},
"email": ""
},
{
"first": "Yih-Ru",
"middle": [],
"last": "Wang",
"suffix": "",
"affiliation": {},
"email": "yrwang@cc.nctu.edu.tw"
},
{
"first": "Hsiu-Min",
"middle": [],
"last": "Yu",
"suffix": "",
"affiliation": {},
"email": ""
},
{
"first": "Horng",
"middle": [],
"last": "Chen",
"suffix": "",
"affiliation": {},
"email": "schen@mail.nctu.edu.tw"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "This paper presents an Hidden Markov Model (HMM)-based variable speech rate Mandarin Chinese text-to-speech (TTS) system. In this system, parameters of spectrum, fundametal frequency and state duration are generated by a context dependent HMM (CDHMM) whose model parameters are linear-interpolated from those of three CDHMMs trained by corpora in three different speech rates (SRs), i.e. fast, medium and slow. In addition, three decision tree (DT)-based pause break predictors trained by using the three SR corpora are used to interpolate the probabilities for inserting pause breaks. The performance of the proposed TTS system were evaluated by several objective and subjective tests. Experimental results suggested that coherence between interpolation weights for CDHMMs and DT-based pasue predictors is crutial for naturalness of the synthesis speech in variable SR. We believe that the proposed variable speech rate Mandarin Chinese TTS system is more suitable than conventional fixed SR TTS systems for applications of human-machine interaction.",
"pdf_parse": {
"paper_id": "O12-2002",
"_pdf_hash": "",
"abstract": [
{
"text": "This paper presents an Hidden Markov Model (HMM)-based variable speech rate Mandarin Chinese text-to-speech (TTS) system. In this system, parameters of spectrum, fundametal frequency and state duration are generated by a context dependent HMM (CDHMM) whose model parameters are linear-interpolated from those of three CDHMMs trained by corpora in three different speech rates (SRs), i.e. fast, medium and slow. In addition, three decision tree (DT)-based pause break predictors trained by using the three SR corpora are used to interpolate the probabilities for inserting pause breaks. The performance of the proposed TTS system were evaluated by several objective and subjective tests. Experimental results suggested that coherence between interpolation weights for CDHMMs and DT-based pasue predictors is crutial for naturalness of the synthesis speech in variable SR. We believe that the proposed variable speech rate Mandarin Chinese TTS system is more suitable than conventional fixed SR TTS systems for applications of human-machine interaction.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "\u53ca\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b(HMM-based approach) (Tokuda et al., 2000 (Imai, 1983 ) \u8f38\u51fa\u5408\u6210\u51fa\u8a9e\u97f3\u8a0a\u865f\u3002 \u7576\u60f3\u8981\u4ee5\u73fe\u6709\u6a21\u578b\u53bb\u5408\u6210\u51fa\uf967\u540c\u7279\u6027\u7684\u8a9e\u97f3\u8a0a\u865f\uff0c\u5247\u53ef\uf9dd\u7528\u8abf\u6574\uf96b\uf969\u7684\u65b9\u5f0f\u9054\u5230\u76ee \u7684\uff0c\u5982\u5167\u63d2(interpolation methods) (Yoshimura et al., 2000) \u3001\u8abf\u9069(adaptation methods) (Tamura et al., 2001 (Yu et al., 2007) \u7684\u65b9\u6cd5\u63d0\u4f9b\uf9ba bottom-up \u7684\u65b9\u5f0f\u5206\u6790\uff0c\u50c5\u5f9e\u97f3\u7bc0\u5c64\u6b21\u8a0e\uf941\u97f3\u9ad8\u8ecc\u8de1\u6703\u5ffd\uf976\u5230\u97fb \uf9d8\u7d50\u69cb\u4e0a\u5c64\u7684\u5f71\u97ff\uff1b\u81f3\u65bc (Li & Zu, 2008) \u548c (Tseng, 2008 ) \u7684\u968e\u5c64\u5f0f\u97fb\uf9d8\u67b6\u69cb\u5247\u63d0\u4f9b\u4e00 \u500b top-down \u7684\u5206\u6790\u65b9\u5f0f\uff0c\u5c0d\u65bc\u5e95\u5c64\u4e4b\u97f3\u7bc0\u5c64\u6b21\u5206\u6790\u8f03\u7f3a\u4e4f\uff0c\u6b64\u5916\uff0c\u50b3\u7d71\u97fb\uf9d8\u968e\u5c64\u7684\u7814\u7a76 \u90fd\u9700\u8981\u4eba\u5de5\u4e8b\u5148\u6a19\u8a18\u97fb\uf9d8\u908a\u754c\uff0c\u56e0\u6b64\uff0c\u5728\u6587\u737b (Chiang et al., 2009 (Yoshimura et al., 2000; Iwano et al., 2002) ",
"cite_spans": [
{
"start": 30,
"end": 50,
"text": "(Tokuda et al., 2000",
"ref_id": "BIBREF12"
},
{
"start": 51,
"end": 62,
"text": "(Imai, 1983",
"ref_id": "BIBREF4"
},
{
"start": 142,
"end": 166,
"text": "(Yoshimura et al., 2000)",
"ref_id": "BIBREF15"
},
{
"start": 191,
"end": 211,
"text": "(Tamura et al., 2001",
"ref_id": "BIBREF13"
},
{
"start": 212,
"end": 229,
"text": "(Yu et al., 2007)",
"ref_id": "BIBREF17"
},
{
"start": 283,
"end": 298,
"text": "(Li & Zu, 2008)",
"ref_id": "BIBREF7"
},
{
"start": 301,
"end": 313,
"text": "(Tseng, 2008",
"ref_id": "BIBREF14"
},
{
"start": 395,
"end": 415,
"text": "(Chiang et al., 2009",
"ref_id": "BIBREF1"
},
{
"start": 416,
"end": 440,
"text": "(Yoshimura et al., 2000;",
"ref_id": "BIBREF15"
},
{
"start": 441,
"end": 460,
