{ "paper_id": "O15-1026", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T08:10:03.588508Z" }, "title": "A Python Implementation of Automatic Speech-text Synchronization Using Speech Recognition and Text-to-Speech Technology", "authors": [ { "first": "Chun-Han", "middle": [], "last": "\u8cf4\u4fca\u7ff0", "suffix": "", "affiliation": { "laboratory": "", "institution": "Chang Gung University", "location": {} }, "email": "" }, { "first": "", "middle": [], "last": "Lai", "suffix": "", "affiliation": { "laboratory": "", "institution": "Chang Gung University", "location": {} }, "email": "" }, { "first": "Chao-Kai", "middle": [], "last": "\u5f35\u671d\u51f1", "suffix": "", "affiliation": { "laboratory": "", "institution": "Chang Gung University", "location": {} }, "email": "" }, { "first": "", "middle": [], "last": "Chang", "suffix": "", "affiliation": { "laboratory": "", "institution": "Chang Gung University", "location": {} }, "email": "" }, { "first": "Renyuan", "middle": [], "last": "\u5442\u4ec1\u5712", "suffix": "", "affiliation": { "laboratory": "", "institution": "Chang Gung University", "location": {} }, "email": "renyuan.lyu@gmail.com" }, { "first": "", "middle": [], "last": "Lyu", "suffix": "", "affiliation": { "laboratory": "", "institution": "Chang Gung University", "location": {} }, "email": "" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "\u672c\u7814\u7a76\u8a2d\u8a08\u4e00\u500b\u65b9\u4fbf\u8655\u7406\u6709\u8072\u66f8\u97f3\u6587\u540c\u6b65\u7684\u6280\u8853\uff0c\u5229\u7528\u96f2\u7aef\u7684\u6587\u5b57\u8f49\u8a9e\u97f3(Text-to-speech)\u6280\u8853\uff0c \u7d50\u5408\u8a9e\u97f3\u8fa8\u8b58(Speech Recognition)\u6280\u8853\uff0c\u8b93\u4f7f\u7528\u8005\u80fd\u5920\u4f7f\u7528\u81ea\u884c\u6e96\u5099\u7684\u6587\u7ae0\u4f86\u88fd\u4f5c\u81ea\u5df1\u7684\u300e\u8ddf \u8ff0\u7df4\u7fd2\u300f(Shadowing technique)\u7684\u5b78\u7fd2\u7d20\u6750\uff0c\u88fd\u4f5c\u9054\u5230\u8a5e\u5c64\u7d1a(Word-level)\u7684\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u3002 \u6b64\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u662f\u85c9\u7531\u300e\u5e36\u6642\u9593\u9ede\u7684\u6587\u5b57\u300f (Timed-text)\u6a94\u6848\u6240\u88fd\u4f5c\uff0c\u800c\u5e36\u6642\u9593\u9ede\u7684\u6587\u5b57\u5247 \u662f\u7531\u4f7f\u7528\u8005\u6240\u63d0\u4f9b\u7684\u6587\u7ae0\u9023\u540c\u5c0d\u61c9\u7684\u8a9e\u97f3\u8072\u6ce2\u6a94\u6848\uff0c\u7d93\u7531\u4e00\u5957\u540d\u70ba CGUAlign \u7684\u97f3\u6587\u540c\u6b65\u6280 \u8853\u4e4b\u8655\u7406\u6240\u7522\u751f\u7684\u3002 CGUAlign \u662f\u904b\u7528 Python \u5c07\u4e00\u6709\u540d\u7684\u8a9e\u97f3\u8fa8\u8b58\u6280\u8853\u2500HTK(Hidden Markov Model Toolkit) \u5305\u88dd\uff0c\u53ea\u8981\u63d0\u4f9b\u6587\u5b57\u6a94\u53ca\u5176\u6717\u8b80\u7684\u8a9e\u97f3\u6a94\uff0c\u5176\u4e2d\u8a9e\u97f3\u6a94\u662f\u7d93\u7531\u96f2\u7aef\u8a9e\u97f3\u5408\u6210\u6280 \u8853\u800c\u5f97\u4f86\u7684\uff0c\u5373\u80fd\u88fd\u4f5c\u51fa\u97f3\u6587\u540c\u6b65\u7684\u5e36\u6642\u9593\u9ede\u7684\u6587\u5b57\u6a94\u6848\uff0c\u96a8\u5f8c\uff0c\u6211\u5011\u4e5f\u5efa\u7acb\u4e00\u500b\u7c21\u6613\u7684\u4ee5 JavaScript \u88fd\u4f5c\u7684\u7db2\u7ad9\uff0c\u80fd\u5920\u904b\u7528\u9019\u500b\u6a94\u6848\u505a\u96fb\u8166\u8f14\u52a9\u8a9e\u8a00\u5b78\u7fd2(Computer-assisted language