{
    "paper_id": "O06-1010",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T08:07:01.579798Z"
    },
    "title": "",
    "authors": [],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "",
    "pdf_parse": {
        "paper_id": "O06-1010",
        "_pdf_hash": "",
        "abstract": [],
        "body_text": [
            {
                "text": "\u6cd5(maximum likelihood, ML)\uff0c\u6c7a\u5b9a\u6700\u4f73\u7684\u591a\u8a9e\u8fa8\uf9fc\u5e8f\uf99c\u3002\u4f5c\u6cd5\u4e0a\uf9d0\u4f3c\u5c0d\u8a9e\u97f3\u8fa8\uf9fc\u5e8f\uf99c\u505a\u9a57\u8b49(verification) \u4e4b\u8655\uf9e4 [3] \uff1b\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u7684\u8868\u73fe\u53d6\u6c7a\u65bc\u5f8c\u7aef\u6700\u4f73\u5e8f\uf99c\u9078\u64c7\u4e4b\u6548\u679c\u3002\uf9dd\u7528\u9078\u64c7\u6700\u5927\u4f3c\u7136\u7684\u65b9\u6cd5\u7f3a\u9ede\u5728\u65bc\uff0c \u591a\u8a9e\u8fa8\uf9fc\u7684\u6548\u679c\u6703\u53d7\u5230 ML \u65b9\u6cd5\u7684\u9650\u5236\uff0c\u4e14\u8fa8\uf9fc\u7684\u591a\u8a9e\uf906\u578b\u9700\u8981\u53e6\u5916\u8003\u616e\uff0c\ufa00\u5272\u51fa\u8a9e\uf906\u5167\uf967\u540c\u8a9e\u8a00\u7684\u6bb5\uf918\u3002 \u7b2c\u4e09\uf9d0\u4f5c\u6cd5\uff1a\u85c9\u7531\u5b9a\u7fa9\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u55ae\u5143\u96c6 [4] \uff0c\u5408\u4f75\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20\u6a21\u578b\uff0c\uf92d\u9032\ufa08\u591a\u8a9e\u8a9e\u97f3\u8fa8 \uf9fc\u3002\u672c\uf941\u6587\u4e43\u57fa\u65bc\u6b64\u65b9\u6cd5\uff0c\u63a2\u8a0e\u5982\u4f55\u5b9a\u7fa9\u51fa\u6709\u6548\u7684\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u55ae\u5143\u6a21\u578b\u3002 \u5728\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\u4e4b\u5efa\uf9f7\u53ef\u4ee5\u6b78\u7d0d\u70ba\u4e09\u7a2e\u65b9\u5f0f\u3002\u9996\u5148\uff0c\u6211\u5011\u53ef\u4ee5\u76f4\u63a5\u5408\u4f75\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20\u96c6\uff0c\u5efa\uf9f7 \u591a\u8a9e\u97f3\u7d20\u6a21\u578b\uff0c\u4f46\u662f\u9019\u7a2e\u65b9\u6cd5\u6c92\u6709\u8003\u616e\u591a\u8a9e\u97f3\u7d20\u9593\uf96b\uf969\u5206\u4eab\u7684\u7279\u6027\u3002\u7b2c\u4e8c\uff0c\u85c9\u7531\u5c0d\u7167\u570b\u969b\u97f3\u7d20\u6a19\u6e96\u5b9a\u7fa9\uff0c \u8003\u616e\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20\uff0c\u9054\u5230\u591a\u8a9e\u97f3\u7d20\u9593\uf96b\uf969\u5171\u7528\u7684\u7279\u6027\uff0c\u4f46\u662f\u6b64\u4f5c\u6cd5\u4e0a\u7f3a\u4e4f\u8cc7\uf9be\u7d71\u8a08\u7684\u5206\u6790\uff0c\u800c\u662f\u7531 \u5c08\u5bb6\u77e5\uf9fc\u6c7a\u5b9a\u5404\u97f3\u7d20\u5b9a\u7fa9\u3002\u570b\u969b\u97f3\u7d20\u6a19\u6e96\u5b9a\u7fa9\u5305\u542b\u6709\uff1aInternational Phonetic Alphabet (IPA) [5] \u3001Speech Assessment Methods Phonetic Alphabet (SAMPA) [6] \u548c Worldbet [7] \u7b49\u3002\u7b2c\u4e09\uff0c\u4f30\u8a08\u591a\u8a9e\u8a9e\u97f3\u97f3\u7d20\u9593\u76f8\u4f3c\u7a0b \ufa01\uff0c\u7531\u4e0b\u800c\u4e0a\u968e\u5c64\u5f0f\u9032\ufa08\u591a\u8a9e\u97f3\u7d20\u5408\u4f75\uff0c\u4ee5\u5b9a\u7fa9\u591a\u8a9e\u97f3\u7d20\u96c6\u3002\u591a\u8a9e\u8a9e\u97f3\u97f3\u7d20\u9593\u76f8\u4f3c\ufa01\u7684\uf97e\u6e2c\uff0c\u53ef\u4ee5\uf9dd\u7528 Bhattacharyya distance [8] \u6216\u8005\u662f Kullback-Leibler (KL) divergence [9] ",
                "cite_spans": [
                    {
                        "start": 72,
                        "end": 75,
                        "text": "[3]",
                        "ref_id": "BIBREF2"
                    },
                    {
                        "start": 188,
                        "end": 191,
                        "text": "[4]",
                        "ref_id": "BIBREF3"
                    },
                    {
                        "start": 451,
                        "end": 454,
                        "text": "[5]",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 508,
                        "end": 511,
                        "text": "[6]",
                        "ref_id": "BIBREF5"
                    },
                    {
                        "start": 523,
                        "end": 526,
                        "text": "[7]",
                        "ref_id": "BIBREF6"
                    },
                    {
                        "start": 614,
                        "end": 617,
                        "text": "[8]",
                        "ref_id": "BIBREF7"
                    },
                    {
                        "start": 655,
                        "end": 658,
                        "text": "[9]",
                        "ref_id": "BIBREF8"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "D 1 2 1 1 2 1 2 1 2 1 2 1 2 ( ) ( ) l n 8 2 2 T bha D \u03bc \u03bc \u03bc \u03bc \u2212 1 \u2211 + \u2211 \u2211 + \u2211 \u23a1 \u23a4 = \u2212 \u2212 + \u23a2 \u23a5 \u23a3 \u23a6 \u2211 \u2211 (\u5f0f 1) \u5176\u4e2d\uff0c \u03bc \u548c \u2211 \u5206\u5225\u8868\u793a\u97f3\u7d20\u6a21\u578b\u7684\u5e73\u5747\u503c\u548c\u8b8a\uf962\uf969\u5411\uf97e\uff0cT \u662f\u8f49\u7f6e\u77e9\u9663\u3002\u53e6\u5916\uff0c\u53ef\u4ee5\uf9dd\u7528 Kullback-Leibler (KL) divergence [9]\uf92d\u6c7a\u5b9a\uf978\u500b\u6a5f\uf961\u5206\u4f48\u7684\u76f8\u4f3c\ufa01 KL D \u3002\u4ee5 KL-divergence \u4f30\u7b97\uf978\u500b\u9ad8\u65af\u5206\u4f48 1 1 ( , ) N \u03bc \u2211 \u548c 2 2 ( , ) N \u03bc \u2211 \u7684\u76f8\u4f3c\ufa01\uff0c\u8868\u793a\u5982\u4e0b\uff1a ( ) ( ) ( ) 1 1 1 1 2 1 2 1 1 2 2 | | 1 ln tr 2 | | T KL D d \u03bc \u03bc \u03bc \u03bc \u2212 \u2212 \u239b \u239e \u2211 = + \u2211 \u2211 + \u2212 \u2211 \u2212 \u239c \u239f \u2211 \u239d \u23a0 \u2212 (",
                        "eq_num": "\u5f0f 2"
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "( | ) i l k P x \u03c9 \uff0c\u5176\u4e2d i l x \u8868\u793a\u7b2c l \u500b\u97f3\u7d20\u4e2d\u4e4b\u7b2c \u500b\u8a13\uf996\u8cc7\uf9be\u8a08\u7b97\u97f3\u7d20\u4e4b\u9593\u53d6\u5c0d\uf969\u7528\u8ddd\uf9ea\u7684\u65b9\u5f0f\u5448\u73fe\u97f3\u7d20\u9593\u5f7c\u6b64 \u7684\u95dc\u4fc2 \uff0c\u4ee5\u5efa\uf9f7\u8072\u5b78\u76f8\u4f3c\ufa01\u77e9\u9663 i log( ( | )) i l k P x \u03c9 ( ) kl N N a \u00d7 = A \u3002\u70ba\u5efa\uf9f7\u4e00\u500b\u5c0d\u7a31\u8072\u5b78\u76f8\u4f3c\ufa01\u77e9\u9663\uff0c\u6211\u5011\u5c0d\u5176 \u8a08\u7b97\u5c0d\u89d2\u5e73\u5747\u503c\u3002 1 1 1 1 log( ( | )) log( ( | )) 2 I J i j l k k l i j kl P x P x I J a \u03c9 \u03c9 = = + = \u2211 \u2211 (",
                        "eq_num": "\u5f0f 3"
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "( , 1 2 1 2 ( , ) , , . . . , , , , . . . , l l l k k k l k l k N N h v v w w w w w w = = ) (\u5f0f 4) \u8003\u616e\u97f3\u7d20\u767c\u8072\u53d7\u5230\u76f8\u9130\u97f3\u7d20\u7684\u5f71\u97ff\uff0c\u4ee5\u4e09\uf99a\u97f3\u7d20 \u70ba\u4e2d\u5fc3\u4e4b\u7a7a\u9593\u76f8\u4f3c\ufa01\u53ef\u4ee5\uf9dd\u7528\u8207\u53f3\u908a\u6587\u8108\u76f8\u95dc\u4e4b\u5411\uf97e \u53ca\u8207\u5de6\u908a\u6587\u8108\u76f8\u95dc\u4e4b\u5411\uf97e \u4e4b\u63cf\u8ff0\uff1b \u548c \u5206\u5225\u8868\u793a\uf9dd\u7528\u89c0\u6e2c\u8996\u7a97\u65bc HAL \u7a7a\u9593\u5167\u7d71\u8a08\u4e4b\u97f3\u7d20\u76f8\u95dc\u6b0a \u91cd\uff0c l \u548c k \u5206\u5225\u8868\u793a\ufa08\u8207\uf99c\u4e4b\uf96a\u5f15\u3002 , l k h l v k v l N w k N w \u5728 HAL \u7a7a\u9593\u4e2d\uff0c\u6b0a\u91cd\u4e4b\u8a08\u7b97\u9700\u8003\u616e\u6b63\u898f\u5316(normalization)\u56e0\u7d20\uff0c\u672c\uf941\u6587\uf9dd\u7528\u5728\u8cc7\u8a0a\u6aa2\uf96a\u4e2d\u76f8\u7576\u91cd\u8981\u4e4b \uf96b\uf969 tf",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "(term frequency and inverse document frequency) [13] \u5176\u4e2d\uff0c \u03b1 \u662f\u4e00\u500b\u6b0a\u91cd\u56e0\u5b50\uff0c\u8ca0\u8cac\u878d\u5408\u8072\u5b78\u76f8\u4f3c\ufa01\u548c\u524d\u5f8c\u6587\u8108\u7684\u95dc\uf997\u3002\u91dd\u5c0d\u76f8\u4f3c\ufa01\u77e9\u9663 \u548c \uff0c\uf941\u6587\u4e2d\u5c0d ",
                "cite_spans": [