"text": "Iwano et al., 2002)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "= = \u00d7 \u2211 (1) \u5176\u4e2d i \u70ba\u6c7a\u7b56\u6a39\u7684 index (i=1\uff1a\u6162\uff0ci=2\uff1a\u4e2d\uff0ci=3\uff1a\u5feb)\uff1b i w \u70ba\u7b2c i \u500b\u6c7a\u7b56\u6a39\u6a21\u578b\u7684\u6b0a \u91cd\u503c\uff1b { } n sp \u2208 \u975c\u97f3\u505c\u9813, \u975e\u975c\u97f3\u505c\u9813 \u70ba\u7b2c n \u500b\u8a5e\u5f8c\u9762\u7684\u975c\u97f3\u505c\u9813\u8207\u5426\uff1b n L \u70ba\u6587\u8108\u8a9e\u8a00 \u8cc7\u8a0a\uff1b ( | ) i n n P sp L \u70ba\u7d93\u7531\u6587\u8108\u8a9e\u8a00\u8cc7\u8a0a ( n L ) \u7d44\u6210\u6c7a\u7b56\u6a39\u554f\u984c\u96c6\u5f8c\uff0c\u7531\u7b2c i \u500b\u6c7a\u7b56\u6a39\u7d50 \u69cb\uf9e8\uff0c\u627e\u5c0b\u5230\u5c0d\u61c9\u4e4b\uf96e\u7bc0\u9ede\u4e0b (leaf node) \u975c\u97f3\u505c\u9813\u548c\u975e\u975c\u97f3\u505c\u9813\u7684\u6a5f\uf961\u503c\u3002 \u800c\u7b2c\u4e8c\u7d44\u6b0a\u91cd\u503c\u5f71\u97ff\u8457\u983b\u8b5c\u3001\u97f3\u9577\u53ca\u57fa\u983b\uff0c\u5728\u8a9e\uf9be\u5206\u6790\u4e2d\u767c\u73fe\u8a9e\u901f\u8d8a\u5feb\uf967\u50c5\u97f3\u9577\u8b8a \u77ed\uff0c\u57fa\u983b\u4e5f\u6703\u96a8\u8457\uf925\u9ad8\uff0c\u76f4\u63a5\u5f71\u97ff\u5230\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b\u7684\uf96b\uf969\uff0c\u5f88\u76f4\u89c0\u5730\uff0c\u7576\u8abf\u6574\u6b0a\u91cd \u503c\u8d8a\u9760\u8fd1\u5feb\u901f\u8a9e\u901f\u8a9e\u97f3\u4e4b\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b\u6642\uff0c\u76f8\u5c0d\u65bc\u50c5\u4f7f\u7528\u6162\u901f\u8a9e\u97f3\u4e4b\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b \u6a21\u578b\uff0c\u6bcf\u500b\u97f3\u7bc0\u7684\u97f3\u9577\u6703\u8b8a\u77ed\u4e14\u97f3\u983b\u6703\u63d0\u9ad8\u3002\u4ee5\uf967\u540c\u6b0a\u91cd\u503c\u5167\u5dee\u4e09\u7a2e\u8a9e\u901f\u4e4b\u6a21\u578b\uf96b\uf969\u65b9 \u6cd5\u5982\u4e0b\u5f0f",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "\uff1a 3 1 i i i a = = \u00d7 \u2211 \u03bc \u03bc (2) 3 2 1 i i i a = = \u00d7 \u2211 U U (3) \u5176\u4e2d i \u70ba CDHMM \u6a21\u578b\u7684 index(i=1\uff1a\u6162\uff0ci=2\uff1a\u4e2d\uff0ci=3\uff1a\u5feb)\uff1b i a \u70ba\u7b2c i \u500b CDHMM \u6a21 \u578b\u7684\u6b0a\u91cd\u503c\uff0c i \u03bc \u53ca i U \u5206\u5225\u70ba CDHMM",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "A Set of Corpus-Based Text to Speech Synthesis Technologies for Mandarin Chinese",
"authors": [
{
"first": "F.-C",
"middle": [],
"last": "Chou",
"suffix": ""
},
{
"first": "C.-Y",
"middle": [],
"last": "Tseng",
"suffix": ""
},
{
"first": "L.-S",
"middle": [],
"last": "Lee",
"suffix": ""
}
],
"year": 2002,
"venue": "IEEE Trans. on Speech and Audoio Processing",
"volume": "10",
"issue": "7",
"pages": "481--494",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chou, F.-C., Tseng, C.-Y., & Lee, L.-S. (2002). A Set of Corpus-Based Text to Speech Synthesis Technologies for Mandarin Chinese. IEEE Trans. on Speech and Audoio Processing, 10(7) , 481-494.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "An Investigation on the Mandarin Prosody of a Parallel Multi-Speaking Rate Speech Corpus",
"authors": [
{
"first": "C.-Y",
"middle": [],
"last": "Chiang",
"suffix": ""
},
{
"first": "C.-C",
"middle": [],
"last": "Tang",
"suffix": ""
},
{
"first": "H.-M",
"middle": [],
"last": "Yu",
"suffix": ""
},
{
"first": "Y.-R",
"middle": [],
"last": "Wang",
"suffix": ""
},
{
"first": "S.-H",
"middle": [],
"last": "Chen",
"suffix": ""
}
],
"year": 2009,
"venue": "Proc. of Oriental COCOSDA 2009",
"volume": "",
"issue": "",
"pages": "148--153",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chiang, C.-Y, Tang, C.-C., Yu, H.-M., Wang, Y.-R., & Chen, S.-H. (2009). An Investigation on the Mandarin Prosody of a Parallel Multi-Speaking Rate Speech Corpus. In Proc. of Oriental COCOSDA 2009, 148-153.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Sinica Treebank: Design criteria, annotation guidelines, and pn-line interface",
"authors": [
{
"first": "C.-R",
"middle": [],
"last": "Huang",
"suffix": ""
},
{
"first": "K.-J",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "F.-Y",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "Z.-M",
"middle": [],
"last": "Gao",
"suffix": ""
},
{
"first": "K.-Y",
"middle": [],
"last": "Chen",
"suffix": ""
}
],
"year": 2000,
"venue": "Proc. of the Second Chinese Language Processing Workshop",
"volume": "",
"issue": "",
"pages": "29--37",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Huang, C.-R., Chen, K.-J., Chen, F.-Y., Gao, Z.-M., & Chen, K.-Y. (2000). Sinica Treebank: Design criteria, annotation guidelines, and pn-line interface. In Proc. of the Second Chinese Language Processing Workshop 2000, 29-37.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "The HMM-based speech synthesis system version 2.0",
"authors": [
{
"first": "H",
"middle": [],
"last": "Zen",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Nose",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Yamagishi",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Sako",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Masuko",
"suffix": ""
},
{
"first": "A",
"middle": [
"W"
],
"last": "Black",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Tokuda",
"suffix": ""
}
],
"year": 2007,
"venue": "Proc. 6th ISCA Workshop Speech Synth",
"volume": "",
"issue": "",