learning, CALL)\u4e4b\u7528\uff0c\u6b64\u7db2\u7ad9\u80fd\u5920\u95b1\u8b80\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\uff0c\u8b93\u4f7f\u7528\u8005\u80fd\u5920\u8f03\u8f15\u9b06\u7684\u505a\u8ddf\u8ff0\u7df4\u7fd2\uff0c \u6700\u5f8c\u6211\u5011\u4e5f\u63d0\u4f9b\u5373\u6642\u7ffb\u8b6f\u7684\u529f\u80fd\u4f86\u9054\u5230\u96fb\u8166\u8f14\u52a9\u8a9e\u8a00\u5b78\u7fd2\u7684\u76ee\u6a19\u3002", "pdf_parse": { "paper_id": "O15-1026", "_pdf_hash": "", "abstract": [ { "text": "\u672c\u7814\u7a76\u8a2d\u8a08\u4e00\u500b\u65b9\u4fbf\u8655\u7406\u6709\u8072\u66f8\u97f3\u6587\u540c\u6b65\u7684\u6280\u8853\uff0c\u5229\u7528\u96f2\u7aef\u7684\u6587\u5b57\u8f49\u8a9e\u97f3(Text-to-speech)\u6280\u8853\uff0c \u7d50\u5408\u8a9e\u97f3\u8fa8\u8b58(Speech Recognition)\u6280\u8853\uff0c\u8b93\u4f7f\u7528\u8005\u80fd\u5920\u4f7f\u7528\u81ea\u884c\u6e96\u5099\u7684\u6587\u7ae0\u4f86\u88fd\u4f5c\u81ea\u5df1\u7684\u300e\u8ddf \u8ff0\u7df4\u7fd2\u300f(Shadowing technique)\u7684\u5b78\u7fd2\u7d20\u6750\uff0c\u88fd\u4f5c\u9054\u5230\u8a5e\u5c64\u7d1a(Word-level)\u7684\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u3002 \u6b64\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u662f\u85c9\u7531\u300e\u5e36\u6642\u9593\u9ede\u7684\u6587\u5b57\u300f (Timed-text)\u6a94\u6848\u6240\u88fd\u4f5c\uff0c\u800c\u5e36\u6642\u9593\u9ede\u7684\u6587\u5b57\u5247 \u662f\u7531\u4f7f\u7528\u8005\u6240\u63d0\u4f9b\u7684\u6587\u7ae0\u9023\u540c\u5c0d\u61c9\u7684\u8a9e\u97f3\u8072\u6ce2\u6a94\u6848\uff0c\u7d93\u7531\u4e00\u5957\u540d\u70ba CGUAlign \u7684\u97f3\u6587\u540c\u6b65\u6280 \u8853\u4e4b\u8655\u7406\u6240\u7522\u751f\u7684\u3002 CGUAlign \u662f\u904b\u7528 Python \u5c07\u4e00\u6709\u540d\u7684\u8a9e\u97f3\u8fa8\u8b58\u6280\u8853\u2500HTK(Hidden Markov Model Toolkit) \u5305\u88dd\uff0c\u53ea\u8981\u63d0\u4f9b\u6587\u5b57\u6a94\u53ca\u5176\u6717\u8b80\u7684\u8a9e\u97f3\u6a94\uff0c\u5176\u4e2d\u8a9e\u97f3\u6a94\u662f\u7d93\u7531\u96f2\u7aef\u8a9e\u97f3\u5408\u6210\u6280 \u8853\u800c\u5f97\u4f86\u7684\uff0c\u5373\u80fd\u88fd\u4f5c\u51fa\u97f3\u6587\u540c\u6b65\u7684\u5e36\u6642\u9593\u9ede\u7684\u6587\u5b57\u6a94\u6848\uff0c\u96a8\u5f8c\uff0c\u6211\u5011\u4e5f\u5efa\u7acb\u4e00\u500b\u7c21\u6613\u7684\u4ee5 JavaScript \u88fd\u4f5c\u7684\u7db2\u7ad9\uff0c\u80fd\u5920\u904b\u7528\u9019\u500b\u6a94\u6848\u505a\u96fb\u8166\u8f14\u52a9\u8a9e\u8a00\u5b78\u7fd2(Computer-assisted language learning, CALL)\u4e4b\u7528\uff0c\u6b64\u7db2\u7ad9\u80fd\u5920\u95b1\u8b80\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\uff0c\u8b93\u4f7f\u7528\u8005\u80fd\u5920\u8f03\u8f15\u9b06\u7684\u505a\u8ddf\u8ff0\u7df4\u7fd2\uff0c \u6700\u5f8c\u6211\u5011\u4e5f\u63d0\u4f9b\u5373\u6642\u7ffb\u8b6f\u7684\u529f\u80fd\u4f86\u9054\u5230\u96fb\u8166\u8f14\u52a9\u8a9e\u8a00\u5b78\u7fd2\u7684\u76ee\u6a19\u3002", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "\"\u53e5\u865f\"(\".\",\"\u3002\") \u3001\"\u554f\u865f\"(\"?\",\"\uff1f\") \u3001\" \u9a5a\u5606\u865f\"(\"!