                    {
                        "start": 48,
                        "end": 52,
                        "text": "[13]",
                        "ref_id": "BIBREF12"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "\uff0c\u91cd\u65b0\u4f30\u8a08\u6bcf\u500b\u5411\uf97e\u7dad\ufa01\u4e4b\u6b0a\u91cd\uff0c\u8868\u793a\u5982\u4e0b\uff1a idf \u00d7 log i i i N w w C = \u00d7 (\u5f0f 5) \u5176\u4e2d\uff0c \u6307\u5728\u5411\uf97e \u6216\u5411\uf97e \u4e2d\u7b2c i \u500b\u7dad\ufa01\u4e4b\u6b0a\u91cd\uff1b \u6307\u5728\u6240\u6709\u5411\uf97e\u4e2d\uff0c\u7b2c i \u500b\u7dad\ufa01\u4e4b\u6b0a\u91cd\uf967\u70ba\uf9b2\u7684\u5411\uf97e \u500b\uf969\uff1b \u70ba\u5411\uf97e\u7e3d\u500b\uf969\u6216\u8fa8\uf9fc\u55ae\u5143\u500b\uf969\u3002 i w l v k v i C N 3.3. \u4e09\uf99a\u97f3\u7d20\u5411\uf97e\uf97e\u5316\u7fa4\u805a\u8003\u616e\u76f8\u4f3c\ufa01\u77e9\u9663\u8cc7\uf9be\u878d\u5408 \u7d93\u904e\u524d\uf978\u5c0f\u7bc0\u5206\u6790\uff0c\u4e09\uf99a\u97f3\u7d20\u53ef\u4ee5\u5728\u8072\u5b78\u7a7a\u9593\u548c\u8a9e\u8a00\u8d85\u7a7a\u9593\u4e2d\uff0c\u7528\u5411\uf97e\u7684\u65b9\u5f0f\u8868\u793a\u5728\u7a7a\u9593\u4e2d\u7684\u76f8\u4f3c\u7a0b \ufa01\u3002\u672c\uf941\u6587\u5206\u6790\u8072\u5b78\u76f8\u4f3c\ufa01\u77e9\u9663 ( ) kl N N a \u00d7 = A \u548c\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u77e9\u9663 ( ) kl N N h \u00d7 = H \uff0c\u540c\u6642\u8003\u616e\u8072\u5b78\u76f8\u4f3c\ufa01 \u548c\u524d\u5f8c\u6587\u8108\u767c\u97f3\u7684\u7279\u6027\uff0c\u57fa\u65bc\u524d\u5f8c\u6587\u8108\u76f8\u95dc\u4e4b\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\uff0c\u5408\u4f75\u76f8\u4f3c\u4e4b\u767c\u97f3\u627e\u51fa\u6700\u70ba\ufa1d\u7c21\u6709\u6548\u7684\u591a\u8a9e\u97f3 \u7d20\u6a21\u578b\u5b9a\u7fa9\u3002\u4f5c\u6cd5\u4e0a\uff0c\uf96b\u8003\u8cc7\uf9be\u878d\u5408\u7684\u65b9\u6cd5[14]\uff0c\u672c\uf941\u6587\uf9dd\u7528\u52a0\u6cd5\u878d\u5408\u7684\u6280\u8853(sum rule)\uff0c\u7d50\u5408\uf978\u76f8\u4f3c\ufa01\u77e9 \u9663 \u548c H \uff0c\u5c07\u8072\u5b78\u76f8\u4f3c\ufa01\u548c\u524d\u5f8c\u6587\u8108\u7684\u97f3\u7d20\u7279\u5fb5\u4f5c\u6574\u5408\uff0c\u8868\u793a\u5982\u4e0b\uff1a A ,",
                        "eq_num": "(1 )"
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "(",
                        "eq_num": "(1 )"
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "\u5176\uf969\u503c\u6b63\u898f\u5283\uff0c\u5c07\u8072\u5b78\u548c\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u77e9\u9663\u7684\u5206\uf969\u7d50\u5408\uff0c\u7a31\u77e5\uf9fc\u878d\u5408\u76f8\u4f3c\ufa01\u77e9\u9663 \u70ba\u4e00\u500b\u5c0d \u7a31\u77e9\u9663\uff0c\ufa08 l \u8207\uf99c \u5747\u8868\u793a\u67d0\u4e00\u97f3\u7d20\u8207\u5176\u4ed6\u97f3\u7d20\u76f8\u4f3c\ufa01\u4e4b\u5411\uf97e\u3002\u70ba\uf9ba\u5efa\uf9f7\u6709\u6548\ufa1d\u7c21\u7684\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\u65bc\u591a\u8a9e \u8a9e \u97f3 \u8fa8 \uf9fc \u4e4b \u61c9 \u7528 \uff0c \u672c \uf941 \u6587 \uf9dd \u7528 \u5411 \uf97e \uf97e \u5316 (vector quantization, VQ) \u7684 \u65b9 \u6cd5 [12] \uff0c \u5f9e \u8cc7 \uf9be \u5206 \u6790 \u7684 \u89d2 \ufa01 (data-driven)\uff0c\u5c07\u539f\u672c\u4e09\uf99a\u97f3\u7d20\u81ea\u52d5\u5730\u4f9d\u64da\u97f3\u7d20\u76f8\u4f3c\ufa01\u5206\u6790\uff0c\u5408\u4f75\u591a\u8a9e\u97f3\u7d20\u5b9a\u7fa9\u3002\u5411\uf97e\uf97e\u5316\u70ba\u662f\u4e00\u7a2e\u975e\u76e3\u7763 \u5f0f\u7684\u7fa4\u96c6\u5206\u6790\u65b9\u6cd5\uff0c\u53ef\u4ee5\u5c07\u5206\u6563\u7684\u8cc7\uf9be\u7fa4\u96c6\u6210\u6709\u610f\u7fa9\u7684\uf9d0\u5225\u3002\u4e09\uf99a\u97f3\u7d20\u5728\u76f8\u4f3c\ufa01\u77e9\u9663\u5206\u6790\u5f8c\uff0c\u53ef\u7528\u5411\uf97e\u65b9 \u5f0f\u8868\u793a\u5176\u7a7a\u9593\u5ea7\u6a19\uff0c\uf941\u6587\u5f15\u7528[15]\u5728\u77e9\u9663\u4e2d\uff0c\uf978\u5411\uf97e\u593e\u89d2\u7684\u8a08\u7b97\u65b9\u6cd5\uff0c\u56e0\u6b64\uf978\u97f3\u7d20\u7684\u76f8\u4f3c\ufa01\u8a08\u7b97\u70ba \uff0c \u8a08\u7b97\u5982\u4e0b\uff1a A H ( ) kl N N s \u00d7 = S k ( , ) l k c s s 1 2 2 1 1 ( , ) N l k i i l k i l k N N l k l k l k s s s s c s s s s s s = = = \u00d7 \u2022 = = \u22c5 \u00d7 \u2211 \u2211 \u2211 (",
                        "eq_num": "\u5f0f 7"
                    }
                ],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "Automatic Segmentation and Identification of Mixed-Language Speech Using Delta-BIC and LSA-Based GMMs",
                "authors": [
                    {
                        "first": "Chung-Hsien",
                        "middle": [],
                        "last": "Wu",
                        "suffix": ""
                    },
                    {
                        "first": "Yu-Hsien",
                        "middle": [],
                        "last": "Chiu",
                        "suffix": ""
                    },
                    {
                        "first": "Chi-Jiun",
                        "middle": [],
                        "last": "Shia",
                        "suffix": ""
                    },
                    {
                        "first": "Chun-Yu",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "IEEE Transactions on audio, speech, and language processing",
                "volume": "14",
                "issue": "1",
                "pages": "266--276",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chung-Hsien Wu, Yu-Hsien Chiu, Chi-Jiun Shia, and Chun-Yu Lin, 2006. Automatic Segmentation and Identification of Mixed-Language Speech Using Delta-BIC and LSA-Based GMMs. IEEE Transactions on audio, speech, and language processing, vol. 14, no. 1, pp. 266-276.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Verbmobil: Foundations of Speech-to-Speech Translation",
                "authors": [
                    {
                        "first": "Alex",
                        "middle": [],
                        "last": "Waibel",
                        "suffix": ""
                    },
                    {
                        "first": "Hagen",
                        "middle": [],
                        "last": "Soltau",
                        "suffix": ""
                    },
                    {
                        "first": "Tanja",
                        "middle": [],
                        "last": "Schultz",
                        "suffix": ""
                    },
                    {
                        "first": "Thomas",
                        "middle": [],
                        "last": "Schaaf",
                        "suffix": ""
                    },
                    {
                        "first": "Florian",
                        "middle": [],
                        "last": "Metze",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Alex Waibel, Hagen Soltau, Tanja Schultz, Thomas Schaaf, and Florian Metze, 2000. Multilingual Speech Recognition. Chapter in Verbmobil: Foundations of Speech-to-Speech Translation, Springer-Verlag.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Vocabulary Independent Discriminative Utterance Verification for Nonkeyword Rejection in Subword based Speech Recognition",
                "authors": [
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Rafid",
                        "suffix": ""
                    },
                    {
                        "first": "Chin-Hui",
                        "middle": [],
                        "last": "Sukkar",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Lee",
                        "suffix": ""
                    }
                ],
                "year": 1996,
                "venue": "IEEE Transactions on Speech and Audio Processing",
                "volume": "4",
                "issue": "6",