"pages": "294--299",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zen, H., Nose, T., Yamagishi, J., Sako, S., Masuko, T., Black, A. W., & Tokuda, K. (2007) The HMM-based speech synthesis system version 2.0. In Proc. 6th ISCA Workshop Speech Synth., 294-299.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Cepstral analysis synthesis on the mel frequency scale",
"authors": [
{
"first": "S",
"middle": [],
"last": "Imai",
"suffix": ""
}
],
"year": 1983,
"venue": "Proc. of ICASSP",
"volume": "",
"issue": "",
"pages": "93--96",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Imai, S. (1983). Cepstral analysis synthesis on the mel frequency scale. In Proc. of ICASSP, 93-96.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Speech-rate-variable HMM-based Japanese TTS system",
"authors": [
{
"first": "K",
"middle": [],
"last": "Iwano",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Yamada",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Togawa",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Furui",
"suffix": ""
}
],
"year": 2002,
"venue": "Proc. of IEEE TTS Workshop",
"volume": "",
"issue": "",
"pages": "219--222",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Iwano, K., Yamada, M., Togawa, T., & Furui, S. (2002). Speech-rate-variable HMM-based Japanese TTS system. In Proc. of IEEE TTS Workshop 2002, 219-222.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Conditional Random Fields Based Label Sequence and Information Feedback",
"authors": [
{
"first": "W",
"middle": [],
"last": "Jiang",
"suffix": ""
},
{
"first": "Y",
"middle": [],
"last": "Guan",
"suffix": ""
},
{
"first": "X.-L",
"middle": [],
"last": "Wang",
"suffix": ""
}
],
"year": 2006,
"venue": "Lecture Notes in Computer Science of Natural Language Processing and Expert Systems",
"volume": "",
"issue": "4114",
"pages": "677--689",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jiang, W., Guan, Y., & Wang, X.-L. (2006) Conditional Random Fields Based Label Sequence and Information Feedback. Lecture Notes in Computer Science of Natural Language Processing and Expert Systems, (4114), 677-689.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Speaking Rate Effects on Discourse Prosody in Standard Chinese",
"authors": [
{
"first": "A.-J",
"middle": [],
"last": "Li",
"suffix": ""
},
{
"first": "Y.-Q",
"middle": [],
"last": "Zu",
"suffix": ""
}
],
"year": 2008,
"venue": "Proc. of the Speech Prosody2008",
"volume": "",
"issue": "",
"pages": "449--452",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Li, A.-J., & Zu, Y.-Q. (2008). Speaking Rate Effects on Discourse Prosody in Standard Chinese. In Proc. of the Speech Prosody2008, 449-452.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Conditional random fields: Probabilistic models for segmenting and labeling sequence data",
"authors": [
{
"first": "J",
"middle": [],
"last": "Lafferty",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Mccallum",
"suffix": ""
},
{
"first": "F",
"middle": [],
"last": "Pereira",
"suffix": ""
}
],
"year": 2001,
"venue": "Proc. of ICML",
"volume": "",
"issue": "",
"pages": "282--289",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lafferty, J., McCallum, A., & Pereira, F. (2001). Conditional random fields: Probabilistic models for segmenting and labeling sequence data. In Proc. of ICML, 282-289.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Wavesurfer -an open source speech tool",
"authors": [
{
"first": "K",
"middle": [],
"last": "Sjlander",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Beskow",
"suffix": ""
}
],
"year": 2000,
"venue": "Proc. of the ICSLP",
"volume": "4",
"issue": "",
"pages": "464--467",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sjlander, K. & Beskow, J. (2000). Wavesurfer -an open source speech tool. In Proc. of the ICSLP 2000, 4, 464-467.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Reference Manual for Speech Signal Processing",
"authors": [
{
"first": "",
"middle": [],
"last": "Sptk Working Group",
"suffix": ""
}
],
"year": 2009,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "SPTK Working Group. (2009). Reference Manual for Speech Signal Processing Toolkit Ver 3.3. available at http://sp-tk.sourceforge.net/",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Mel-generalized cepstral analysis -A unified approach to speech spectral estimation",
"authors": [
{
"first": "K",
"middle": [],
"last": "Tokuda",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kobayashi",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Masuko",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Imai",
"suffix": ""
}
],
"year": 1994,
"venue": "Proc. of ICSLP'94",
"volume": "",
"issue": "",
"pages": "1043--1046",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tokuda, K., Kobayashi, T., Masuko, T. & Imai, S. (1994). Mel-generalized cepstral analysis - A unified approach to speech spectral estimation. In Proc. of ICSLP'94, 1043-1046",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Speech parameter generation algorithms for HMM-Based speech synthesis",
"authors": [
{
"first": "K",
"middle": [],