\",\"\uff01\") \u3001 \"\u7834\u6298\u865f\"(\"-\",\"\u2500\") \u3001\"\u5192\u865f\"(\":\",\"\uff1a\") \u3001\"\u9017\u865f\"(\",\",\"\uff0c\")\u3002", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "Step2:\u82e5\u6700\u7d42\u5b57\u4e32\u9577\u5ea6\u9084\u662f\u6709\u8d85\u904e 100 \u7684\uff0c\u5247\u6703\u5f9e\u8d85\u904e 100 \u7684\u5b57\u4e32\u4ee5\u4e2d\u9593\u7684\"\u7a7a\u767d\" \u5207\u5272\u3002 our talk turns to our children and our families however different we ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [], "bib_entries": { "BIBREF1": { "ref_id": "b1", "title": "The HTK Book version 3, Microsoft Corporation", "authors": [ { "first": "Steve", "middle": [], "last": "Young", "suffix": "" } ], "year": 2000, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Steve Young, The HTK Book version 3, Microsoft Corporation, 2000. http://htk.eng.cam.ac.uk/", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "SailAlign: Robust long speech-text alignment", "authors": [ { "first": "A", "middle": [], "last": "Katsamanis", "suffix": "" }, { "first": "M", "middle": [ "P" ], "last": "Black", "suffix": "" }, { "first": "P", "middle": [ "G" ], "last": "Georgiou", "suffix": "" }, { "first": "L", "middle": [], "last": "Goldstein", "suffix": "" }, { "first": "S", "middle": [], "last": "Narayanan", "suffix": "" } ], "year": 2011, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "A. Katsamanis, M. P. Black, P. G. Georgiou, L. Goldstein, S. Narayanan\"SailAlign: Robust long speech-text alignment\" University of Southern California, Los Angeles, CA, USA , Jan. 28-31, 2011.", "links": null } }, "ref_entries": { "FIGREF0": { "uris": null, "type_str": "figure", "num": null, "text": "Thank you very much, Gertrude Mongella, for your dedicated work that has brought us to this point, distinguished delegates, and guests: 0:0:0.000000,0:0:1.619000Thank you very much, 0:0:1.619000,0:0:3.022000 Gertrude Mongella, 0:0:3.022000,0:0:6.549000 for your dedicated work that has brought us to this point," }, "FIGREF1": { "uris": null, "type_str": "figure", "num": null, "text": "" }, "FIGREF2": { "uris": null, "type_str": "figure", "num": null, "text": "FFmpeg\u3002 def ffmpeg_AudioDuration(filename): os.system(\"ffmpeg -report -y -i ./TTS-MP3/{0}.mp3\" + \"./FFmpeg-WAV/{1}.wav\".format(filename,filename)) dirlist= os.listdir() for i in dirlist : if i.find('ffmpeg')!=-1 and i.find('.log') !=-1 : report_name= i break f=open(report_name,\"r\") for i in f: if i.find(\"Duration:\") != -1: duration= i.split(\" Duration: \")[1].split(\",\")[0] hour= int(duration.split(\":\")[0]) min = int(duration.split(\":\")[1]) sec = float(duration.split(\":\")[2]) total_ms= int(hour* 3600000 + min*60000 + sec*1000) print(total_ms) f.close() os.system(\"copy \"+report_name+\" .\\\\FFmpeg-WAV\\\\\"+report_name) os.system(\"del \"+report_name) return total_ms \u5716\u516b\u3001Audio Converter \u7bc4\u4f8b\u7a0b\u5f0f\u78bc (\u4e8c) CGUAlign \u8a9e\u97f3\u8fa8\u8b58-Force Alignment \u672c\u7ae0\u7bc0\u5c07\u8aaa\u660e\u5982\u4f55\u5c07( \u4e00 ) \u96f2 \u7aef \u8a9e \u97f3 \u5408 \u6210 (Text-to-speech,TTS) \u5f97\u5230\u7684\u53e5\u5c64\u7d1a" }, "TABREF0": { "html": null, "num": null, "type_str": "table", "content": "
\u81ea\u6211\u5167\u5316\u5b78\u7fd2\u53ca\u589e\u9032\u8a9e\u8a00\u80fd\u529b\u3002 \u9664\u6b64\u4e4b\u5916\uff0cGoogle Translate \u4e5f\u63d0\u4f9b\u6717\u8b80\u7684\u529f\u80fd\uff0c\u5373\u6587\u5b57\u8f49\u8a9e\u97f3(Text-to-speech)\u7684\u4eba (\u56db) CGUAlign[3]