                "pages": "420--429",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Rafid A. Sukkar and Chin-Hui Lee, 1996. Vocabulary Independent Discriminative Utterance Verification for Nonkeyword Rejection in Subword based Speech Recognition. IEEE Transactions on Speech and Audio Processing, vol. 4, no. 6, pp. 420-429.",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "Generation of robust phonetic set and decision tree for Mandarin using chi-square testing",
                "authors": [
                    {
                        "first": "Yeou-Jiunn",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "Chung-Hsien",
                        "middle": [],
                        "last": "Wu",
                        "suffix": ""
                    },
                    {
                        "first": "Yu-Hsien",
                        "middle": [],
                        "last": "Chiu",
                        "suffix": ""
                    },
                    {
                        "first": "Hsiang-Chuan",
                        "middle": [],
                        "last": "Liao",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Speech Communication",
                "volume": "38",
                "issue": "",
                "pages": "349--364",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Yeou-Jiunn Chen, Chung-Hsien Wu, Yu-Hsien Chiu, and Hsiang-Chuan Liao, 2002. Generation of robust phonetic set and decision tree for Mandarin using chi-square testing. Speech Communication, vol. 38(3-4), pp. 349-364.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "Mathews' Chinese-English Dictionary, Caves",
                "authors": [
                    {
                        "first": "R",
                        "middle": [
                            "H"
                        ],
                        "last": "Mathews",
                        "suffix": ""
                    }
                ],
                "year": 1975,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Mathews, R. H., 1975. Mathews' Chinese-English Dictionary, Caves, 13th printing.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Computer-Coded Phonemic Notation of Individual Languages of the European Community",
                "authors": [
                    {
                        "first": "J",
                        "middle": [
                            "C"
                        ],
                        "last": "Wells",
                        "suffix": ""
                    }
                ],
                "year": 1989,
                "venue": "J. IPA",
                "volume": "19",
                "issue": "",
                "pages": "32--54",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "J. C. Wells, 1989. Computer-Coded Phonemic Notation of Individual Languages of the European Community. J. IPA, 19, pp. 32-54.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "ASCII Phonetic Symbols for the World's Languages: Worldbet",
                "authors": [
                    {
                        "first": "James",
                        "middle": [
                            "L"
                        ],
                        "last": "Hieronymus",
                        "suffix": ""
                    }
                ],
                "year": 1993,
                "venue": "Journal of the International Phonetic Association",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "James L. Hieronymus, 1993. ASCII Phonetic Symbols for the World's Languages: Worldbet. Journal of the International Phonetic Association.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Phone clustering using the Bhattacharyya distance",
                "authors": [
                    {
                        "first": "Brian",
                        "middle": [],
                        "last": "Mak",
                        "suffix": ""
                    },
                    {
                        "first": "Etienne",
                        "middle": [],
                        "last": "Barnard",
                        "suffix": ""
                    }
                ],
                "year": 1996,
                "venue": "Proc. ICSLP",
                "volume": "",
                "issue": "",
                "pages": "2005--2008",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Brian Mak and Etienne Barnard, 1996. Phone clustering using the Bhattacharyya distance. in Proc. ICSLP, pp. 2005-2008.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "A Distance Measure Between GMMs Based on the Unsented Transform and its Application to Speaker Recognition",
                "authors": [
                    {
                        "first": "Jacob",
                        "middle": [],
                        "last": "Goldberger",
                        "suffix": ""
                    },
                    {
                        "first": "Hagai",
                        "middle": [],
                        "last": "Aronowitz",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Proc. of EUROSPEECH 2005",
                "volume": "",
                "issue": "",
                "pages": "1985--1988",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jacob Goldberger and Hagai Aronowitz, 2005. A Distance Measure Between GMMs Based on the Unsented Transform and its Application to Speaker Recognition. in Proc. of EUROSPEECH 2005, pp. 1985-1988, Lisbon, Portugal.",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "The Sound Pattern of English",
                "authors": [
                    {
                        "first": "N",
                        "middle": [],
                        "last": "Chomsky",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Halle",
                        "suffix": ""
                    }
                ],
                "year": 1968,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chomsky, N. and Halle, M., 1968. The Sound Pattern of English. New York: Harper & Row.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Modelling parsing constraints with high-dimensional context space",
                "authors": [
                    {
                        "first": "C",
                        "middle": [],
                        "last": "Burgess",
                        "suffix": ""
                    },
                    {
                        "first": "K",
                        "middle": [],
                        "last": "Lund",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Language and Cognitive Processes",
                "volume": "12",
                "issue": "",
                "pages": "177--210",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Burgess, C. and Lund, K., 1997. Modelling parsing constraints with high-dimensional context space. Language and Cognitive Processes, 12:177-210.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Quantization",
                "authors": [
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Robert",
                        "suffix": ""
                    },
                    {
                        "first": "David",
                        "middle": [
                            "L"
                        ],
                        "last": "Gray",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Neuhoff",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "IEEE Transactions on Information Theory",
                "volume": "44",
                "issue": "6",
                "pages": "2325--2383",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Robert M. Gray and David L. Neuhoff, 1998. Quantization. IEEE Transactions on Information Theory, vol. 44, no. 6, pp. 2325-2383.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "Term-weighting Approaches in Automatic Text Retrieval",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Salton",