"last": "Tokuda",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Yoshimura",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Masuko",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kobayashi",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kitamura",
"suffix": ""
}
],
"year": 2000,
"venue": "Proc. of ICASSP",
"volume": "",
"issue": "",
"pages": "1315--1318",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tokuda, K., Yoshimura, T., Masuko, T., Kobayashi, T., & Kitamura, T. (2000). Speech parameter generation algorithms for HMM-Based speech synthesis. In Proc. of ICASSP, 1315-1318.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Adaptation of pitch and spectrum for HMM-based speech synthesis using MLLR",
"authors": [
{
"first": "M",
"middle": [],
"last": "Tamura",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Masuko",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Tokuda",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kobayashi",
"suffix": ""
}
],
"year": 2001,
"venue": "Proc of ICASSP",
"volume": "",
"issue": "",
"pages": "805--808",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tamura, M., Masuko, T., Tokuda, K., & Kobayashi, T. (2001). Adaptation of pitch and spectrum for HMM-based speech synthesis using MLLR. In Proc of ICASSP, 805-808.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Corpus Phonetic Investigations of Discourse Prosody and Higher Level Information. Language and Linguistics, Institute of Linguistics",
"authors": [
{
"first": "C.-Y",
"middle": [],
"last": "Tseng",
"suffix": ""
}
],
"year": 2008,
"venue": "",
"volume": "9",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tseng, C.-Y. (2008). Corpus Phonetic Investigations of Discourse Prosody and Higher Level Information. Language and Linguistics, Institute of Linguistics, 9(3).",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Speaker interpolation for HMM-based speech synthesis system",
"authors": [
{
"first": "T",
"middle": [],
"last": "Yoshimura",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Masuko",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Tokuda",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kobayashi",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kitamura",
"suffix": ""
}
],
"year": 2000,
"venue": "J. Acoust. Soc. Jpn. (E)",
"volume": "21",
"issue": "4",
"pages": "199--206",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yoshimura, T., Masuko, T., Tokuda, K., Kobayashi, T., & Kitamura, T. (2000). Speaker interpolation for HMM-based speech synthesis system. J. Acoust. Soc. Jpn. (E), 21(4), 199-206.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Simultaneous modeling of phonetic and prosodic parameters, and characteristic conversion for HMM-based text-to-speech systems",
"authors": [
{
"first": "T",
"middle": [],
"last": "Yoshimura",
"suffix": ""
}
],
"year": 2002,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yoshimura, T. (2002) Simultaneous modeling of phonetic and prosodic parameters, and characteristic conversion for HMM-based text-to-speech systems. Ph.D thesis, Nagoya Institute of Technology.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Modeling Incompletion Phenomenon in Mandarin Dialog Prosody",
"authors": [
{
"first": "J",
"middle": [],
"last": "Yu",
"suffix": ""
},
{
"first": "L.-X",
"middle": [],
"last": "Huang",
"suffix": ""
},
{
"first": "J.-H",
"middle": [],
"last": "Tao",
"suffix": ""
},
{
"first": "X",
"middle": [],
"last": "Wang",
"suffix": ""
}
],
"year": 2007,
"venue": "Proc. of the Interspeech2007",
"volume": "",
"issue": "",
"pages": "462--465",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yu, J., Huang, L.-X., Tao, J.-H., & Wang, X. (2007). Modeling Incompletion Phenomenon in Mandarin Dialog Prosody. In Proc. of the Interspeech2007, 462-465.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "The HTK Book",
"authors": [
{
"first": "S",
"middle": [
"J"
],
"last": "Young",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Evermann",
"suffix": ""
},
{
"first": "M",
"middle": [
"J F"
],
"last": "Gales",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Hain",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Kershaw",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Moore",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Odell",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Ollason",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Povey",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Valtchev",
"suffix": ""
},
{
"first": "P",
"middle": [
"C"
],
"last": "Woodland",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Young, S. J., Evermann, G., Gales, M. J. F., Hain, T., Kershaw, D., Moore, G., Odell, J., Ollason, D., Povey, D., Valtchev, V., & Woodland, P. C. (2006). The HTK Book, version 3.4. Cambridge University Engineering Department, Cambridge, UK. \u4e2d\u592e\u7814\u7a76\u9662\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0chttp://ckipsvr.iis.sinica.edu.tw/, last visit 2009/09/09",
"links": null
}
},
"ref_entries": {
"TABREF1": {