\u4e8c\u3001\u76f8\u95dc\u7814\u7a76 \u5de5\u8a9e\u97f3\u5408\u6210\u6717\u8b80\u7684\u529f\u80fd\uff0c\u53e6\u5916\u4e5f\u63d0\u4f9b\u67e5\u8a62\u6587\u5b57\u62fc\u97f3\u7684\u529f\u80fd\uff0c\u5373\u80fd\u5920\u63d0\u4f9b\u975e\u62fc\u97f3\u8a9e\u8a00 \u7684\u7f85\u99ac\u62fc\u97f3\u67e5\u8a62\u3002 \u4e09\u3001\u7814\u7a76\u65b9\u6cd5 CguAlign \u662f\u6a21\u4eff[4]\u4ee5 \u904b\u7528\u7684\u4e00\u5957\u6280\u8853\uff0c\u70ba\u672c\u5be6\u9a57\u5ba4\u4e00\u500b\u65b9\u4fbf\u8655\u7406\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u7684\u6280\u8853\uff0c\u539f\u672c\u662f\u70ba\u4e86\u5c07 (\u4e00) \u8ddf\u8ff0\u7df4\u7fd2(Shadowing Technique) Google Translate \u6240\u63d0\u4f9b\u7684\u4e09\u7a2e\u529f\u80fd\u2500\u7ffb\u8b6f\u3001\u6717\u8b80\u3001\u62fc\u97f3\u5df2\u7d93\u5f88\u9069\u5408\u505a\u521d\u6b65\u7684\u8a9e\u8a00 \u672c\u7ae0\u7bc0\u5167\u5bb9\u65e8\u5728\u4ecb\u7d39\u6574\u9ad4\u7684\u7814\u7a76\u65b9\u6cd5\uff0c\u5168\u7ae0\u5206\u70ba\u4e09\u7bc0\uff0c \u5f9e Youtube \u4e0a\u6240\u53d6\u5f97\u7684\u53e5\u5c64\u7d1a Timed-text sbv \u6a94\uff0c\u4ee5\u7a0b\u5f0f\u81ea\u52d5\u5207\u97f3\u53d6\u4ee3\u50b3\u7d71\u4eba\u5de5\u624b (\u4e00) \u96f2\u7aef\u8a9e\u97f3\u5408\u6210(Text-to-speech,TTS) Shadowing technique \u662f\u4e00\u7a2e\u8a9e\u8a00\u5b78\u7fd2\u6280\u5de7\uff0c\u4e00\u822c\u6211\u5011\u7a31\u4e4b\u70ba\u8ddf\u8ff0\u7df4\u7fd2\u6216\u8005\u662f\u5f71\u5b50\u7df4 \u7fd2\uff0c\u8207\u76ee\u524d\u53f0\u7063\u6240\u5e38\u898b\u7684\u8b1b\u8ff0\u5f0f\u6559\u5b78\u6cd5\u4e0d\u540c\uff0c\u5b83\u6bd4\u8f03\u76f8\u4f3c\u65bc\u6240\u8b02\u7684\u807d\u8aaa\u5f0f\u6559\u5b78\u6cd5\uff0c \u5b78\u7fd2\uff0c\u4f46\u662f\u5176\u9802\u591a\u53ea\u80fd\u88fd\u4f5c\u51fa\u53e5\u5c64\u7d1a(Sentence-level)\u7684\u6548\u679c\uff0c\u5176\u53e5\u5c64\u7d1a\u662f\u6307\u5728\u97f3\u6587 \u52d5\u7684\u65b9\u5f0f\u5207\u97f3\u6210\u8a5e\u5c64\u7d1a Timed-text \u6a94\u7684\u65b9\u6cd5\uff0c\u6b64\u65b9\u6cd5\u9664\u4e86\u53ef\u4ee5\u6e1b\u5c11\u4eba\u529b\u8cc7\u6e90\uff0c\u9084 (\u4e8c) CGUAlign \u8a9e\u97f3\u8fa8\u8b58-ForceAlignment
\u4f46\u5176\u53c8\u8207\u807d\u8aaa\u5f0f\u6559\u5b78\u6cd5\u6709\u4e00\u9ede\u9ede\u4e0d\u540c\uff0c\u8ddf\u8ff0\u7df4\u7fd2\u8207\u807d\u8aaa\u6559\u5b78\u6cd5\u8f03\u70ba\u4e0d\u540c\u7684\u5730\u65b9\u5728\u65bc \u540c\u6b65\u64ad\u653e\u6642\u7576\u4e0b\u8a9e\u97f3\u5167\u5bb9\u662f\u4ee5\u53e5\u5b50\u70ba\u55ae\u4f4d\u7684\u986f\u793a\u65bc\u756b\u9762\u4e2d\u3002\u7531\u800c\u672c\u7814\u7a76\u5247\u662f\u9032\u4e00\u6b65 (\u4e09) \u7db2\u7ad9\u5448\u73fe(Website presentation) \u80fd\u5920\u5927\u5e45\u6e1b\u5c11\u4eba\u5de5\u624b\u52d5\u5207\u97f3\u6240\u6d6a\u8cbb\u7684\u6642\u9593\u3002\u53ea\u9700\u8f38\u5165\u6587\u5b57\u6a94\u4ee5\u53ca\u8072\u97f3\u6a94\uff0c\u7d93\u904e\u97f3\u6587 \u807d\u8aaa\u6559\u5b78\u662f\u6559\u5e2b\u4ee5\u81ea\u8eab\u7684\u6f14\u7e79\u53e3\u8aaa\u5167\u5bb9\u4f86\u4fc3\u4f7f\u5b78\u751f\u53cd\u8986\u7df4\u7fd2\u8a9e\u8a00\u5167\u5bb9\u800c\u8ddf\u8ff0\u7df4\u7fd2 \u6bd4\u8f03\u50be\u5411\u65bc\u5b78\u7fd2\u8005\u81ea\u4e3b\u8a13\u7df4\u7684\u65b9\u5f0f\u3002 \u5229\u7528\u9019\u4e09\u7a2e\u529f\u80fd\uff0c\u5229\u7528\u62fc\u97f3\u548c\u4eba\u5de5\u8a9e\u97f3\u5408\u6210\u7684\u529f\u80fd\u80fd\u5920\u88fd\u4f5c\u51fa Google Translate \u6240 \u7121\u6cd5\u9054\u5230\u7684\u8a5e\u5c64\u7d1a(Word-level)\u7684\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\uff0c\u5176\u8a5e\u5c64\u7d1a\u662f\u6307\u5728\u97f3\u6587\u540c\u6b65\u64ad\u653e \u5c0d\u9f4a\u7684\u8655\u7406\uff0c\u5373\u53ef\u5f97\u5230\u5e36\u6709\u6642\u9593\u9ede\u7684 Timed-text \u6587\u5b57\u6a94\u6848\u3002\u4f46\u6b64\u6280\u8853\u7121\u6cd5\u8655\u7406\u904e \u5716\u4e09\u70ba\u7cfb\u7d71\u6574\u9ad4\u6d41\u7a0b\u5716\uff0c\u539f\u59cb\u6587\u5b57\u6a94\u7d93\u7531(\u4e00)\u6703\u5f97\u5230\u53e5\u5c64\u7d1a\u7684\u5e36\u6709\u6642\u9593\u9ede\u7684\u6587\u5b57\u7684