                        "suffix": ""
                    },
                    {
                        "first": "C",
                        "middle": [],
                        "last": "Buckley",
                        "suffix": ""
                    }
                ],
                "year": 1988,
                "venue": "Information Processing Management",
                "volume": "24",
                "issue": "5",
                "pages": "513--523",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "G. Salton and C. Buckley, 1988. Term-weighting Approaches in Automatic Text Retrieval. Information Processing Management, vol. 24, no. 5, pp. 513-523.",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "On Combining Classifiers",
                "authors": [
                    {
                        "first": "Josef",
                        "middle": [],
                        "last": "Kittler",
                        "suffix": ""
                    },
                    {
                        "first": "Mohamad",
                        "middle": [],
                        "last": "Hatef",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [
                            "W"
                        ],
                        "last": "Robert",
                        "suffix": ""
                    },
                    {
                        "first": "Jiri",
                        "middle": [],
                        "last": "Duin",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Matason",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "IEEE Transactions on Pattern Analysis and Machine Intelligence",
                "volume": "20",
                "issue": "3",
                "pages": "226--239",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Josef Kittler, Mohamad Hatef, Robert P.W. Duin, and Jiri MatasOn, 1998. On Combining Classifiers. IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 20, no. 3, pp. 226-239.",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "Exploiting latent semantic information in statistical language modeling",
                "authors": [
                    {
                        "first": "Jerome",
                        "middle": [
                            "R"
                        ],
                        "last": "Bellegarda",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "Proc. IEEE",
                "volume": "88",
                "issue": "",
                "pages": "1279--1296",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jerome R. Bellegarda, 2000. Exploiting latent semantic information in statistical language modeling. Proc. IEEE, vol. 88, no. 8, pp. 1279-1296.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "A modified K-means clustering algorithm for use in isolated work recognition",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Jay",
                        "suffix": ""
                    },
                    {
                        "first": "Lawrence",
                        "middle": [
                            "R"
                        ],
                        "last": "Wilpon",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Rabiner",
                        "suffix": ""
                    }
                ],
                "year": 1985,
                "venue": "IEEE Transactions on Acoustics, Speech, and Signal Proc",
                "volume": "33",
                "issue": "3",
                "pages": "587--594",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jay G. Wilpon and Lawrence R. Rabiner, 1985. A modified K-means clustering algorithm for use in isolated work recognition. IEEE Transactions on Acoustics, Speech, and Signal Proc., vol. 33, no. 3, pp. 587-594.",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "Speech feature smoothing for robust ASR",
                "authors": [
                    {
                        "first": "Chia-Ping",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "Jeff",
                        "middle": [],
                        "last": "Bilmes",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [
                            "W"
                        ],
                        "last": "Daniel",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Ellis",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Proc. ICASSP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chia-Ping Chen, Jeff Bilmes and Daniel P. W. Ellis, 2005. Speech feature smoothing for robust ASR. in Proc. ICASSP, Philadelphia PA.",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "Statistical Methods for Speech Recognition",
                "authors": [
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Johnston",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Johnston, D., 1997. Statistical Methods for Speech Recognition. The MIT Press, Cambridge, MA.",
                "links": null
            },
            "BIBREF19": {
                "ref_id": "b19",
                "title": "Progress in dynamic programming search for LVCSR",
                "authors": [
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Ney",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Ortmanns",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "Proceedings of the IEEE",
                "volume": "88",
                "issue": "",
                "pages": "1224--1240",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "H. Ney and S. Ortmanns, 2000. Progress in dynamic programming search for LVCSR. Proceedings of the IEEE, vol. 88, no. 8, pp. 1224-1240.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF1": {
                "uris": null,
                "type_str": "figure",
                "num": null,
                "text": "\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\u4e4b\u591a\u8a9e\u8fa8\uf9fc\u97f3\u7d20\u6b63\u78ba\uf961 (\u62ec\u5f27\u5167\u8868\u8fa8\uf9fc\u55ae\u5143\u4e4b\u500b\uf969) ====================== English Across Taiwan, EAT ====================== --------------------------------------Triphone Tree-Search Results------------------------------------ACCURACY"
            },
            "TABREF0": {
                "type_str": "table",
                "html": null,
                "text": "\u7684\u65b9\u6cd5\uff0c\u8a08\u7b97\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\u9593\u7684\u8ddd \uf9ea\uff0c\u6c7a\u5b9a\u76f8\u4f3c\ufa01\u4ee5\u5b9a\u7fa9\u591a\u8a9e\u97f3\u7d20\u96c6\u3002\u6b64\u4f5c\u6cd5\u4e0a\uff0c\u540c\u6642\u8003\u616e\u591a\u8a9e\u97f3\u7d20\u9593\uf96b\uf969\u5206\u4eab\u7684\u7279\u6027\uff0c\u4e26\uf9dd\u7528\u8cc7\uf9be\u7d71\u8a08\u5206 \u6790\u6c7a\u5b9a\u97f3\u7d20\u5b9a\u7fa9\u3002\u4f46\u662f\u7f3a\u9ede\u5728\u65bc\u8a08\u7b97\u6a21\u578b\uf96b\uf969\u9593\u7684\u8ddd\uf9ea\uff0c\u8207\u5be6\u969b\u8fa8\uf9fc\u6f14\u7b97\u6cd5\u5728\u57f7\ufa08\u6642\uff0c\u6240\u8003\u616e\u7684\u8072\u5b78\u76f8\u4f3c \ufa01(acoustic likelihood)\uf967\u7b26\u3002 \u672c\uf941\u6587\u63a2\u8a0e\u4e2d\u82f1\u6587\u4e4b\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u7814\u7a76\uff0c\u5f9e\u4e2d\u82f1\u6587\u57fa\u672c\u97f3\u7d20\u4f5c\u5206\u6790\u3002\u4e2d\u6587\u53ef\u4ee5\u5206\u70ba 37 \u500b\u97f3\u7d20\uff0c\u82f1 \u6587\u53ef\u5206\u70ba 39 \u500b\u97f3\u7d20\u3002\u8003\u616e\u8a9e\u97f3\u767c\u97f3\u5171\u8072\u7684\u73fe\u8c61(co-articulation)\uff0c\u672c\uf941\u6587\u5b9a\u7fa9\u524d\u5f8c\u6587\u76f8\u95dc\u4e4b\u4e09\uf99a\u97f3\u7d20\u6a21\u578b (contextual tri-phone models)\uff0c\u9032\u4e00\u6b65\u5c0d\u8a9e\u97f3\u767c\u97f3\u76f8\u4f3c\ufa01\u4f5c\u8072\u5b78\u76f8\u4f3c\ufa01(acoustic likelihood)\u5206\u6790\u3002\u6b64\u5916\uf901\u5c0e\u5165 \u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u5206\u6790(hyperspace analog to language, HAL)\uff0c\u8003\uf97e\u4e09\uf99a\u97f3\u8fa8\uf9fc\u55ae\u5143\u524d\u5f8c\u6587\u8108\u4e4b\u95dc\u4fc2\uff0c\u4ee5\u6539 \u5584\u904e\u53bb\u55ae\u7d14\u8003\uf97e\u6a21\u578b\uf96b\uf969\u8072\u5b78\u76f8\u4f3c\ufa01\uf92d\uf97e\u6e2c\u8a9e\u97f3\u97f3\u7d20\u9593\u76f8\u4f3c\ufa01\u4e4b\u65b9\u5f0f\uff0c\u4ee5\u6c7a\u5b9a\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\uff0c\u7b26\u5408\u8a9e\u97f3\u767c \u97f3\u4e2d\u53d7\u524d\u5f8c\u6587\u5f71\u97ff\u4e4b\u7279\u6027\u3002\u6700\u5f8c\uff0c\u4ee5\u8cc7\uf9be\u878d\u5408\u7684\u6280\u8853\u5408\u4f75\u5b9a\u7fa9\u767c\u97f3\u76f8\u4f3c\u7684\u97f3\u7d20\u3002\u5be6\u9a57\u8a55\u4f30\uff0c\uf9dd\u7528\u81ea\ufa08\u958b\u767c \u7684\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u7cfb\u7d71\uff0c\u4f7f\u7528\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b(hidden Markov model, HMM)\uff0c\u5efa\uf9f7\u4ee5\u97f3\u7d20\u70ba\u57fa\u790e\u7684\u8072\u5b78\u6a21",