"text": "\u3001\u57fa\u983b\u6a21\u578b(F0 parameter model)\u53ca\u97f3\u9577\u6a21\u578b(duration model)\u3002\u6b32\u5408\u6210\u8a9e \u97f3\u6642\uff0c\uf9dd\u7528\u4e0a\u8ff0\u8a13\uf996\u597d\u7684\u4e09\u7a2e\u6a21\u578b\uff0c\u4f9d\u64da\u8f38\u5165\u6587\u672c\u7684\u8a9e\u8a00\uf96b\uf969\u6216\u9810\u4f30\u4e4b\u97fb\uf9d8\u6a19\u8a18\u627e\u5230\u9069 \u7576 CDHMM \u6a21\u578b\u4e26\uf905\u63a5\u4e4b\uff0c\u518d\u4ee5\u7279\u6b8a\u7684\u6f14\u7b97\u6cd5\u7531\uf905\u63a5\u4e4b CDHMM \uf96b\uf969\u7522\u751f frame spectrum \u53ca frame F0 \uf96b\uf969\uff0c\u6700\u5f8c\u5c07 spectrum \u548c f0 \uf96b\uf969\u8f38\u5165 MLSA \uf984\u6ce2\u5668(Mel Log Spectrum Approximation filter)",
"num": null,
"content": "<table><tr><td>\u53ef\u8b8a\u901f\u4e2d\u6587\u6587\u5b57\u8f49\u8a9e\u97f3\u7cfb\u7d71</td><td>29</td></tr><tr><td colspan=\"2\">\u97f3\u5408\u6210\u65b9\u6cd5\uff1b\u5927\u578b\u8a9e\uf9be\u5eab\u5408\u6210\u6cd5\u7531\uf93f\u88fd\u597d\u7684\u8a9e\uf9be\u5eab\u4e2d\uff0c\u6311\u9078\u9069\u7576\u7684\u8a9e\u97f3\u4fe1\u865f\u7247\u6bb5\uf905\u63a5\u5408</td></tr><tr><td colspan=\"2\">\u6210\uff0c\u56e0\u6b64\u53ef\u539f\u97f3\u91cd\u73fe\uff0c\u6709\u6975\u4f73\u7684\u5408\u6210\u97f3\u8cea\uff0c\u4f46\u662f\u5982\u679c\u8981\u5408\u6210\u51fa\uf967\u540c\u7279\u6027\u7684\u8a9e\u97f3\uff0c\u5982\uf967\u540c</td></tr><tr><td colspan=\"2\">\u8b1b\u8a71\u901f\ufa01\u53ca\u591a\u7a2e\u60c5\u7dd2\u7b49\u61c9\u7528\uff0c\u5247\u9808\uf93f\u88fd\u5927\uf97e\u7684\u8a9e\uf9be\u4f5c\u70ba\u6311\u9078\u55ae\u5143\u7684\u57fa\u790e\uff0c\u7136\u800c\u6b32\u6536\u96c6\uf967</td></tr><tr><td colspan=\"2\">\u540c\u7279\u6027\u4e4b\u8a9e\uf9be\u4e26\uf967\u5bb9\uf9e0\uff0c\u56e0\u6b64\uff0c\u5c0d\u65bc\u5408\u6210\uf967\u540c\u7279\u6027\u8a9e\u97f3\u7684\u61c9\u7528\uff0c\u55ae\u5143\u9078\u53d6\u4e26\uf967\u662f\u4e00\u500b\u9069</td></tr><tr><td>\u5408\u7684\u65b9\u6cd5\u3002</td><td/></tr><tr><td colspan=\"2\">\u57fa\u65bc\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b\u8a9e\u97f3\u5408\u6210\u5668\u662f\u4e00\u7a2e\u7d71\u8a08\u5f0f\uf96b\uf969\u8a9e\u97f3\u5408\u65b9\u6cd5\uff0c\u662f\u76ee\u524d\u6700\u70ba\u5ee3\u6cdb</td></tr><tr><td colspan=\"2\">\u63a1\u7528\u7684\u5408\u6210\u65b9\u6cd5\uff0c\u5b83\u4ee5\u6587\u8108\u76f8\u95dc\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b(Context-dependent HMMs, CDHMMs)</td></tr><tr><td colspan=\"2\">\uf92d\u6a21\u64ec\uf967\u540c\u8a9e\u8a00\uf96b\uf969\u6216\u97fb\uf9d8\u67b6\u69cb\u4e0b\u7684\u8072\u5b78\u4fe1\u865f\uff0c\u5f9e\u8a9e\uf9be\u5eab\u8a13\uf996\u5f97\u5230\u983b\u8b5c\u6a21\u578b(spectral</td></tr><tr><td>parameter model)</td><td/></tr><tr><td/><td>) \u7684\u8a9e</td></tr></table>",
"type_str": "table",
"html": null
},
"TABREF4": {
"text": "state \u4e4b mean vector \u53ca covariance matrix\u3002 label construction \u6b65\u9a5f\u5f8c \u7522\u751f\u6587\u672c\u6a19 \u793a (label) \uff0c\u4f9d\u64da\u6587\u672c\u6a19 \u793a\u4f7f\u7528\u4e09\u7a2e \u8a9e\u901f\u4e4b CDHMM \u6a21\u578b\u3001state duration \u6a21\u578b\u53ca CDHMM \u6a21\u578b\uf96b\uf969\u6b0a\u91cd\uff0c\u7531\u6587\u672c\u76f8\u95dc\u6c7a\u7b56\u6a39\u627e\u5230\u9069 \u7576\u7684\u6a21\u578b\uff0c\u9996\u5148\u9810\u4f30\u51fa\u6bcf\u500b\u8072\u6bcd\u3001\u97fb\u6bcd\u6216\u975c\u97f3\u505c\u9813\u7684\u9577\ufa01\uff0c\u518d\uf9dd\u7528 maximum likelihood \u6cd5 (Tokuda et al., 2000) \u7522\u751f\u6bcf\u500b\u97f3\u6846\u7684 logF0 \u53ca MGC \u983b\u8b5c\uf96b\uf969\u3002 \u70ba\u5be6\u9a57\u7d50\u679c\u3002 \u7531\u6574\u9ad4\uf92d\u770b Inside test \u7684 RMSE \u90fd\u4f4e\u65bc Outside test \u9019\u662f\u56e0\u70ba\u904e\ufa01\u8a13\uf996(overfitting) \u7684\u95dc\u4fc2\uff0c\u7d93\u89c0\u5bdf\u767c\u73fe\u975c\u97f3\u505c\u9813\u97f3\u9577\u7684\u9810\u6e2c\uf967\uf941 Outside test \u548c Inside test \u5728\u8a9e\u901f\u5feb\u7684 RMSE",
"num": null,
"content": "<table><tr><td>\u53ef\u8b8a\u901f\u4e2d\u6587\u6587\u5b57\u8f49\u8a9e\u97f3\u7cfb\u7d71</td><td>\u6c5f\u632f\u5b87 \u7b49 39</td></tr><tr><td colspan=\"2\">\u7531\u5be6\u9a57\u7d50\u679c\u767c\u73fe\uff0c\u5c0d\u65bc\u5feb\u901f\u5408\u6210\u8a9e\u97f3\u9810\u4f30\u975c\u97f3\u505c\u9813\u7684\u7d50\u679c\u662f\u6700\u5dee\u7684\uff0c\u932f\u8aa4\u5927\u591a\u662f\u5728 AR \u4ee5\u53ca\u57fa\u983b\u7684\u7d71\u8a08\u5dee\uf962\u4e26\uf967\u5927\uff0c\u56e0\u6b64\uf967\u8003\u616e 0.5-0.5-0 \u9019\u500b\u6b0a\u91cd\u503c\u7d44\u5408\uff0c\u6240\u4ee5\u672c\u5be6\u9a57\u53ea</td></tr><tr><td colspan=\"2\">\u9810\u6e2c\u76ee\u6a19\u8a9e\uf906\u6709\u975c\u97f3\u505c\u9813\u7684\u90e8\u4efd\uff0c\u4e3b\u8981\u539f\u56e0\u53ef\u80fd\u662f\u56e0\u70ba\u5feb\u901f\u8a9e\uf9be\uf9e8\u97f3\u7bc0\u9593\u7684\u975c\u97f3\u505c\u9813\u8f03 \u6709 16 \u7a2e\u975c\u97f3\u505c\u9813-CDHMM \u6b0a\u91cd\u7684\u7d44\u5408\u3002</td></tr><tr><td colspan=\"2\">\u5c11\uff0c\u6240\u4ee5\u9020\u6210\uf9ba\u6c7a\u7b56\u6a39\u5b78\u7fd2\u5230\u97f3\u7bc0\u9593\u7121\u975c\u97f3\u505c\u9813\u7684\u6a5f\uf961\u8f03\u5927\uff0c\u5728\u9810\u6e2c\u7d50\u679c\u4e5f\u662f\u504f\u5411\u6c92\u6709 \u6bcf\u4e00\u500b\u5408\u6210\u6587\u672c\u5747\u70ba outside test \u7684\u8a9e\uf906\uff0c\u4e00\u500b\u6587\u672c\u4f9d\u64da\uf978\u7d44\u6b0a\u91cd\u503c\u8b8a\u5316\u6703\u7522\u751f 