\u8a9e\u8a00\u5b78\u7fd2\u8005\u8ddf\u8ff0\u7684\u5b78\u7fd2\u5c0d\u8c61\u4e0d\u4e00\u5b9a\u70ba\u771f\u4eba\uff0c\u4e5f\u53ef\u80fd\u50c5\u662f\u4e00\u500b\u8a9e\u97f3\u6216\u5f71\u50cf\u6a94\uff0c\u5728\u8ddf\u8ff0 \u6642\u7576\u4e0b\u7684\u8a9e\u97f3\u5167\u5bb9\u9664\u4e86\u4ee5\u53e5\u5b50\u70ba\u55ae\u4f4d\u7684\u986f\u793a\u65bc\u756b\u9762\u4e2d\u4e26\u9644\u52a0\u53e5\u5b50\u4e2d\u6bcf\u500b\u8a5e\u7684\u986f\u793a \u9577\u7684\u8072\u97f3\u6a94\uff0c\u56e0\u6b64\u5e0c\u671b\u7ad9\u5728\u96f2\u7aef\u8a9e\u97f3\u5408\u6210\u6280\u8853\u4e0a\uff0c\u5c07\u4ee5\"\u53e5\"\u70ba\u5c64\u7d1a\u7684 TTS \u6539\u826f\uff0c\u4f7f \u6a94\u6848\uff0c\u518d\u7d93\u7531(\u4e8c)\u6703\u5f97\u5230\u8a5e\u5c64\u7d1a\u7684\u5e36\u6709\u6642\u9593\u9ede\u7684\u6587\u5b57\u7684\u6a94\u6848\uff0c\u6700\u5f8c\u7d93\u7531(\u4e09)\u80fd\u5920\u4ee5
\u7684\u904e\u7a0b\u4e2d\uff0c\u8ddf\u8ff0\u8005\u4ee5\u81ea\u6211\u6240\u80fd\u7684\u767c\u97f3\u6280\u5de7\u4ee5\u53ca\u95b1\u8b80\u80fd\u529b\u53bb\u76e1\u53ef\u80fd\u5730\u6a21\u4eff\u6240\u8981\u5b78\u7fd2\u7684 \u6548\u679c\uff0c\u800c\u8a5e\u5c64\u7d1a\u7684\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u80fd\u5920\u8b93\u5b78\u7fd2\u8005\u66f4\u5bb9\u6613\u7684\u8033\u807d\u3001\u773c\u770b\u4f86\u505a\u8ddf\u8ff0\u6280\u5de7 \u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\u7684\u65b9\u5f0f\u700f\u89bd\u6b64\u5e36\u6709\u6642\u9593\u9ede\u7684\u6587\u5b57\u7684\u6a94\u6848\uff0c\u4e26\u4ee5\u6b64\u505a\u4e00\u500b\u96fb\u8166\u8f14\u52a9\u8a9e \u5176\u80fd\u5920\u9054\u5230\"\u5b57\"\u7684\u5c64\u7d1a\u3002\u5716\u4e8c\u70ba CGUAlign \u4e4b\u6d41\u7a0b\u5716\u3002 \u8a9e\u8a00\u5c0d\u8c61\u6216\u8005\u662f\u5f71\u97f3\u5167\u5bb9\uff0c\u9019\u7a2e\u5b78\u7fd2\u65b9\u5f0f\u6709\u5982\u9e1a\u9d61\u5b78\u820c\uff0c\u662f\u4e00\u7a2e\u53cd\u8986\u7df4\u7fd2\u4ee5\u53ca\u81ea\u6211 \u5167\u5316\u7684\u904e\u7a0b\uff0c\u5728\u5176\u4ed6\u7684\u7814\u7a76\u4e2d[1]\u6211\u5011\u4e5f\u53ef\u4ee5\u767c\u73fe\u5230\u5229\u7528\u9019\u6a23\u7684\u8a9e\u8a00\u5b78\u7fd2\u6280\u5de7\u662f\u4e00 \u7684\u7df4\u7fd2\u3002 \u5716\u4e8c.a\u3001\u5207\u5230\u53e5\u7684 sbv \u6a94 \u8a00\u5b78\u7fd2\u7684\u52d5\u4f5c\u3002
\u7a2e\u5feb\u901f\u5167\u5316\u65b9\u5f0f\u53bb\u5b78\u7fd2\u4e00\u7a2e\u8a9e\u8a00\u7684\u65b9\u6cd5\u3002 (\u4e09) HTK (Hidden Markov Model Toolkit)[2]
HTK \u7684\u5168\u540d\u70ba Hidden Markov Model Toolkit\uff0c\u662f\u4e00\u5957\u61c9\u7528\u65bc\u8a9e\u97f3\u8a13\u7df4\u8207\u8fa8\u8b58\u7684\u514d (\u4e8c) Google Translate \u73fe\u5728 Google \u5728\u8a31\u591a\u65b9\u9762\u5ee3\u6cdb\u5730\u88ab\u4f7f\u7528\uff0c\u4e0d\u53ea\u662f\u5728\u641c\u5c0b\u5f15\u64ce\u7684\u529f\u80fd\u4e0a\uff0c\u8a31\u591a\u4eba\u5728\u9047 \u5230\u8a9e\u8a00\u4e0a\u554f\u984c\u7684\u6642\u5019\uff0c\u5f80\u5f80\u6703\u85c9\u7531 Google \u6240\u63d0\u4f9b\u7684\u7ffb\u8b6f\u529f\u80fd\u2500 