                "content": "<table><tr><td colspan=\"3\">\u5b9a\u7fa9\u548c 39 \u500b\u82f1\u6587\u97f3\u7d20\u5b9a\u7fa9\u3002\u6b64\u65b9\u6cd5\u7d50\u5408\u4e2d\u82f1\uf978\u7a2e\u8a9e\u8a00\u4e4b\u97f3\u7d20\uff0c\u5efa\uf9f7\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u8072\u5b78\u6a21\u578b\u3002\u4f5c\u6cd5\u4e0a\u7684</td></tr><tr><td colspan=\"3\">\u7f3a\u9ede\uff0c\u5728\u65bc\u5404\u76ee\u6a19\u8a9e\u8a00\u4e2d\u76f8\u4f3c\u4e4b\u97f3\u7d20\uff0c\u6a21\u578b\uf96b\uf969\u7121\u6cd5\u5206\u4eab\uff0c\u800c\u4e14\u7576\u9700\u8981\u7d50\u5408\u7684\u76ee\u6a19\u8a9e\u8a00\u8b8a\u591a\u7684\u6642\u5019\uff0c\u6240\u9700</td></tr><tr><td colspan=\"2\">\u8981\u5b9a\u7fa9\u7684\u97f3\u7d20\u6a21\u578b\u6703\u5927\uf97e\u96a8\u4e4b\u589e\u52a0\u3002</td><td/></tr><tr><td colspan=\"2\">2.2. \u4ee5 IPA \u70ba\u57fa\u6e96\u5b9a\u7fa9\u591a\u8a9e\u97f3\u7d20</td><td/></tr><tr><td colspan=\"3\">\u7b2c\u4e8c\u7a2e\u591a\u8a9e\u97f3\u7d20\u5b9a\u7fa9\u65b9\u5f0f\u662f\u57fa\u65bc\u5c08\u5bb6\u7684\u77e5\uf9fc\uff0c\u5c07\u500b\u5225\u7368\uf9f7\u7684\u55ae\u4e00\u8a9e\u8a00\u5c0d\u61c9\u5230 IPA \u6a19\u6e96\u7684\u7b26\u865f\u5b9a\u7fa9\uff0c\u85c9</td></tr><tr><td colspan=\"3\">\u6b64\u5404\u8a9e\u8a00\u9593\u53ef\u4ee5\u5206\u4eab\u76f8\u540c\u7684\u97f3\u7d20\u5b9a\u7fa9\u3002\u5982(\u8868 2)\u6240\u793a\u662f\u4ee5 IPA \u70ba\u6a19\u6e96\u4e4b\u4e2d\u82f1\u591a\u8a9e\u97f3\u7d20\u7684\u5b9a\u7fa9\u3002</td></tr><tr><td/><td>\u8868 2. \u4ee5 IPA \u70ba\u6a19\u6e96\u4e4b\u4e2d\u82f1\u591a\u8a9e\u97f3\u7d20\u5b9a\u7fa9</td><td/></tr><tr><td>\u97f3\u7d20\uf9d0\u5225</td><td colspan=\"2\">IPA \u70ba\u6a19\u6e96\u4e4b\u4e2d\u82f1\u591a\u8a9e\u97f3\u7d20</td></tr><tr><td>\u6709\u8072\u7834\uf9a0\u97f3</td><td>B, D, G</td><td/></tr><tr><td>\u7121\u8072\u7834\uf9a0\u97f3</td><td>P, T, K</td><td/></tr><tr><td>\u6469\u64e6\u97f3</td><td>F, S, SH, H, X, V, TH, DH</td><td/></tr><tr><td>\uf96c\u64e6\u97f3</td><td>Z, ZH, C, CH, J, Q, CH, JH</td><td/></tr><tr><td>\u9f3b\u97f3</td><td>M, N, NG</td><td/></tr><tr><td>\uf9ca\u97f3</td><td>R, L</td><td/></tr><tr><td>\uf904\u97f3</td><td>W, Y</td><td/></tr><tr><td>\u524d\u90e8\u6bcd\u97f3</td><td>I, ER, V, EI, IH, EH, AE</td><td/></tr><tr><td colspan=\"3\">\u4e2d\u90e8\u6bcd\u97f3 \u578b\uff0c\u4e26\u914d\u5408\u591a\u8a9e\u8a9e\u8a00\u6a21\u578b\u548c\u591a\u8a9e\u767c\u97f3\u8fad\u5178\u6587\u6cd5\u6a39\uff0c\u9032\ufa08\uf99a\u7e8c\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u3002 ENG, AN, ANG, EN, AH, UH \u63a5\u4e0b\uf92d\u7684\u6587\u7ae0\u7d50\u69cb\u5c07\u5206\u5225\u63a2\u8a0e\u5982\u4e0b\uff1a\u7b2c\u4e8c\u7bc0\uff0c\u63a2\u8a0e\u904e\u53bb\u5c0d\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u7814\u7a76\u3002\u7b2c\u4e09\u7bc0\uff0c\uf96f\u660e\uf941\u6587\u65b9 \u80cc\u90e8\u5713\u5507\u6bcd\u97f3 O</td></tr><tr><td colspan=\"3\">\u6cd5\u5efa\uf9f7\ufa1d\u7c21\u6709\u6548\u7684\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\u65bc\u81ea\u52d5\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u61c9\u7528\u3002\u7b2c\u56db\u7bc0\uff0c\u91dd\u5c0d\u672c\uf941\u6587\u6240\u63d0\u65b9\u6cd5\u5efa\uf9f7\u4e4b\u591a\u8a9e\u97f3\u7d20\u6a21 \u80cc\u90e8\u975e\u5713\u5507\u6bcd\u97f3 A, U, OU, AI, AO, E, EE, OY, AW</td></tr><tr><td colspan=\"3\">\u578b\u9032\ufa08\u8fa8\uf9fc\u7d50\u679c\u8a55\u4f30\uff0c\u5be6\u9a57\u4e26\u8207\u4e4b\u524d\u65b9\u6cd5\u6bd4\u8f03\u3002\u7b2c\u4e94\u7bc0\u662f\u8a0e\uf941\uf96f\u660e\u8207\u7d50\uf941\u3002</td></tr><tr><td colspan=\"3\">(\u8868 2) \u5167\u4e4b\u97f3\u7d20\uf9d0\u5225\uf96b\u8003 Chomsky \u5b9a\u7fa9[10]\u3002\u5982\u6b64\u898f\u5247\u5730\u5c07\u4e2d\u82f1\uf978\u7a2e\u8a9e\u8a00\u7684\u97f3\u7d20\u7d50\u5408\uff0c\u5171\u8a08\u6709 52 \u500b\u4e2d\u82f1\u96d9 2. \u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u97f3\u7d20\u5b9a\u7fa9\u76f8\u95dc\u7814\u7a76 \u8a9e\u97f3\u7d20\u5b9a\u7fa9\u3002\u4f5c\u6cd5\u4e0a\u53ef\u4ee5\u6709\u6548\u5730\u5c07\u90e8\u5206\u7684\u4e2d\u82f1\u6587\u97f3\u7d20\u5408\u4f75\uff0c\u5171\u4eab\u8a9e\u8a00\u9593\u5f7c\u6b64\u7684\u5171\u540c\u97f3\u7d20\uff0c\u6e1b\u5c11\u8a9e\u97f3\u97f3\u7d20\u6a21</td></tr><tr><td colspan=\"3\">\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u97f3\u7d20\u5b9a\u7fa9\u7684\u65b9\u6cd5\uff0c\u4e3b\u8981\u53ef\u5206\u70ba\u4e09\u7a2e\u65b9\u5f0f\uff1a(\u4e00)\u76f4\u63a5\u7d50\u5408\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20\u5b9a\u7fa9\uff1b(\u4e8c) \u578b\u7684\u5b9a\u7fa9\u548c\u8a13\uf996\u3002\u4f46\u6b64\u4f5c\u6cd5\u7684\u7f3a\u9ede\u662f\u5efa\u69cb\u5728\u5c08\u5bb6\u77e5\uf9fc\u7684\u5206\u6790\uff0c\u800c\u975e\u5f9e\u8cc7\uf9be\u7279\u6027\u7d71\u8a08\u7684\u89d2\ufa01\u5b9a\u7fa9\u3002\u4e5f\u5c31\u662f\uf96f\uff0c</td></tr><tr><td colspan=\"3\">\u4f9d\u64da\u570b\u969b\u97f3\u7d20\u6a19\u6e96\u5b9a\u7fa9\uff0c\u627e\u51fa\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20\uf997\u96c6\uff1b(\u4e09)\u5f9e\u8cc7\uf9be\u5206\u6790\u7684\u89d2\ufa01(data-driven)\uff0c\u5408\u4f75\u500b\u5225\u55ae \u76f4\u63a5\u5c0d\u7167 IPA \u5b9a\u7fa9\u7522\u751f\u7684\u591a\u8a9e\u97f3\u7d20\u96c6\uff0c\u4e26\u6c92\u6709\u8003\u616e\u5230\u97f3\u7d20\u6a21\u578b\u9593\u983b\u8b5c\u7279\u6027\u3002\u5c08\u5bb6\u77e5\uf9fc\u5206\u6790\u7684\u591a\u8a9e\u97f3\u7d20\u96c6\uff0c</td></tr><tr><td colspan=\"3\">\u4e00\u8a9e\u8a00\u4e4b\u76f8\u4f3c\u97f3\u7d20\u3002\u73fe\u5206\u5225\u4ecb\u7d39\u5982\u4e0b\uff1a \u8207\u6700\u5f8c\u9032\ufa08\u8a9e\u97f3\u8fa8\uf9fc\uff0c\u5f9e\u8cc7\uf9be\u5206\u6790\u89d2\ufa01\u5efa\uf9f7\u7684\u7d71\u8a08\u6a21\u578b\u8a08\u7b97\uf967\u4e00\u81f4\u3002\u56e0\u6b64\uff0c\u63a1\u7528\u76f4\u63a5\u5c0d\u7167 IPA \u5b9a\u7fa9\u4e4b\u591a\u8a9e</td></tr><tr><td colspan=\"2\">2.1. \u76f4\u63a5\u7d50\u5408\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20 \u97f3\u7d20\u6a21\u578b\u4e26\uf967\u80fd\u78ba\u5be6\u5730\u5448\u73fe\u7d71\u8a08\u8a13\uf996\u8cc7\uf9be\u4e0a\u7684\u5206\u4f48\u3002</td><td/></tr><tr><td colspan=\"2\">\u5982(\u8868 1)\u6240\u793a\uff0c\u6bd4\u7167\u4e2d\u6587\u548c\u82f1\u6587\u55ae\u4e00\u8a9e\u8a00\u97f3\u7d20\u7684\u5b9a\u7fa9\u3002 2.3. \uf97e\u6e2c\u97f3\u7d20\u76f8\u4f3c\ufa01\u5b9a\u7fa9\u591a\u8a9e\u97f3\u7d20\u96c6</td><td/></tr><tr><td colspan=\"3\">\u9664\uf9ba\u76f4\u63a5\u6df7\u5408\u591a\u8a9e\u97f3\u7d20\u5b9a\u7fa9\uff0c\u4ee5\u53ca\uf9dd\u7528 IPA \u570b\u969b\u6a19\u6e96\u5b9a\u7fa9\u7684\u591a\u8a9e\u97f3\u7d20\uff0c\u904e\u53bb\u7814\u7a76\u4e5f\u66fe\uf9dd\u7528\u4f30\u6e2c\u4e09\uf99a\u97f3 \u8868 1. \u7d50\u5408\u4e2d\u82f1\u6587\u97f3\u7d20\u5b9a\u7fa9 \u7d20\u6a21\u578b\u9593\u7684\u76f8\u4f3c\ufa01\uff0c\u4ee5 HMM \u6a21\u578b\uf96b\uf969\u8ddd\uf9ea\u8a08\u7b97\uff0c\uf9dd\u7528\u905e\u8ff4\u65b9\u6cd5\u5408\u4f75\u4e09\uf99a\u97f3\u7d20\u6a21\u578b (triphone)\uff0c\u5efa\u69cb\u51fa\u591a</td></tr><tr><td colspan=\"3\">\u97f3\u7d20\uf9d0\u5225 \u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u7684\u97f3\u7d20\u96c6[8][9]\u3002\uf978\u500b\u9ad8\u65af\u5206\u4f48\u7684\u76f8\u4f3c\u53ef\u4ee5\uf9dd\u7528\u5e73\u5747\u503c\u548c\u8b8a\uf962\uf969\u51fd\uf969\uff0c\uf92d\u63cf\u8ff0\u5f7c\u6b64\u7684\u76f8\u4f3c\u7a0b\ufa01\u3002 \u4e2d\u6587 \u82f1\u6587 \u6709\u8072\u7834\uf9a0\u97f3 b_M, d_M, g_M b, d, g \uf9dd\u7528 Bhattacharyya distance [8]\uf92d\u8a08\u7b97\u97f3\u7d20\u6a21\u578b\u9593\u7684\u8ddd\uf9ea \uff0c\u8868\u793a\u5982\u4e0b\uff1a bha</td></tr><tr><td>\u7121\u8072\u7834\uf9a0\u97f3</td><td>p_M, t_M, k_M</td><td>p, t, k</td></tr><tr><td>\u6469\u64e6\u97f3</td><td>f_M, s_M, sh_M, h_M, x_M</td><td>f, v, th, dh, s, sh, hh</td></tr><tr><td>\uf96c\u64e6\u97f3</td><td>c_M, ch_M, j_M, q_M, z_M, zh_M</td><td>ch, jh, z, zh</td></tr><tr><td>\u9f3b\u97f3</td><td>m_M, n_M</td><td>m, n, ng</td></tr><tr><td>\uf9ca\u97f3</td><td>r_M, l_M</td><td>r, l</td></tr><tr><td>\uf904\u97f3</td><td/><td>w, y</td></tr><tr><td>\u524d\u90e8\u6bcd\u97f3</td><td>i_M, v_M, ei_M, er_M</td><td>ih, eh, ae, iy, ey</td></tr><tr><td>\u4e2d\u90e8\u6bcd\u97f3</td><td>an_M, ang_M, en_M, eng_M</td><td>ah, uh, er</td></tr><tr><td>\u80cc\u90e8\u5713\u5507\u6bcd\u97f3</td><td>o_M</td><td>ao</td></tr></table>",
                "num": null
            },
            "TABREF3": {
                "type_str": "table",
                "html": null,
                "text": "\u9ea5\u514b\u98a8\u8a9e\uf9be\uf93f\u88fd 16KHz \u53d6\u6a23\u983b\uf961 16bits \u7684\u53d6\u6a23\u9ede\u97f3\u6a94\uff0c\u96fb\u8a71\u8a9e\uf9be\uf93f\u88fd 8KHz \u53d6\u6a23\u983b\uf961 16bits \u7684\u53d6\u6a23\u9ede\u97f3\u6a94\uff0c \u5176\u4e2d\u96fb\u8a71\u8a9e\uf9be\u53c8\u53ef\u7d30\u5206\u70ba\u56fa\u5b9a\u5f0f\u96fb\u8a71(PSTN)\u8a9e\uf9be\u53ca\ufa08\u52d5\u96fb\u8a71(GSM )\u8a9e\uf9be\uff0c\u96fb\u8a71\u8a9e\uf9be\u90e8\u4efd\u662f\u900f\u904e Dialogic \u96fb \u8a71\u8a9e\u97f3\u4ecb\u9762\u5361\uff0c\uf93f\u5f97\u7684 8KHz\uff0c8Bits\uff0cMulaw \u683c\u5f0f\u7684\u53d6\u6a23\u9ede,\u7d93\u7a0b\u5f0f\u8f49\u6210 8KHz\uff0c16bits\uff0cpcm \u683c\u5f0f\u7684\u53d6\u6a23\u9ede\uff1b \u9ea5\u514b\u98a8\u8a9e\uf9be\u662f\u7531\u500b\u4eba\u96fb\u8166\u53ca\u9ea5\u514b\u98a8\uff0c\u76f4\u63a5\u5f9e\u500b\u4eba\u96fb\u8166\u7684\u97f3\u6548\u5361\uf93f\u88fd 16KHz\uff0c16bits \u7684\u8072\u97f3\u8a0a\u865f\u3002\u6700\u5f8c\u5c07\u6240 \u6709\u53d6\u6a23\u9ede\u4ee5 \u8868\u793a\u591a\u8a9e\u97f3\u7d20\u5b9a\u7fa9\u7684 HMM \u6a21\u578b\uff0c\u7814\u7a76\u4e0a\u61c9\u7528 3 \u500b\uf9fa\u614b(state)\uf92d\u63cf\u8ff0\u6bcf\u4e00\u500b HMM \u6a21\u578b\uff0c\u6bcf \u4e00\u500b\uf9fa\u614b\u5305\u542b\u6709 16 \u500b\u9ad8\u65af(mixture)\u3002\u6b64\u591a\u8a9e\u4e4b\u6a39\uf9fa\u7d50\u69cb\u767c\u97f3\u8fad\u5178\u8209\uf9b5\u5171\u6709\uff1asay(sil_S_EY, S_EY_sil)\u3001\u5df4\uf989 (sil_B_IY, B_IY_L, B_L_IY, L_IY_sil)\u3001top(sil_T_AA, T_AA_P, AA_P_sil)\u7b49\u8a5e\u7d44\u3002\u672c\u5be6\u9a57\u5408\u4f75\u82f1\u6587\u767c\u97f3\u8fad\u5178 \u8207\u4e2d\u6587\u767c\u97f3\u8fad\u5178\uff0c\u5efa\uf9f7\u5305\u542b 29,104 \u500b\u4e2d\u82f1\u6587\u8a5e\u4e4b\u591a\u8a9e\u767c\u97f3\u8fad\u5178\u3002\u672c\u5716\u793a\u8209\uf9b5\uf96f\u660e\u7531\u975c\u97f3(silence, sil)\u70ba\u8d77\u9ede\uff0c \u8fa8\uf9fc\u591a\u8a9e\u8a9e\uf906\"say (sil_S_EY, S_EY_sil) \u5df4\uf989 (sil_B_IY, B_IY_L, IY_L_IY, L_IY_sil)\"\u70ba\uf9b5\uff0c \u70ba\u6a39\u7684\u6839\u7bc0 \uf9dd\u7528\u524d\u5f8c\u6587\u8108\u5206\u6790\u65b9\u6cd5 HAL \u6bd4\u8072\u5b78\u76f8\u4f3c\ufa01\u65b9\u6cd5 ACL \u6709\u8f03\u9ad8\u7684\u6e96\u78ba\uf961\uff0c\u800c\u540c\u6642\u7d50\u5408\u8072\u5b78\u76f8\u4f3c\ufa01\u8207\u524d\u5f8c\u6587\u8108 \u5206\u6790 FUN \u53ef\u4ee5\u6709\u6700\u4f73\u7684\u8fa8\uf9fc\u6548\u679c\u3002\u7576\u7fa4\u96c6\uf969 16 Y = \u6642\uff0c\uf941\u6587\u6240\u63d0\u4e4b\u65b9\u6cd5(FUN)\u53ef\u4ee5\u6709\u6700\u597d\u7684\u8fa8\uf9fc\u6548\u679c\u3002\u56e0 \u6b64\uff0c\uf941\u6587\u8a2d\u5b9a\u7fa4\u96c6\u5206\u6790 16 Y = \uff0c\u7fa4\u96c6\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\uff0c\u5206\u6790\u5982(\u5716 5)\u6240\u793a\u3002\u7d93\u904e\u5206\uf9d0\u5b8c\u5f8c\uff0c\u5404\u500b IPA \u5b9a\u7fa9\u4e4b\u97f3 \u7d20\u4e2d\uff0c\u6240\u5305\u542b\u4e4b\u4e09\uf99a\u97f3\u500b\uf969\u3002\u7531\u4e0a\u5716\u53ef\u77e5\uff0c\u4ee5 55 \u500b IPA \u6a19\u6e96\u5b9a\u7fa9\u6240\u7522\u751f\u4e4b 997 \u500b\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\uff0c\uf9dd\u7528\u8cc7 \uf9be\u878d\u5408\u65b9\u6cd5\u53ef\u4ee5\u5408\u4f75\u70ba 260 \u500b\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\u3002",
                "content": "<table><tr><td colspan=\"4\">\u900f\u904e\u591a\u8a9e\u767c\u97f3\u8fad\u5178\uff0c\u53ef\u4ee5\u5efa\u69cb\u51fa\u591a\u8a9e\u767c\u97f3\u4e4b\u6587\u6cd5\u6a39(grammar tree) [20]\u3002\u5982\u4e0b(\u5716 4)\u6240\u793a\u3002\u5728\u8fa8\uf9fc\u7684\uf9ca\u7a0b\u4e0a\uff0c</td></tr><tr><td colspan=\"4\">wav \u683c\u5f0f\u97f3\u6a94\u5132\u5b58\u3002\u672c\uf941\u6587\u7814\u7a76\u63a1\u7528\u9ea5\u514b\u98a8\u8a9e\uf9be\u90e8\u5206\u3002 \u6bcf\u4f4d\u8a9e\u8005\u6536\uf93f 80 \uf906\u8a9e\u97f3\u8a9e\uf9be\uff0c\u8a9e\uf9be\u5167\u5bb9\u8a2d\u8a08\u6709\u82f1\u6587\uf969\u5b57\uf99a\u7e8c\u8a9e\u97f3\u3001\u82f1\u6587\u5b57\u6bcd\uf99a\u7e8c\u8a9e\u97f3\u3001\u4e2d\u82f1\u6587\u6df7\u5408 \uf906\u3001\u82f1\u6587\u55ae\u5b57\u3001\u7247\u8a9e\u6216\uf906\u5b50\u7b49\uff0c\u5982(\u8868 5)\u6240\u793a\u3002\uf941\u6587\u4e3b\u8981\u63a2\u8a0e\u4e2d\u82f1\u593e\u96dc\u7684\u591a\u8a9e\u61c9\u7528\uff0c\u5be6\u9a57\u62bd\u53d6\u8a9e\uf9be\u5167\u4e2d\u82f1\u6587 \u6df7\u5408\uf906\u578b (\u8868 5 \u4e4b 6 \u548c 7)\uff0c\u8a9e\uf9be\u7de8\u865f#58 \u81f3#70 \u7684\u97f3\u6a94\u8cc7\u8a0a\u3002 4.4. \u8072\u5b78\u8207\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u5206\u6790\u65bc\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc \u6bcf\u4e00\u500b\u5206\u652f(arc)\u9ede\uff1b \u7dda\u689d\u8868\u793a\u591a\u8a9e\u97f3\u7d20\u4e5f\u5c31\u662f\u8a13\uf996\u7684\u8072\u5b78\u6a21\u578b\uff0c \u6307\u97f3\u7d20\u7684\u7bc0\u9ede\uff1b \u8868\u793a\uf96e\u7d50\u9ede\uff0c\u6307\u51fa\u5f9e\u6839\u7bc0\u9ede\u5230\u6b64 \u672c\uf941\u6587\u7814\u7a76\u63a2\u8a0e\u4e2d\u6587\u548c\u82f1\u6587\u7684\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u61c9\u7528\uff0c\u5be6\u9a57\u9996\u5148\u6e2c\u8a66\u4f7f\u7528\u55ae\u97f3\u7d20\u6a21\u578b(monophone)\u7684\u5b9a</td></tr><tr><td colspan=\"4\">\uf96e\u7d50\u9ede\u4e4b\u767c\u8072\u97f3\u7d20\u53ef\u80fd\u69cb\u6210\u7684\u6240\u6709\u591a\u8a9e\u8a5e\u5f59\uff1b \u7fa9\uff0c\u4f9d\u64da(\u8868 1)\u548c(\u8868 2)\u7b49\uf967\u540c\u6a19\u8a18\u65b9\u6cd5\u7684\u5167\u5bb9\uff0c\u5206\u5225\u53ef\u4ee5\u5b9a\u7fa9\uff1a (\u4e00) \u76f4\u63a5\u7d50\u5408\u500b\u5225\u55ae\u4e00\u8a9e\u8a00\u4e4b\u97f3\u7d20(MIX)\uff1b \u8868\u793a\u6a39\u8207\u6a39\u4e4b\u9593\uf99a\u7d50\u7684\u8a9e\u8a00\u6a21\u578b\u3002 \u8868 5. EAT \u8a9e\uf9be\u4e2d\u591a\u8a9e\uf906\u578b\u7bc4\uf9b5 4.3. \uf9dd\u7528\u8072\u5b78\u8207\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u5206\u6790\u7fa4\u805a\u4e09\uf99a\u97f3\u7d20\u6a21\u578b (\u4e8c) \u4ee5 IPA \u70ba\u57fa\u6e96\u5b9a\u7fa9\u591a\u8a9e\u97f3\u7d20\u4e4b\u65b9\u6cd5 (IPA)\u3002\u5be6\u9a57\u7d50\u679c\u5982(\u8868 7)\u6240\u793a\uff1a</td></tr><tr><td colspan=\"4\">EAT \u8a9e\uf9be\uf906\u578b 100% four eight three zero one two nine for instance Safe len ins del sub len \u2212 \u2212 \u00d7 \u2212 \u8a9e\u97f3\u8fa8\uf9fc\u53ef\u80fd\u767c\u751f\u7684\u932f\u8aa4\u6709\u4e09\u7a2e\u578b\u614b\uff0c\u5206\u5225\u662f\u63d2\u5165\u932f\u8aa4(insertion)\u3001\u522a\u9664\u932f\u8aa4(deletion)\u4ee5\u53ca\u66ff\u63db\u932f\u8aa4 1 2 3 Accuracy = (\u5f0f 