16</td></tr><tr><td colspan=\"2\">\u7b2c\u4e00\u7d44\u6b0a\u91cd\u503c\u5f71\u97ff\u975c\u97f3\u505c\u9813\u7684\u8b8a\u5316\uff0c\u800c\u7b2c\u4e8c\u7d44\u6b0a\u91cd\u503c\u5f71\u97ff\uf9ba\u97f3\u9577\u3001\u983b\u8b5c\u53ca\u57fa\u983b\uff0c\u5728 \u81ea\u7136\u7684\u8a9e\u97f3\u8a0a\u865f\u4e2d\uff0c\u6162\u901f\u8a9e\uf9be\u975c\u97f3\u505c\u9813\u8f03\u591a\uff0c\u97f3\u7bc0\u97f3\u9577\u4e5f\u6703\uf925\u9577\uff0c\u5feb\u901f\u8a9e\uf9be\u5247\u76f8\u53cd\u3002\u5728 \u7d66\u5b9a\u6b0a\u91cd\u503c\u4e5f\u9700\u8981\u6309\u7167\u8a9e\u901f\u7684\u898f\u5247\uff0c\u7576\u60f3\u8981\u5408\u6210\u8a9e\u901f\u8f03\u5feb\u7684\u8a9e\u97f3\u8a0a\u865f\u6642\uff0c\u589e\u52a0\u5feb\u901f\u8a9e\u901f \u4e4b\u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u7684\u6bd4\u91cd\uff0c\u4f7f\u975c\u97f3\u505c\u9813\u9810\u4f30\u51fa\u7684\uf969\uf97e\u8f03\u5c11\uff0c\u53ea\u5728\u9069\u5408\u7684\u4f4d\u7f6e\u7d66\u5b9a\u975c\u97f3\u505c \u9813\uff0c\u540c\u6642\uff0c\u6211\u5011\u4e5f\u8abf\u6574\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b\u6b0a\u91cd\uff0c\u589e\u52a0\u5feb\u901f\u8a9e\uf9be\u7684\u6bd4\u91cd\uff0c\u800c\u53ef\u4ee5\u7522\u751f\u51fa\u8f03 \u77ed\u7684\u97f3\u9577\u53ca\u8f03\u9ad8\u7684\u57fa\u983b\uff0c\u9019\uf978\u7d44\u6b0a\u91cd\u503c\u9700\u8981\u6709\u6b63\u76f8\u95dc\u624d\u6703\u5339\u914d\uff0c\uf978\u7d44\uf967\u5339\u914d\u7684\u6b0a\u91cd\u503c\u6703 \u5408\u6210\u51fa\uf967\u81ea\u7136\u7684\u8a9e\u97f3\u8a0a\u865f\uff0c\u56e0\u6b64\u5728\uf967\u540c\u8a9e\u901f\u4e0b\uf978\u7d44\u6b0a\u91cd\u503c\u7684\u5339\u914d\u662f\u76f8\u7576\u91cd\u8981\u7684\uff0c\u5728\u4e4b\u5f8c \u7684\u5be6\u9a57\u6703\u5c0d\u9019\uf978\u7d44\u6b0a\u91cd\u503c\u5339\u914d\u4f5c\u4e3b\u89c0\u6e2c\u8a66\u7684\u5be6\u9a57\u3002 3.3 Label Construction \u6b32\u5408\u6210\u7684\u6587\u5b57\u7d93\u7531\u6587\u672c\u5206\u6790\u5f8c\uff0c\u53ef\u5f97\u5230\u5c0d\u61c9\u6587\u8108\u76f8\u95dc\u7684\u8a9e\u8a00\uf96b\uf969\u8cc7\u8a0a\uff0c\u4f7f\u7528\u4e4b\u524d\u4ee5\u5167\u63d2 \u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u6240\u9810\u4f30\u4e4b\u975c\u97f3\u505c\u9813\uff0c\u653e\u5165\u6587\u672c\u6a19\u793a(label)\uff0c\u6700\u7d42\u7522\u751f\u7684 label \u5177\u6709\u6b32\u5408\u6210 \u6587\u672c\u4e2d\u6bcf\u500b\u8072\u6bcd\u3001\u97fb\u6bcd\u53ca\u975c\u97f3\u505c\u9813\u7684\u6587\u8108\u76f8\u95dc\u8a9e\u8a00\uf96b\uf969\u3002 Spectrum Approximation filter) (Imai, 1983)\u7522\u751f\u5408\u6210\u8a9e\u97f3\u3002 4. \u5be6\u9a57\u7d50\u679c\u53ca\u8a0e\uf941 \u5be6\u9a57\u8a9e\uf9be\u5df2\u65bc 1.4 \u4e2d\u4ecb\u7d39\uff0c\u5c0d\u65bc\u6bcf\u7a2e\u8a9e\u901f\u53d6\u5176\u7d04 328 \u500b\u8a9e\uf906\u70ba\u8a13\uf996\u8a9e\uf9be\uff0c\u53e6 20 \uf906\u70ba\u6e2c\u8a66 \u8a9e\uf9be\uff0c\u70ba\uf9ba\u8a55\u4f30\u672c\u53ef\u8b8a\u901f\u6f22\u8a9e\u8a9e\u97f3\u5408\u6210\u7cfb\u7d71\u7684\u6548\u80fd\uff0c\u6211\u5011\u5206\u5225\u5c0d\u5408\u6210\u8a9e\u97f3\u9032\ufa08\u5ba2\u89c0\u53ca\u4e3b \u89c0\u7684\u6e2c\u8a66\u3002\u5728\u5ba2\u89c0\u6e2c\u8a66\u65b9\u9762\uff0c\u6211\u5011\uf97e\u6e2c\uf9ba\u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u7684\u9810\u4f30\u6b63\u78ba\u6027\uff0c\u53e6\u5916\uff0c\u4e5f\uf97e\u6e2c \uf9ba\u6574\u500b\u7cfb\u7d71\u5408\u6210\u8a9e\u97f3\u548c\u76ee\u6a19\u8a9e\u97f3\u7684\uf97e\u5316\u8aa4\u5dee\u3002\u800c\u5728\u4e3b\u89c0\u6e2c\u8a66\u65b9\u9762\uff0c\u5c0d\u7cfb\u7d71\uf978\u7d44\u6b0a\u91cd\u503c\u5339 \u914d\u7684\uf9fa\u6cc1\u4f5c\u4e3b\u89c0\u6e2c\u8a66\u7684\u5be6\u9a57\uff0c\u5408\u6210\u97f3\u6a94\u7684\u5c55\u793a\u8acb\uf99a\u7d50 http://140.113.144.71\u3002 4.1 \u5ba2\u89c0\u6e2c\u8a66 \u5728\u7b2c\u4e00\u500b\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u5c0d\u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u7684\u6548\u80fd\u9032\ufa08\u8a55\u4f30\uff0c\u8a08\u7b97\u5408\u6210\u97f3\u6a94\u548c\u76ee\u6a19\u8a9e\uf906\u7684 \u975c\u97f3\u505c\u9813\u9810\u4f30\u7684\u6b63\u78ba\uf961\u53ca\u6df7\u6dc6\u7a0b\ufa01\uff0c\u56e0\u70ba\u53ea\u6709\u55ae\u7d14\u4e09\u7a2e\u8a9e\u901f\u7684\u76ee\u6a19\u8a9e\uf906\uff0c\u6c92\u6709\u5be6\u969b\u4ecb\u65bc \u9019\u4e09\u7a2e\u8a9e\u901f\u7684\u76ee\u6a19\u8a9e\uf906\uff0c\u6240\u4ee5\u53ea\u6709\u5c0d\u65bc\u4e09\u7a2e\uf967\u540c\u8a9e\u901f\u76ee\u6a19\u8a9e\uf906\u7684\u9810\u4f30\u7d50\u679c\u4f5c\u89c0\u5bdf\uff0c\u4ee5\u5408 \u6210\u5feb\u901f\u8a9e\u97f3\u70ba\uf9b5\uff0c\u7576\u6e2c\u8a66\u5feb\u901f\u8a9e\uf9be\u6642\uff0c\u6211\u5011\u8abf\u6574\u5feb\u901f\u7684\u975c\u97f3\u505c\u9813\u9810\u4f30\u6c7a\u7b56\u6a39\u6b0a\u91cd\u503c\u70ba 1\uff0c \u5176\u4ed6\u8a9e\u901f\u4e4b\u6b0a\u91cd\u70ba 0\uff0c\u4e2d\u6162\u901f\u6e2c\u8a66\u4ea6\u540c\u3002\u8868 4 \u70ba\u9810\u4f30\u975c\u97f3\u505c\u9813\u5c0d\u65bc\u5feb\u4e2d\u6162\u8a9e\u901f\u7684\u7d50\u679c\u3002 \u8868 4. \uf967\u540c\u8a9e\u901f\u4e0b\u9810\u4f30\u975c\u97f3\u505c\u9813\u7684\u7d50\u679c\uff0cXX*\u4ee3\u8868\u9810\u6e2c\u70ba\u975c\u97f3\u505c\u9813\u6216\u975e\u975c\u97f3\u505c\u9813 (\u4ee5 \u767e\u5206\u6bd4\u8868\u793a) \uff0cTotal \u70ba Non-SP \u6216 SP \u7684\u7e3d\u500b\uf969\u3002 \u6162 Inside Outside Non-SP* SP* Total Non-SP* SP* Total Non-SP 90.05 9.95 28108 Non-SP 89.66 10.34 1885 SP 30.19 69.81 20486 SP 33.57 66.43 1415 \u4e2d Inside Outside Non-SP* SP* Total Non-SP* SP* Total Non-SP 92.77 7.23 29119 Non-SP 91.55 8.45 1977 SP 37.81 62.19 19314 SP 39.61 60.39 1323 \u5feb Inside Outside Non-SP* SP* Total Non-SP* SP* Total \u975c\u97f3\u505c\u9813\uff0c\u53e6\u5916\u53ef\u80fd\u7684\u539f\u56e0\uff0c\u662f\u8003\u616e\u5230\u5feb\u901f\u8a9e\uf9be\uf967\uf941\u662f AR \u548c SR \u8b8a\u5316\u90fd\u662f\u6700\u5927\u7684\uff0c\u8a9e \uf906\u548c\u8a9e\uf906\u9593\u8a9e\u901f\u6709\u8f03\u5927\u7684\u5dee\uf962\uff0c\u56e0\u70ba\u8a9e\u901f\u548c\u975c\u97f3\u505c\u9813\u7684\u591a\u5be1\u6709\u95dc\u4fc2\uff0c\u8a9e\u901f\u7684\u5dee\uf962\u6f5b\u5728\u6703 \u9020\u6210\u5feb\u901f\u8a9e\uf9be\u975c\u97f3\u505c\u9813\u9810\u4f30\u4e0a\u7684\u56f0\u96e3\u3002\u5728\u6162\u901f\u8a9e\uf9be\u4e0a\u96d6\u7136\u5728\u975e\u975c\u97f3\u505c\u9813\u9810\u6e2c\u4e0a\uf976\u8f38\u5feb\u901f \u8a9e\uf9be\uff0c\u4f46\u5728\u6709\u975c\u97f3\u505c\u9813\u9810\u6e2c\u4e0a\u6bd4\u5feb\u901f\u8a9e\uf9be\u6e96\u5f97\u591a\uff0c\u53ef\u80fd\u662f\u56e0\u70ba\u8a9e\u8005\u65bc\uf929\uf95a\u6162\u901f\u8a9e\uf9be\u6642\uff0c \u6703\u5c07\u8a5e\u6216\u97fb\uf9d8\u8a5e\u7684\u7d50\u69cb\u6e05\u695a\uf9a3\u51fa\uff0c\u6240\u4ee5\u5728\u8a9e\uf9be\u4e0a\u7522\u751f\u8f03\u4e00\u81f4\u6027\u7684\u975c\u97f3\u505c\u9813\uff0c\u8f03\u5bb9\uf9e0\u5f9e\u8a9e \u8a00\uf96b\uf969\u5b78\u7fd2\u5230\u898f\u5247\uff0c\u56e0\u6b64\u6e96\u78ba\ufa01\u6bd4\u5feb\u901f\u8981\u9ad8\u7684\u591a\u3002 \u7b2c\u4e8c\u500b\u5ba2\u89c0\u6e2c\u8a66\uff0c\u6211\u5011\u5206\u5225\u6e2c\uf97e\u5408\u6210\u548c\u76ee\u6a19\u8a9e\uf906\u5176\u57fa\u983b\u3001\u505c\u9813\u975c\u97f3\u7684\u9577\ufa01\u4ee5\u53ca\u97f3\u7bc0 \u7684\u9577\ufa01\u7684\u8aa4\u5dee\uff0c\u4f7f\u7528\u5747\u65b9\u6839\u8aa4\u5dee(Root Mean Square Error, RMSE)\u7528\uf92d\u8a55\u4f30\u8aa4\u5dee\u503c\uff0c\u56e0 \u8a9e\uf9be\u5eab\u53ea\u6709\u4e09\u7a2e\u8a9e\u901f\uff0c\u6240\u4ee5\u5728\u6e2c\uf97e\u5feb\u901f\u8a9e\uf9be\u7684 RMSE \u6642\uff0c\u9810\u6e2c\u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u7684\u6b0a\u91cd\u548c \u96b1\u85cf\u99ac\u53ef\u592b\u5f0f\u6a21\u578b\u7684\u6b0a\u91cd\uff0c\u5747\u8a2d\u5b9a\u5feb\u901f\u6b0a\u91cd\u503c\u70ba 1\uff0c\u5176\u4ed6\u6b0a\u91cd\u8a2d\u70ba 0\uff0c\u4e2d\u6162\u901f\u8a9e\uf9be\u4e5f\u4f7f\u7528 \u540c\u6a23\u7684\u65b9\u6cd5\u6e2c\uf97e\uff0c\u8868 5 \u5747\u70ba\u6700\u4f4e\uff0c\u7531\u65bc\u5feb\u901f\u8a9e\u901f\u5728\u975c\u97f3\u505c\u9813\u7684\u97f3\u9577\u4e26\uf967\u9577\uff0c\u5c31\u7b97\u975c\u97f3\u505c\u9813\u6c92\u6709\u6b63\u78ba\u9810\u4f30\u51fa\uf92d\uff0c \u8aa4\u5dee\u4e5f\uf967\u6703\u592a\u5927\uff0c\u800c\u6162\u901f\u7684\u975c\u97f3\u505c\u9813\u5c31\uf967\u4e00\u6a23\uff0c\u975c\u97f3\u505c\u9813\u97f3\u9577\u8f03\u9577\uff0c\u6c92\u6709\u6b63\u78ba\u9810\u4f30\u5230\u975c \u97f3\u505c\u9813\u8aa4\u5dee\u5c31\u6703\u8f03\u5927\uff0c\u6211\u5011\u89c0\u5bdf\u97f3\u7bc0\u97f3\u9577\u7684 RMSE \u4e5f\u770b\u5230\u540c\u6a23\u7684\u7d50\u679c\uff0c\u5728\u8a9e\u901f\u5feb\u7684\u97f3\u7bc0 \u97f3\u9577 RMSE \u5747\u70ba\u6700\u4f4e\uff0c\u56e0\u70ba\u5feb\u901f\u8a9e\uf9be\u97f3\u7bc0\u97f3\u9577\u90fd\u8f03\u77ed\uff0c\u8a08\u7b97\u8aa4\u5dee\u4e5f\uf967\u6703\u592a\u5927\u3002 \u8868 5. \u5feb\u4e2d\u6162\u8a9e\uf9be\u4f5c\u6e2c\u8a66\u4e4b RMSE \u503c \u6e2c\u8a66\u9805\u76ee \u8a9e\u901f Fast Median Slow Inside F0 (Hz) 36.28 34.38 35.21 Outside F0 (Hz) 42.66 42.78 45.23 Inside sp duration (ms) 44.97 64.19 84.17 Outside sp duration (ms) 56.55 60.02 85.55 Inside syllable duration (ms) 37.53 41.44 44.19 Outside syllable duration (ms) 39.23 42.66 47.08 4.2 \u4e3b\u89c0\u6e2c\u8a66 \u4e3b\u89c0\u6e2c\u8a66\u76ee\u7684\u70ba\u6e2c\u8a66\u7cfb\u7d71\uf978\u7d44\u6b0a\u91cd\u503c\uf967\u540c\u7684\u7d44\u5408\uff0c\u4ee5\u4e3b\u89c0\u6e2c\u8a66\u5224\u5225\u5408\u6210\u8a9e\u97f3\u7684\u81ea\u7136\ufa01\uff0c \u7a2e\uf967\u540c\u8a9e\u901f\u8b8a\u5316\u7684\u5408\u6210\u97f3\u6a94\uff0c\u5404\u5206\u70ba\u56db\u7d44\u4f5c\u6e2c\u8a66\uff0c\u4ee5\u540c\u6a23\u96b1\u85cf\u99ac\u53ef\u592b\u6a21\u578b\u7684\u6b0a\u91cd\u503c\u70ba\u540c \u4e00\u7d44\uff0c\u76ee\u7684\u70ba\u56fa\u5b9a\u4e00\u7d44\u6b0a\u91cd\u503c\uff0c\u89c0\u5bdf\uf967\u540c\u6b0a\u91cd\u9810\u6e2c\u975c\u97f3\u505c\u9813\u7684\u5339\u914d\u7a0b\ufa01\u3002\u4e3b\u89c0\u6e2c\u8a66\u4e2d\u8a9e \u97f3\u81ea\u7136\ufa01\u7684\u8a55\u5206\u70ba\u4e94\u5206\u5236\uff0c\u5206\uf969\u70ba\u4e00\u81f3\u4e94\uff0c\u4e00\u70ba\u6700\uf967\u81ea\u7136\uff0c\u4e94\u70ba\u6700\u81ea\u7136\uff0c\u7e3d\u5171\u5c0d 6 \u4eba\u4f5c \u4e3b\u89c0\u6e2c\u8a66\uff0c\u6bcf\u500b\u6e2c\u8a66\u8005\u7531\u4e5d\uf906\u6587\u672c\u4e2d\u9078\u807d\uf978\uf906\u6587\u672c\u7684\u8a9e\uf906\uff0c\u5176\u4e2d\u4e00\uf906\u6587\u672c\u8207\u53e6\u4e00\u500b\u6e2c\u8a66 \u8005\u91cd\u8907\uff0c\u56e0\u70ba\u6bcf\u6587\u672c\u6709 16 \u7a2e\u8a9e\u901f\u6b0a\u91cd\u7d44\u5408\uff0c\u6240\u4ee5\u6bcf\u500b\u4eba\u807d 32 \uf906\u6e2c\u8a66\u8a9e\uf906\uff0c\u6574\u500b\u6e2c\u8a66\u8a9e \uf906\u5171\u6709 192 \uf906\uff0c\u6e2c\u8a66\u7d50\u679c\u5982\u8868 6 \u6240\u793a\u3002 \u88686. \u4e3b\u89c0\u6e2c\u8a66\u7684\u5e73\u5747\u503c\u00b1\u4e00\u500b\u6a19\u6e96\u5dee\uff0cx-x-x\u4e2d\u7684x\u9806\u5e8f\u4ee3\u8868\u6162\u3001\u4e2d\u3001\u5feb\u6b0a\u91cd\u503c \u9810\u6e2c\u975c\u97f3\u505c\u9813\u6b0a\u91cd\u503c \u96b1\u85cf\u99ac\u53ef\u592b\u6b0a\u91cd\u503c 1-0-0 0-1-0 0-0.5-0.5 0-0-1 1-0-0 2.33\u00b10.61 3.08\u00b11.36 2.79\u00b10.98 2.21\u00b10.70 0-1-0 2.54\u00b10.88 3.38\u00b10.96 3.25\u00b10.391 2.21\u00b10.52 0-0.5-0.5 2.67\u00b10.60 3.08\u00b10.99 3.67\u00b10.79 2.54\u00b11.43 0-0-1 2.83\u00b10.88 2.88\u00b11.00 3.71\u00b10.93 3.25\u00b11.66 \u7531\u4e3b\u89c0\u5be6\u9a57\u767c\u73fe\uff0c\uf978\u7d44\u6b0a\u91cd\u503c\u5fc5\u9808\u6709\u6b63\u76f8\u95dc\u7684\u95dc\u4fc2\uff0c\u5408\u6210\u51fa\u7684\u8a9e\u97f3\u624d\u6703\u81ea\u7136\uff0c\u7576\uf978 \u7d44\u6b0a\u91cd\u503c\uf967\u76f8\u5339\u914d\u7684\u6642\u5019\uff0c\u5408\u6210\u51fa\u7684\u8a9e\u97f3\u5927\u591a\uf967\u81ea\u7136\uff0c\u56e0\u6b64\u901a\u5e38\u5728\u8868\uf9d1\u5c0d\u89d2\u7dda\u9644\u8fd1\u6703\u6709 \u6700\u5927\u7684\u81ea\u7136\ufa01\uff0c\u4f46\u7576\u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u6b0a\u91cd\u503c\u70ba 0-0-1 \u548c\u96b1\u85cf\u99ac\u53ef\u592b\u6b0a\u91cd\u503c\u70ba 1-0-0 \u662f\u6bd4 \u8f03\uf9a8\u4eba\u8a1d\uf962\u7684\u7d50\u679c\uff0c\u731c\u6e2c\u5728\u96b1\u85cf\u99ac\u53ef\u592b\u6b0a\u91cd\u503c\u70ba 1-0-0 \u6642\u8a9e\u901f\u5f88\u6162\uff0c\u9020\u6210\u6e2c\u8a66\u8005\u807d\u5f97\u53ad \u7169\uff0c\u7531\u8868\uf9d1\u56fa\u5b9a\u96b1\u85cf\u99ac\u53ef\u592b\u6b0a\u91cd\u503c 1-0-0 \u89c0\u5bdf\uff0c\u767c\u73fe\u7121\uf941\u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u6b0a\u91cd\u5982\u4f55\u8abf\u6574\uff0c \u53d7\u6e2c\u8005\u6240\u7d66\u4e88\u7684\u81ea\u7136\ufa01\u90fd\u504f\u4f4e\uff0c\u5728\u9019\u7a2e\u6b0a\u91cd\u503c\u7d44\u5408\u4e0b\uff0c\u53d7\u6e2c\u8005\u89ba\u5f97\u53ad\u7169\u5206\uf969\u90fd\u7d66\u7684\u8f03\u4f4e\u3002 5. \u7d50\uf941\u8207\u672a\uf92d\u65b9\u5411 \u672c\u7cfb\u7d71\u70ba\u53ef\u8b8a\u901f\u4e2d\u6587\u6587\u5b57\u8f49\u8a9e\u97f3\uff0c\u7d93\u7531\u6b0a\u91cd\u503c\u8abf\u6574\u6240\u5408\u6210\u51fa\u7684\u8a9e\u97f3\uff0c\u5408\u6210\u51fa\uf92d\u7684\u8a9e\u97f3\u57fa \u672c\u4e0a\u5c1a\u4f73\uff0c\u5728\u4e3b\u89c0\u5be6\u9a57\u4e2d\uf978\u7d44\u6b0a\u91cd\u503c\u7686\u8abf\u70ba 0-0.5-0.5 \u6240\u5408\u6210\u51fa\u7684\u8a9e\u97f3\u81ea\u7136\ufa01\u4e5f\u662f\uf9a8\u4eba\u6eff \u610f\u7684\uff0c\u5176\u8a9e\u901f\u4ecb\u65bc\u4e2d\u901f\u53ca\u5feb\u901f\u4e4b\u9593\uff0c\u7cfb\u7d71\u53ef\u9810\u6e2c\u51fa\u9069\u7576\u7684\u975c\u97f3\u505c\u9813\u3001\u983b\u8b5c\u53ca\u5176\u4ed6\u97fb\uf9d8\uf96b \uf969\uff0c\u9054\u5230\u5408\u6210\u51fa\uf967\u540c\u8a9e\u901f\u7684\u81ea\u7136\u8a9e\u97f3\u3002\u7531\u5ba2\u89c0\u5be6\u9a57\u767c\u73fe\u5feb\u901f\u7684\u975c\u97f3\u505c\u9813\u9810\u4f30\u7d50\u679c\u8f03\u5dee\uff0c \u672a\uf92d\u7684\u7814\u7a76\u6703\u4ee5\u6162\u901f\u70ba\u57fa\u6e96\u9810\u4f30\u5176\u4ed6\u8a9e\u901f\u7684\u975c\u97f3\u505c\u9813\uff0c\u56e0\u70ba\u5728\u6162\u901f\u6642\uf95a\u7a3f\u4eba\u6703\u5b8c\u6574\u5206\u6790 \u8a5e\u548c\u97fb\uf9d8\u8a5e\u7d50\u69cb\u5f8c\uf9a3\u51fa\u8a9e\uf906\uff0c\u8003\u616e\u76f8\u5c0d\u975c\u97f3\u505c\u9813\u7684\u8b8a\u5316\uff0c\u5982\u67d0\u4e9b\u97f3\u7bc0\u9593\u6216\u8a5e\u9593\uf967\u7ba1\u5728\u6162 \u901f\u9084\u662f\u5feb\u901f\u90fd\u9700\u8981\u975c\u97f3\u505c\u9813\uff0c\u800c\u6709\u4e9b\u975c\u97f3\u505c\u9813\u5728\u5feb\u901f\u6642\u53cd\u800c\u6d88\u5931\uff0c\u8003\u616e\u9019\u4e9b\u76f8\u5c0d\u7684\u8b8a\u5316 \u518d\u9032\u4e00\u6b65\u9032\ufa08\u975c\u97f3\u505c\u9813\u9810\u4f30\u662f\u9700\u8981\u7684\u3002 3.4 \u5728 \u5c07\u4e0a\u4e00\u6b65\u5f97\u5230\u7684\u6bcf\u500b\u97f3\u6846\u4e4b logF0 \u548c MGC \u983b\u8b5c\uf96b\uf969\u8f38\u5165\u81f3 MSLA filter (Mel-Log Non-SP 96.34 3.66 35380 Non-SP 94.83 5.17 2496 \u5c0d\u65bc\u5feb\u4e2d\u6162\uf978\u7d44\u6b0a\u91cd\u503c\u8a2d\u70ba\uff1a1-0-0\u30010-1-0\u30010-0-1\u30010-0.5-0.5(x-x-x \u4e2d\u7684 x \u9806\u5e8f\u4ee3\u8868\u6162 \u975c\u97f3\u505c\u9813\u6c7a\u7b56\u6a39\u662f\u7531\u8a9e\u8a00\uf96b\uf969\u9810\u4f30\u8a5e\u4e4b\u9593\u975c\u97f3\u505c\u9813\u7684\u51fa\u73fe\u8207\u5426\uff0c\u96d6\u7136\u672c\u7cfb\u7d71\u6240\u9810\u4f30</td></tr><tr><td colspan=\"2\">SP \u901f\u3001\u4e2d\u901f\u3001\u4ee5\u53ca\u5feb\u901f\u6b0a\u91cd\u503c) \uff0c\u56e0\u70ba\u8003\u616e\u7684\u7d44\u5408\uf969\uf97e\u904e\u591a\uff0c\u800c\u4e14\u6162\u901f\u8ddf\u4e2d\u901f\u8a9e\uf9be\u4f9d\u64da SR\u3001 49.5 50.5 11613 SP 52.74 47.26 804 \u7684\u975c\u97f3\u505c\u9813\u7d50\u679c\u5c1a\u4f73\uff0c\u4f46\u4ee5\u9019\u500b\u7cfb\u7d71\u6240\u9810\u4f30\u51fa\uf92d\u975c\u97f3\u505c\u9813\u7279\u6027\u4e26\u6c92\u6709\u8003\u616e\u5be6\u969b\u975c\u97f3\u505c\u9813</td></tr></table>",
"type_str": "table",
"html": null
}
}
}
} |