Google Translate \u8cbb\u8edf\u9ad4\u3002HTK \u65bc 1989 \u5e74\u958b\u59cb\u7531\u82f1\u570b\u528d\u6a4b\u5927\u5b78\u5de5\u7a0b\u7cfb (Cambridge University \u5716\u56db\u3001\u96f2\u7aef\u8a9e\u97f3\u5408\u6210\u6d41\u7a0b\u5716 Engineering Department, CUED)\u7684\u6a5f\u5668\u667a\u80fd\u5be6\u9a57\u5ba4 (Machine Intelligence Lab\uff0c\u6216\u662f \u5927\u773e\u8f03\u70ba\u719f\u6089\u7684 Speech Vision and Robotics Group)\u9032\u884c\u958b\u767c\uff0c\u8a72\u5718\u968a\u5229\u7528\u96b1\u85cf\u5f0f\u99ac 1. \u6587\u5b57\u5207\u5272
\u5e6b\u5fd9\u3002 Google Translate \u6240\u63d0\u4f9b\u7684\u7ffb\u8b6f\u529f\u80fd\u975e\u5e38\u5f37\u5927\uff0c\u63d0\u4f9b\u8fd1\u767e\u7a2e\u7684\u8a9e\u8a00\u76f8\u4e92\u7684\u7ffb \u53ef\u592b\u6a21\u578b (HMM)\u5efa\u9020\u51fa\u4e00\u5957 HMM-based \u7684\u8a9e\u97f3\u8fa8\u8b58\u7cfb\u7d71\u30021999 \u5e74\u5341\u4e00\u6708\uff0c\u5fae\u8edf
\u8b6f\uff0c\u800c\u4e14\u5728\u53d6\u5f97\u6b64\u529f\u80fd\u7684\u4fbf\u5229\u6027\u4e0a\uff0c\u4e5f\u662f\u7121\u8207\u502b\u6bd4\uff0c\u64da Google \u7d71\u8a08\uff0c\u81f3 2015 \u5e74 6 \u8cfc\u5165\u64c1\u6709\u6b64\u8edf\u9ad4\u7684 Entropic \u516c\u53f8\uff0c\u4e26\u65bc\u7fcc\u5e74\u5c07 HTK \u5b9a\u4f4d\u70ba\u514d\u8cbb\u8edf\u9ad4\uff0c\u671f\u671b HTK \u56e0 Google Translate TTS \u7121\u6cd5\u76f4\u63a5\u8f38\u5165\u9577\u5ea6\u5927\u65bc 100 \u7684\u5b57\u4e32\uff0c\u56e0\u6b64\u9700\u8981\u5148\u505a\u6587\u5b57\u5206
\u6708 Google Translate \u6bcf\u5929\u9700\u8981\u8655\u7406\u8d85\u904e 1000 \u5104\u7b46\u5b57\u8a5e\u3002 \u5716\u4e00\u662f Google Translate \u7684\u4f7f\u7528\u4ecb\u9762\uff0c\u5176\u63d0\u4f9b\u7684\u5373\u6642\u7ffb\u8b6f\u529f\u80fd\uff0c\u8b93\u4f7f\u7528\u8005\u53ef\u4ee5\u5728\u5de6 \u4f5c\u70ba\u8a9e\u97f3\u8fa8\u8b58\u7684\u5171\u540c\u5e73\u53f0\uff0c\u4fbf\u80fd\u8c50\u5bcc HTK \u7684\u529f\u80fd\u6027\uff0c\u4ee5\u53ca\u63d0\u5347\u8a9e\u97f3\u8fa8\u8b58\u7b49\u76f8\u95dc\u6280 \u5272\uff0c\u5c07\u5176\u9577\u5ea6\u964d\u4f4e\u65bc\u5c0f\u65bc 100\uff0c\u4e26\u7a31\u6b64\u70ba\u53e5\u5c64\u7d1a\u7684\u7d14\u6587\u5b57\u6a94\uff0c\u57fa\u672c\u7684\u5207\u5272\u65b9\u6cd5\u53ea\u5148 \u8853\u3002\u70ba\u4e86\u9054\u5230\u9019\u500b\u76ee\u6a19\uff0cHTK \u5efa\u7f6e\u5b98\u65b9\u7db2\u7ad9\uff0c\u4ee5\u63d0\u4f9b\u958b\u653e\u7684\u5b8c\u6574\u529f\u80fd\u539f\u59cb\u78bc\u53ca\u8aaa \u6309\u7167\u6a19\u9ede\u7b26\u865f\u4f5c\u5207\u5272\u3002 \u660e\u66f8\u3002 Step1:\u6309\u7167\u6a19\u9ede\u7b26\u865f\u505a\u5207\u5272\u4f8b\u5982:
\u908a\u7684\u8f38\u5165\u6b04\u4f4d\u8f38\u5165\u6587\u5b57\uff0c\u7ffb\u8b6f\u7d50\u679c\u6703\u5373\u6642\u5728\u53f3\u908a\u7684\u7d50\u679c\u6846\u986f\u793a\uff0c\u5c07\u6ed1\u9f20\u9f20\u6a19\u79fb\u5230\u7ffb \u4e00\u3001\u7dd2\u8ad6 \u96a8\u8457\u5730\u7403\u6751\u7684\u8da8\u52e2\u4f86\u81e8\uff0c \u300c\u8a9e\u8a00\u5b78\u7fd2\u300d \u662f\u73fe\u4eca\u793e\u6703\u666e\u7f85\u5927\u773e\u6240\u9700\u8981\u9762\u81e8\u7684\u4e00\u9805\u8ab2\u984c\uff0c \u8b6f\u7d50\u679c\u6587\u5b57\u4e0a\u53ef\u4ee5\u770b\u5230\u5176\u5c0d\u61c9\u7684\u539f\u6587\uff0c\u5716\u4e00\u7bc4\u4f8b\u70ba\u5c07\u4e2d\u6587\u7ffb\u8b6f\u6210\u82f1\u6587\u7684\u7bc4\u4f8b\u3002 \u7531\u65bc\u8a9e\u97f3\u8fa8\u8b58\u7684\u539f\u7406\u5305\u542b\u76f8\u7576\u9ad8\u6df1\u7684\u6578\u5b78\uff0c\u76f8\u5c0d\u5730\u4f7f\u5f97\u7a0b\u5f0f\u78bc\u4e5f\u4e0d\u6613\u64b0\u5beb\uff0c\u9020\u6210\u9032 \u5165\u9580\u6abb\u9ad8\uff0c\u8907\u96dc\u5ea6\u4e0d\u6613\u638c\u63a7\u7684\u60c5\u6cc1\u7522\u751f\u3002\u4f46\u81ea\u5f9e HTK \u5728 2000 \u5e74\u5b9a\u4f4d\u6210\u958b\u653e\u539f\u59cb\u78bc \u5716\u4e8c.b\u3001trs \u6a94