9) (substitution)\u3002\u5be6\u9a57\u4e2d\u97f3\u7d20\u6b63\u78ba\uf961(accuracy)\u7684\u8a08\u7b97[21]\uff0c\u65b9\u5f0f\u5982\u4e0b\uff1a \u8868 7. \u55ae\u97f3\u7d20\u6a21\u578b\u4e4b\u591a\u8a9e\u8fa8\uf9fc\u97f3\u7d20\u6b63\u78ba\uf961 (\u62ec\u5f27\u5167\u8868\u8fa8\uf9fc\u55ae\u5143\u4e4b\u500b\uf969)</td></tr><tr><td colspan=\"3\">4 \u5176\u4e2d\uff0c le \u70ba\u8fa8\uf9fc\u7d50\u679c\uff0c\u97f3\u7d20\u5e8f\uf99c\u7684\u9577\ufa01\u3002 in \u70ba\u6bd4\u8f03\u8f03\u6b63\u78ba\u7d50\u679c\u591a\u8fa8\uf9fc\u51fa\u7684\u97f3\u7d20\uff0c\u5c6c\u65bc\u63d2\u5165\u932f\u8aa4\uff0c Silicon Graphics n s</td><td>del</td><td>\u70ba</td></tr><tr><td colspan=\"4\">5 \u6bd4\u8f03\u6b63\u78ba\u7d50\u679c\u5c11\u8fa8\uf9fc\u5230\u7684\u97f3\u7d20\uff0c\u5c6c\u65bc\u522a\u9664\u932f\u8aa4\u3002 R. S. R. T. E. K. 6 \u6790\uf967\u540c\u7fa4\u96c6\u689d\u4ef6\u4e0b\u7684\u7fa4\u805a\u97f3\u7d20\u500b\uf969\uff0c\uf9dd\u7528\u8abf\u6574 k \u7fa4\u805a(modified k-means, MKM)\u5206\uf9d0\u65b9\u6cd5[16]\uff0c\u7fa4\u805a\u4e09\uf99a\u97f3 sub \u70ba\u6bd4\u8f03\u6b63\u78ba\u7d50\u679c\u8fa8\uf9fc\u932f\u8aa4\u7684\u97f3\u7d20\uff0c\u5c6c\u65bc\u66ff\u63db\u932f\u8aa4\u3002\u5206 \u51a0\u8ecd\u5bb6\u5ead T.V.\u79c0\u5165\u570d\uf90a\u9418\u734e 7 \u7d20\u6a21\u578b\u70ba\u6709\u6548\u591a\u8a9e\u8fa8\uf9fc\u6a21\u578b\u3002\u5be6\u9a57\u8072\u5b78\u76f8\u4f3c\ufa01\u8a08\u7b97(ACL)\u3001\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u8a08\u7b97(HAL)\u53ca\u8cc7\uf9be\u878d\u5408\u6280\u8853 \u5e6b\u6211\u67e5\u4e00\u4e0b Bryan \u7684\u5206\u6a5f (FUN)\u7b49\uf967\u540c\u65b9\u6cd5\uff0c\u5728\u6536\u6582\u9580\u6abb\u503c\u70ba 0.01 \u03b8 = \u7684\u60c5\u6cc1\u4e0b\uff0c\u5be6\u9a57\uf967\u540c\u6700\u5927\u7fa4\u96c6\uf969 Y \u3002(\u8868 6)\u5be6\u9a57\u5206\u6790\u5404\u7a2e\uf967\u540c</td></tr><tr><td colspan=\"3\">8 \u65b9\u6cd5\u7fa4\u96c6\u4e4b\u591a\u8a9e\u97f3\u7d20\u500b\uf969\u53ca\u97f3\u7d20\u8fa8\uf9fc\u7684\u6b63\u78ba\uf961\uff0c\u5982\u4e0b\u6240\u793a\uff1a The vote at the September meeting was eleven zero</td></tr><tr><td colspan=\"4\">\u539f\u672c\u97f3\u6a94\u5167\u5bb9\u7686\u5c6c\u65bc raw \u683c\u5f0f\uff0c\u56e0\u6b64\u6211\u5011\u4e8b\u5148\u5c0d\u97f3\u6a94\u4f5c dc-offset \u53ca silence removal \u7684\u8655\uf9e4\u3002\u4e26\u4e14\u6839\u64da\u82f1\u6587 \u8868 6. \uf967\u540c\u7fa4\u96c6\uf969\u76ee\u9650\u5236\u689d\u4ef6\u4e0b\u7fa4\u805a\u97f3\u7d20\u500b\uf969\u53ca\u8fa8\uf9fc\u6b63\u78ba\uf961 (Y :\u6700\u5927\u7fa4\u96c6\uf969\u76ee, 0.01 \u03b8 = )</td></tr><tr><td colspan=\"4\">) \u767c\u97f3\u8fad\u5178\u8207\u4e2d\u6587\u767c\u97f3\u8fad\u5178\uff0c\u5c07\u6587\u5b57\u8a3b\u89e3\u8f49\u6210\u97f3\u7d20\u6a19\u8a18\u3002\u7531\u65bc\u8a9e\uf9be\u5167\u6709\u90e8\u4efd\u97f3\u6a94\u53ca\uf93f\u97f3\u54c1\u8cea\uf967\uf97c\uff0c\u5be6\u9a57\u4ee5\u4eba \u5de5\u7684\u65b9\u5f0f\u5148\ufa08\u6821\u5c0d\u3002\u6700\u5f8c\uff0c\uf941\u6587\u6240\u63a1\u7528\u4e4b\u5be6\u9a57\u8a9e\uf9be\u5305\u542b\u6709\u8a13\uf996\u7528\u4e2d\u82f1\u6587\u6df7\u5408\uf906\u578b\u5171\u6709 2,018 \uf906\uff0c\u5be6\u9a57\u8a55\u4f30 8 Y = 16 Y = 32 Y =</td></tr><tr><td colspan=\"4\">\u5176\u4e2d\uff0c\u5411\uf97e l s \u8868\u793a\u76ee\u524d\u76f8\u4f3c\ufa01\u77e9\u9663\u5728\ufa08\uf96a\u5f15 l \u7684\u97f3\u7d20\uff0c\u5411\uf97e k s \u8868\u793a\u76f8\u4f3c\ufa01\u77e9\u9663\u5728\uf99c\uf96a\u5f15 \u7684\u97f3\u7d20\uff0c\u5168\u90e8\u97f3 \u7d20\u7e3d\u5171\u6709 n \u540d\u3002\u672c\u7814\u7a76\uf9dd\u7528\u8abf\u6574\u6027 k \u7fa4\u805a(modified k-means, MKM)\u5206\uf9d0\u65b9\u6cd5[16]\uff0c\u5b9a\u7fa9\u6536\u6582\u689d\u4ef6\u70ba\u5206\u7fa4\u5167 \u7684\u8cc7\uf9be\u8b8a\uf962\ufa01\u4f4e\u65bc\u5b9a\u7fa9\u4e4b\u9580\u6abb\u503c\uff0c\u5247\u9054\u6210\u5206\u7fa4\u7d42\u6b62\uff0c\u6700\u5f8c\u5b8c\u6210\uf941\u6587\u6240\u63d0\u4e4b\u6709\u6548\u591a\u8a9e\u97f3\u7d20\u96c6\uff0c\u5176\u4e2d\u6536\u6582\u689d\u4ef6 \u70ba\uff1a \u6e2c\u8a66\u5171\u6709 100 \uf906\u3002 \u6b63\u78ba\uf961 \u97f3\u7d20\u500b\uf969 \u6b63\u78ba\uf961 \u97f3\u7d20\u500b\uf969 \u6b63\u78ba\uf961 \u97f3\u7d20\u500b\uf969 k 1 1 1 1 1 ( ) / Y Y 4.2. \u97f3\u7d20\u70ba\u57fa\u6e96\u4e4b\u81ea\u52d5\u8a9e\u97f3\u8fa8\uf9fc\u67b6\u69cb ACL 62.22% 161 63.12% 288 64.37% 531 \u70ba\uf9ba\u8a55\u4f30\u97f3\u7d20\u5b9a\u7fa9\u7684\u597d\u58de\uff0c\u672c\uf941\u6587\u4f7f\u7528\u81ea\ufa08\u958b\u767c\u7684\u591a\u8a9e\u97f3\u7d20\u8fa8\uf9fc\u7cfb\u7d71\uff0c\u63a2\u8a0e\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u3002\u63a1\u7528\u4e0a\u8ff0 HAL 62.52% 159 64.23% 286 64.57% 530 \u4e4b\u5be6\u9a57\u8a9e\uf9be\u4e2d\uff0c\u6211\u5011\uf9dd\u7528 IPA \u97f3\u7d20\u6a19\u6e96\u5b9a\u7fa9\uff0c\u627e\u51fa\u591a\u8a9e\u97f3\u7d20\u4e4b\uf997\u96c6\u3002\u5b9a\u7fa9\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\u5171 N=997 \u500b\uff0c\u8a13\uf996 FUN 64.44% 119 66.07% 260 64.74% 515 \u8a9e\uf9be\u5c11\u65bc 5 \u6b21\u7684\u4e09\uf99a\u97f3\u7d20\uf967\u4e88\u8003\u616e\u3002\u5728\u8a9e\u97f3\uf96b\uf969\u64f7\u53d6\u7684\u90e8\u4efd\uff0c\u5c0d\u65bc\u8f38\u5165\u7684\u8a9e\u97f3\u8a0a\u865f\u8a08\u7b97 26 \u7dad\u7684\u6885\u723e\u5012\u983b Y t t t y y y y y y \u03b8 \u2212 \u2212 = = = \u0394 \u2212 \u0394 \u0394 &lt; \u2211 \u2211 \u2211 (\u5f0f 8) \u5176\u4e2d\uff0c \u8868\u793a\u5728\u7b2c \u6b21\u905e\u8ff4\u4e2d\uff0c \u7fa4\u96c6\u4e2d\u7b2c \u7fa4\u4e4b\u96c6\u5408\u5167\u500b\uf969\u5206\uf969\u503c t y \u0394 t Y y ( , ) t y l k c s s \u0394 = \u2211 \uff0c \u8868\u793a\u904b \u7b97\u905e\u8ff4\u6b21\uf969\uff0c \u6307\u8a2d\u5b9a\u4e4b\u6700\u5927\u905e\u8ff4\u6b21\uf969\uff0c max 1,..., t t \u8b5c\uf96b\uf969(mel-frequency ceptral coefficient, MFCC)\uff0c\u5176\u4e2d\u5305\u542b 12 \u968e\u7684\u6885\u723e\u5012\u983b\u8b5c\uf96b\uf969\uff0c\u52a0\u4e0a 12 \u968e\u7684\u4e00\u6b21\u5fae\u5206 \u5be6\u9a57\u8003\u616e\u8072\u5b78\u76f8\u4f3c\ufa01\u77e9\u9663\u5206\uf969\u8a08\u7b97\uff0c\uf9dd\u7528\u8072\u5b78\u76f8\u4f3c\ufa01\u7fa4\u805a\u65b9\u6cd5 ACL\uff0c\u5728 8, 16, 32 Y = \u7684\u60c5\u6cc1\u4e0b\uff0c\u5206\u5225\u53ef\u4ee5 \u6885\u723e\u5012\u983b\u8b5c\uf96b\uf969\uff0c\u4ee5\u53ca\u4e00\u968e\u7684\u80fd\uf97e\u548c\u5176\u4e00\u6b21\u5fae\u5206\uf96b\uf969\uff0c\u4e26\u4e14\u5c0d\uf96b\uf969\u505a MVA [18]\u8655\uf9e4\u4ee5\u589e\u52a0\u8fa8\uf9fc\u7684\u5f37\u5065\u6027\u3002 \u7fa4\u805a\u70ba 161\uff0c288 \u53ca 531 \u500b\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\uff0c\u5176\u97f3\u7d20\u8fa8\uf9fc\u6b63\u78ba\uf961\u5206\u5225\u70ba 62.22%\uff0c63.12%\u53ca 64.37%\u3002\uf9dd\u7528\u8a9e\u8a00 = max t \u8d85\u7a7a\u9593\u5206\u6790\u65b9\u6cd5 HAL\uff0c\u5728 \u7684\u60c5\u6cc1\u4e0b\uff0c\u5206\u5225\u53ef\u4ee5\u7fa4\u805a\u70ba 159\uff0c286 \u53ca 530 \u500b\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\uff0c\u5176 8, 16, 32 Y = \u03b8 \u70ba\u6536\u6582\u4e4b\u9580\u6abb\u503c\u3002 \u97f3\u7d20\u8fa8\uf9fc\u6b63\u78ba\uf961\u5206\u5225\u70ba 62.52%\uff0c64.23%\u53ca 64.57%\u3002\uf9dd\u7528\u8cc7\uf9be\u878d\u5408\u65b9\u6cd5 FUN\uff0c\u5728 \u7684\u60c5\u6cc1\u4e0b\uff0c 8, 16, 32 Y =</td></tr><tr><td colspan=\"4\">\u5206\u5225\u53ef\u4ee5\u7fa4\u805a\u70ba 159\uff0c286 \u53ca 530 \u500b\u591a\u8a9e\u97f3\u7d20\u6a21\u578b\uff0c\u5176\u97f3\u7d20\u8fa8\uf9fc\u6b63\u78ba\uf961\u5206\u5225\u70ba 64.44%\uff0c66.07%\u53ca 64.74%\u3002 4. \u5be6\u9a57\u8a55\u4f30 \u70ba\uf9ba\u8a55\u4f30\u7814\u7a76\u65b9\u6cd5\uff0c\uf941\u6587\u63d0\u51fa\u5e7e\u9805\u5be6\u9a57\u9a57\u8b49\uff1a\u9996\u5148\uff0c\u5be6\u9a57\u55ae\u7368\u8003\u616e\u8072\u5b78\u76f8\u4f3c\ufa01\u3001\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u8207 16</td></tr><tr><td colspan=\"4\">\u672c\uf941\u6587\u6240\u63d0\u7d50\u5408\u8072\u5b78\u8207\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u5206\u6790\u4e4b\u65b9\u6cd5\uff0c\u6bd4\u8f03\u5176\u8fa8\uf9fc\u7d50\u679c\u3002\u518d\u8005\uff0c\u6bd4\u8f03\u8207\u524d\u5f8c\u6587\u8108\u7368\uf9f7\u4e4b\u97f3 14</td></tr><tr><td colspan=\"4\">\u7d20\u96c6\u548c\uf941\u6587\u6240\u63d0\u8207\u524d\u5f8c\u6587\u8108\u76f8\u95dc\u4e4b\u97f3\u7d20\u96c6\u5728\u591a\u8a9e\u8fa8\uf9fc\u6e96\u78ba\uf961\u7684\u5dee\u5225\u3002 4.1. \u591a\u8a9e\u8a9e\u97f3\u8a9e\uf9be\u5206\u6790 \u672c\uf941\u6587\u4f7f\u7528\u7684\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u8a13\uf996\u8a9e\uf9be\uff0c\u53f0\u7063\u8154\u82f1\u6587(English Across Taiwan, EAT)\u8a9e\uf9be\u5eab\uff0c\u5176\u4e2d\u5305\u542b\u82f1 \u6587\u9577\uf906,\u82f1\u6587\u77ed\uf906,\u82f1\u6587\u55ae\u8a5e\u53ca\u4e2d\u82f1\u593e\u96dc\uf906\u7b49[17]\u3002\u5f9e 2004 \uf98e 5 \u6708\u958b\u59cb\u6536\u96c6\uff0c\u81f3 2005 \uf98e 1 \u6708\u521d\u6b65\u5b8c\u6210\u6536\u96c6\uff0c \u7531\u5e2b\u5927\u3001\u4ea4\u5927\u3001\u6e05\u5927\u3001\u6210\u5927\u548c\u53f0\u5927\u7b49\u4e94\u6240\u5b78\u6821\uf96b\u8207\u8a9e\uf9be\u4e4b\uf93f\u88fd\u6536\u96c6\uff0c\u7d93\u5de5\u7814\u9662\u96fb\u901a\u6240\u5f59\u6574\u3002\u5206\u5225\u7531\u82f1\u8a9e\u7cfb \u53ca\u975e\u82f1\u8a9e\u7cfb\u5b78\u751f\uf93f\u88fd\uff0c\u8a9e\uf9be\u4f9d\u6027\u5225\u505a\u5206\uf9d0\uff0c\uf93f\u88fd\u6709\u9ea5\u514b\u98a8\u8a9e\uf9be\u53ca\u96fb\u8a71\u8a9e\uf9be\uff0c\u6b78\u7d0d\u5982\u4e0b\u8868\u6240\uf99c\uff1a \u8868 4. EAT \u8a9e\uf9be\u9ea5\u514b\u98a8\u97f3\u6a94\u8cc7\uf9be\u7d71\u8a08 MIC 16khz 16bits \u8a9e\uf9be 4 6 8 10 Number of clustered tri-phones 12</td></tr><tr><td>2</td><td>\u82f1\u8a9e\u7cfb</td><td>\u975e\u82f1\u8a9e\u7cfb</td></tr><tr><td colspan=\"4\">\u7537\u6027 11,977 \u5716 4. \uf9dd\u7528\u6a39\uf9fa\u7d50\u69cb\u767c\u97f3\u8fad\u5178\u6587\u6cd5\u6a39\u65bc\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u67b6\u69cb\u5716 \uf981\u6027 \u7537\u6027 \uf981\u6027 \uf906\uf969 30,094 25,432 0 10 20 30 40 50 15,540 \u4eba\uf969 166 406 368 \u672c\uf941\u6587\u8abf\u6574\u4e00\u822c\u8a9e\u97f3\u8fa8\uf9fc\u4f7f\u7528\u7684\u8a9e\u8a00\u6a21\u578b(language model)\uff0c\u5728\u8a08\u7b97\u4e0a\uf9dd\u7528\u5747\u7b49\u6a5f\uf961(equal probability) IPA phone model 224 \u7684\u65b9\u6cd5[19]\uff0c\u78ba\u4fdd\u53ef\u4ee5\u771f\u6b63\u5448\u73fe\uf967\u540c\u591a\u8a9e\u97f3\u7d20\u5b9a\u7fa9\u7684\u8072\u5b78\u6a21\u578b(acoustic model)\uff0c\u5c0d\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u7684\u5f71\u97ff\u3002\u5728 \u591a\u8a9e\u8a9e\u97f3\u97f3\u7d20\u8fa8\uf9fc\uff0c\u9700\u8981\u4f9d\u64da\u5b9a\u7fa9\u7684\u591a\u8a9e\u97f3\u7d20\u7d50\u5408\u5404\u500b\u76ee\u6a19\u8a9e\u8a00\u7684\u767c\u97f3\u8fad\u5178\uff0c\u5efa\u69cb\u51fa\u4e00\u500b\u591a\u8a9e\u767c\u97f3\u8fad\u5178\u3002 \u5716 5. \uf9dd\u7528\u8072\u5b78\u76f8\u4f3c\ufa01\u53ca\u524d\u5f8c\u6587\u8108\u5206\u6790\u7fa4\u96c6\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\u5206\u4f48\u5716\uff0c 16 Y =</td></tr></table>",
                "num": null
            },
            "TABREF5": {
                "type_str": "table",
                "html": null,
                "text": "Triphone sets)\u7684\u591a\u8a9e\u8fa8\uf9fc\u6548\u679c\u53ef\u9054 68.07%\u7684\u6b63\u78ba\uf961\u3002\uf9dd\u7528\u8072\u5b78\u76f8 \u4f3c\ufa01\u77e9\u9663\u7fa4\u96c6\u97f3\u7d20\u5b9a\u7fa9(ACL phone sets) \uff0c\u5728\u591a\u8a9e\u8fa8\uf9fc\u6548\u679c\u4e0a\u53ef\u9054 63.12%\u7684\u6b63\u78ba\uf961\uff0c\u800c\uf9dd\u7528\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8 \u4f3c\ufa01\u77e9\u9663\u7fa4\u96c6\u97f3\u7d20\u5b9a\u7fa9(HAL phone sets) \uff0c\u5728\u4e2d\u82f1\u6587\u591a\u8a9e\u8fa8\uf9fc\u6548\u679c\u4e0a\u53ef\u9054 64.23%\u7684\u6b63\u78ba\uf961\u3002\u9032\u4e00\u6b65\uf9dd\u7528\u8cc7 \uf9be\u878d\u5408\u65b9\u6cd5\u65bc\u8072\u5b78\u53ca\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u77e9\u9663\u7fa4\u96c6\u5206\u6790\u4e4b\u97f3\u7d20\u5b9a\u7fa9(FUN phone sets) \uff0c\u5728\u591a\u8a9e\u8fa8\uf9fc\u6548\u679c\u53ef\u4ee5 \u63d0\u5347\u81f3 66.07%\u7684\u6b63\u78ba\uf961\u3002\u6574\u9ad4\u800c\u8a00\uff0c\u63a1\u7528\u4e09\uf99a\u97f3\u7d20\u7684\u8fa8\uf9fc\u6548\u679c\u6bd4\u55ae\u97f3\u7d20(IPA \u6216 MIX)\u5b9a\u7fa9\u597d\u3002\u53c8\u5f9e\u8a9e\u8a00\u5206 \u6790(HAL)\u6548\u679c\u6703\u8f03\u8072\u5b78\u5206\u6790(ACL)\u6548\u679c\uf92d\u5f97\u986f\u8457\uff0c\u4e14\uf9dd\u7528\u8cc7\uf9be\u878d\u5408\u65b9\u6cd5\u7d50\u5408\u8072\u5b78\u76f8\u4f3c\ufa01\u53ca\u524d\u5f8c\u6587\u8108\u5206\u6790\uff0c\u5c0d \u65bc\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u53ef\u4ee5\u6709\u660e\u986f\u7684\u63d0\u5347\u3002",
                "content": "<table><tr><td/><td/><td>INSERTION</td><td colspan=\"2\">DELETION SUBSTITUTION</td></tr><tr><td>Triphone sets (997)</td><td>68.07%</td><td>15.87%</td><td>4.43%</td><td>11.63%</td></tr><tr><td>ACL phone sets (288)</td><td>63.12%</td><td>19.73%</td><td>4.88%</td><td>12.32%</td></tr><tr><td>HAL phone sets (286)</td><td>64.23%</td><td>20.67%</td><td>4.75%</td><td>10.48%</td></tr><tr><td>FUN phone sets (260)</td><td>66.07%</td><td>16.94%</td><td>4.41%</td><td>12.71%</td></tr><tr><td colspan=\"5\">====================== English Across Taiwan, EAT ======================</td></tr><tr><td>\u7531\u5be6\u9a57\u7d50\u679c\u53ef\u77e5\uff0c\u5408\u4f75\u524d\u7684\u4e09\uf99a\u97f3\u7d20\u6a21\u578b(5. \u7d50\uf941\u53ca\u672a\uf92d\u5c55\u671b</td><td/><td/><td/><td/></tr><tr><td colspan=\"5\">\u672c\uf941\u6587\u63d0\u51fa\u61c9\u7528\u8072\u5b78\u76f8\u4f3c\ufa01\u53ca\u524d\u5f8c\u6587\u8108\u5206\u6790\u65bc\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u4e4b\u6709\u6548\u97f3\u7d20\u5b9a\u7fa9\uff0c\u4ee5 EAT \u4e2d\u82f1\u6587\u96d9\u8a9e\u8a9e</td></tr><tr><td colspan=\"5\">\uf9be\u70ba\uf9b5\u3002\u57fa\u65bc IPA \u6a19\u6e96\u5b9a\u7fa9\u4e4b\u591a\u8a9e\u55ae\u97f3\u7d20\u96c6\uff0c\u672c\u7814\u7a76\u8003\u616e\u4ee5\u767c\u97f3\u524d\u5f8c\u6587\u76f8\u4f9d\u4e09\uf99a\u97f3\u7d20\u6a21\u578b\u3002\u4ee5\u6b64\u5b9a\u7fa9\uff0c\u6211</td></tr><tr><td colspan=\"5\">\u5011\u5206\u5225\u4ee5\u8072\u5b78\u76f8\u4f3c\ufa01\u53ca\u524d\u5f8c\u6587\u8108\u5206\u6790\uff0c\u97f3\u7d20\u9593\u76f8\u4f3c\ufa01\u9ad8\u7684\u97f3\u7d20\u5408\u4f75\uff0c\u671f\u671b\u627e\u51fa\ufa1d\u7c21\u6709\u6548\u7684\u591a\u8a9e\u8a9e\u97f3\u8fa8\uf9fc\u97f3</td></tr><tr><td colspan=\"5\">\u7d20\u96c6\u3002\uf9dd\u7528\u97f3\u7d20 HMM \u6a21\u578b\uff0c\u4ee5\u76f4\u63a5\u6821\u6e96\u65b9\u6cd5\ufa00\u97f3\u4e26\u8a08\u7b97\u4e8b\u5f8c\u6a5f\uf961\u503c\uff0c\u5efa\uf9f7\u8072\u5b78\u76f8\u4f3c\ufa01\u77e9\u9663\u3002\uf9dd\u7528\u8a9e\u8a00\u8d85</td></tr><tr><td colspan=\"5\">\u7a7a\u9593\u76f8\u4f3c\ufa01\u5206\u6790(hyperspace analog to language, HAL)\uff0c\u627e\u51fa\u97f3\u7d20\u524d\u5f8c\u767c\u97f3\u7279\u6027\u6240\u9020\u6210\u7684\u8b8a\u97f3\u5f71\u97ff\uff0c\u5efa\uf9f7\u8a9e</td></tr><tr><td colspan=\"5\">\u8a00\u767c\u97f3\u4e0a\u76f8\u4f3c\ufa01\u77e9\u9663\u3002\u4e4b\u5f8c\uff0c\u4ee5\u8cc7\uf9be\u878d\u5408\u65b9\u6cd5\uff0c\u540c\u6642\u8003\u616e\u8072\u5b78\u548c\u8a9e\u8a00\u8d85\u7a7a\u9593\u76f8\u4f3c\ufa01\u77e9\u9663\u3002\uf9dd\u7528\u5411\uf97e\uf97e\u5316\u7fa4</td></tr><tr><td colspan=\"5\">\u96c6\u5206\u6790\uff0c\u627e\u51fa\u540c\u4e00\uf9d0\u5225\u4e4b\u97f3\u7d20\u5b9a\u7fa9\uff0c\u5efa\uf9f7\u6709\u6548\u800c\ufa1d\u7c21\u7684\u591a\u8a9e\u97f3\u7d20\u96c6\u3002\u5be6\u9a57\u8b49\u660e\uf9dd\u7528\u7d50\u5408\u8072\u5b78\u548c\u8a9e\u8a00\u8d85\u7a7a\u9593</td></tr><tr><td colspan=\"5\">\u76f8\u4f3c\ufa01\u77e9\u9663\u5206\u6790\u65b9\u6cd5\uff0c\u53ef\u4ee5\u9054\u5230\uf97c\u597d\u7684\u591a\u8a9e\uf99a\u7e8c\u8a9e\u97f3\u8fa8\uf9fc\u7684\u6548\u679c\u3002\u672a\uf92d\u53ef\u4ee5\u5c07\u65b9\u6cd5\u61c9\u7528\u5728\u55ae\u4e00\u8a9e\u8a00\u8a9e\u97f3\u8fa8</td></tr></table>",
                "num": null
            }
        }
    }
}