\u7684\u514d\u8cbb\u8edf\u9ad4\u5f8c\uff0c\u5927\u5e45\u964d\u4f4e\u4e86\u9032\u5165\u9580\u6abb\uff0c\u4e26\u52a0\u901f\u63d0\u6607\u8a9e\u97f3\u6280\u8853\u7684\u767c\u5c55\uff0c\u7d9c\u89c0\u76ee\u524d\u570b\u5167 \u4e5f\u662f\u4e00\u7a2e\u8da8\u52e2\uff0c\u56e0\u6b64\u57f9\u990a\uf97c\u597d\u7684\u591a\u570b\u8a9e\u8a00\u80fd\u529b\uff0c\u5df2\u6210\u70ba\u7576\u4eca\u793e\u6703\u4e0d\u53ef\u6216\u7f3a\u7684\u76ee\u6a19\u3002 \u91dd\u5c0d\u65bc\u53f0\u7063\u4eba\u800c\u8a00\uff0c\u82f1\u8a9e\u5b78\u7fd2\u7684\u9700\u6c42\u66f4\u662f\u986f\u5f97\u6bd4\u5176\u4ed6\u8a9e\u8a00\u4f86\u5f97\u66f4\u70ba\u91cd\u8981\uff0c\u4e8b\u5be6\u4e0a\u6211 \u5916\u8a9e\u97f3\u6280\u8853\u76f8\u95dc\u7684\u5be6\u9a57\u5de5\u5177\u548c\u7cfb\u7d71\u958b\u767c\uff0c\u7d55\u5927\u90e8\u5206\u90fd\u4ee5 HTK \u70ba\u4e3b\u6d41\uff1b\u7531\u6b64\u53ef\u77e5\uff0c \u5716\u4e09\u3001\u7cfb\u7d71\u6d41\u7a0b\u5716
\u5011\u77e5\u9053\uff0c \u300c\u8a9e\u8a00\u5b78\u7fd2\u300d\u4e26\u975e\u53ea\u662f\u5982\u540c\u4e00\u822c\u8ab2\u7a0b\u7684\u5b78\u7fd2\uff0c\u53c8\u5206\u70ba\u300c\u807d\u300d \u3001 \u300c\u8aaa\u300d \u3001 \u300c\u8b80\u300d \u3001 HTK \u5728\u8a9e\u97f3\u6280\u8853\u7684\u7814\u7a76\u9818\u57df\u5360\u4e86\u4e0d\u53ef\u6216\u7f3a\u7684\u5730\u4f4d\u3002 \u5716\u4e8c\u3001CGUAlign \u6d41\u7a0b\u5716
\u300c\u5beb\u300d \uff0c\u5176\u9700\u8981\u7d93\u904e\u300c\u81ea\u6211\u5167\u5316\u300d \u3001 \u300c\u7df4\u7fd2\u300d \u3001 \u300c\u6f14\u7e79\u300d\u7b49\u904e\u7a0b\u624d\u80fd\u6839\u6df1\u8482\u56fa\u7684\u8a18\u61b6\u5728 (\u4e00) \u96f2\u7aef\u8a9e\u97f3\u5408\u6210(Text-to-speech,TTS)
\u6211\u5011\u8166\u6d77\u4e2d\uff0c\u800c\u5728\u820a\u6709\u7684\u81ea\u6211\u5b78\u7fd2\u4e2d\uff0c\u53c8\u7f3a\u4e4f\u7368\u7279\u7684\u8a9e\u8a00\u5b78\u7fd2\u74b0\u5883\uff0c\u7f3a\u4e4f\u7df4\u7fd2\u7684\u5c0d
\u8c61\uff0c\u5982\u679c\u8981\u8acb\u4ed6\u4eba\u4f86\u6307\u5c0e\u6559\u5b78\uff0c\u5f80\u5f80\u53c8\u6240\u8cbb\u4e0d\u8cb2\uff0c\u800c\u6578\u4f4d\u5316\u96f2\u7aef\u5b78\u7fd2\u5728\u7576\u4eca\u7684\u4e16\u4ee3 \u672c\u7bc0\u5c07\u8aaa\u660e\u5982\u4f55\u5c07\u7d14\u6587\u672c\u7d93\u7531\u6587\u5b57\u7684\u9810\u5148\u8655\u7406\uff0c\u900f\u904e Google Translate \u7684\u96f2\u7aef\u6587\u5b57 \u662f\u4e00\u500b\u71b1\u9580\u7684\u8da8\u52e2\uff0c\u5982\u300c\u8996\u8a0a\u6559\u5b78\u300d \u3001 \u300c\u7dda\u4e0a\u5b78\u7fd2\u300d \uff0c\u9019\u4e9b\u90fd\u662f\u7db2\u8def\u666e\u53ca\u8207\u8cc7\u8a0a\u767c\u5c55 \u8f49\u8a9e\u97f3(Text-to-speech,TTS)\u7684\u670d\u52d9\uff0c\u53d6\u5f97 TTS \u7684\u8a9e\u97f3\u6a94\uff0c\u4e26\u5c07\u6b64\u8a9e\u97f3\u6a94\u548c\u6587\u5b57\u6a94\u7d50 \u4e0b\u7684\u91cd\u8981\u7522\u7269\uff0c\u82e5\u6211\u5011\u53ef\u4ee5\u5229\u7528\u9069\u5ea6\u7684\u96fb\u8166\u56de\u994b\u7d50\u5408\u6578\u4f4d\u5316\u5b78\u7fd2\uff0c\u4e5f\u8a31\u80fd\u70ba\u66f4\u591a\u4f7f \u7528\u8005\u9020\u5c31\u4e00\u500b\u65b0\u5f62\u614b\u7684\u81ea\u6211\u5b78\u7fd2\u65b9\u5f0f\u3002 \u5408\u7522\u751f\u53e5\u5c64\u7d1a\u7684 Timed-text \u6a94\u6848\u4ee5\u4f9b\u4e0b\u4e00\u968e\u6bb5 CGUAlign \u4f7f\u7528\u3002
\u672c \u7814 \u7a76 \u8a2d \u8a08 \u4e00 \u500b \u65b9 \u4fbf \u8655 \u7406 \u6709 \u8072 \u66f8 \u97f3 \u6587 \u540c \u6b65 \u7684 \u6280 \u8853 \uff0c \u5229 \u7528 \u96f2 \u7aef \u7684 \u6587 \u5b57 \u8f49 \u8a9e \u97f3 \u5716\u4e00\u3001Google Translate \u7db2\u9801\u4ecb\u9762
(Text-to-speech)\u6280\u8853\uff0c\u7d50\u5408\u8a9e\u97f3\u8fa8\u8b58(Speech Recognition)\u6280\u8853\uff0c\u8b93\u4f7f\u7528\u8005\u80fd\u5920\u4f7f\u7528
\u81ea\u884c\u6e96\u5099\u7684\u6587\u672c\u4f86\u88fd\u4f5c\u81ea\u5df1\u7684\u8ddf\u8ff0\u7df4\u7fd2\u7684\u5b78\u7fd2\u7d20\u6750\uff0c\u88fd\u4f5c\u9054\u5230\u8a5e\u5c64\u7d1a(Word-level)
\u7684\u97f3\u6587\u540c\u6b65\u6709\u8072\u66f8\uff0c\u5176\u4e0d\u50c5\u53ef\u4ee5\u63d0\u4f9b\u97f3\u6587\u540c\u6b65\u7684\u96fb\u5b50\u66f8\u4f9b\u4f7f\u7528\u8005\u95b1\u8b80\u6587\u7ae0\uff0c\u4e5f\u53ef\u4ee5 \u8b93\u4f7f\u7528\u8005\u85c9\u7531\u6717\u8aa6\u6587\u7ae0\u7684\u65b9\u5f0f\uff0c\u4e26\u900f\u904e\u8ddf\u8ff0\u7df4\u7fd2\u7684\u5be6\u4f5c\u548c\u5373\u6642\u7ffb\u8b6f\u7684\u6548\u679c\uff0c\u4ee5\u9054\u5230 \u5716\u4e8c.c\u3001lrc \u6a94
", "text": "Perl \u5305\u88dd HTK \u7684\u65b9\u6cd5\uff0cCguAlign \u6539\u7528 Python \u5c07 HTK \u5305\u88dd\u3001" }, "TABREF1": { "html": null, "num": null, "type_str": "table", "content": "
Translate TTS Parameters \u5728\u6b64\u7bc4\u4f8b\u7a0b\u5f0f\u78bc\u4e2d\uff0c\u5148\u5c0d\u8a2d\u5b9a GOOGLE_TTS_URL \u8f38\u5165\u7db2\u5740\uff0c\u7528 payload \u8f38\u5165\u5c0d
parameters Google Translate \u7684 TTS \u4e4b\u53c3\u6578\u5982\u4e0a\u8ff0\u8868\u4e00\u6240\u793a\uff0c\u6700\u5f8c\u904b\u7528 urllib.request.urlopen() \u610f\u7fa9
tl \u767c\u51fa request \u53d6\u5f97\u53e5\u5b50\u5167\u5bb9\u7684 mp3 \u8a9e\u97f3\uff0c\u7136\u5f8c\u7528 read()\u8b80\u53d6 mp3 \u8a9e\u97f3\u7684\u5167\u5bb9\u4e26\u7528 Target Language\uff0c\u76ee\u6a19\u8a9e\u8a00\uff0c\u8868\u793a\u8981\u6587\u5b57 TTS \u7684\u8a9e\u8a00\u7a2e\u985e\u3002
q len( )\u8a08\u7b97\u51fa\u53e5\u5b50 mp3 \u97f3\u8a0a\u5167\u5bb9\u7684\u9577\u5ea6\u3002 Query\uff0c\u6b32 TTS \u7684\u6587\u5b57\u3002
totalTotal number of text segments\uff0c\u6587\u7ae0\u5206\u6bb5\u7684\u500b\u6578\u3002
idx 3. Creat Timed-text FileIndex of text segments\uff0c\u6587\u7ae0\u5206\u6bb5\u7684\u6307\u6a19\u3002
textlenString length in this segment\uff0c\u6b64 Query \u7684\u5b57\u4e32\u9577\u5ea6\u3002
\u5229\u7528\u4e0a\u4e00\u6b65\u9a5f\u6240\u8490\u96c6\u7684\u6bcf\u4e00\u500b segment \u7684 byteNum \u5927\u5c0f\uff0c\u4e26\u8a08\u7b97 byteNum \u7684\u7e3d\u548c\uff0c
\u80fd \u5920 \u8a08 \u7b97 \u51fa \u6bcf \u4e00 \u6bb5 segment \u5728 \u7e3d \u8a9e \u97f3 \u9577 \u5ea6 \u4e2d \u7684 \u6642 \u9593 \u9577 \u5ea6 \u5176 \u516c \u5f0f \u5982 \u4e0b \uff0c
1 2import urllib.request SegmentLength(i)\u70ba\u7b2c i \u6bb5\u8a9e\u97f3\u7684\u9577\u5ea6\uff0cByteNum(i)\u70ba\u7b2c i \u6bb5\u8a9e\u97f3\u7684\u6a94\u6848\u5927\u5c0f\uff0c import urllib.parse Sum(ByteNum)\u70ba\u7e3d\u5171\u7684\u6a94\u6848\u5927\u5c0f\uff0cTotalLength \u70ba\u7e3d\u8a9e\u97f3\u9577\u5ea6\u3002
3 4savefile=\"./TTS.mp3\" f= open(savefile, 'wb+') Sum ByteNum ByteNum i i gth SegmentLen \uf03d ( ) ( ) ()\uf0b4TotalLength
5\u6587\u5b57= \"Chung Gung University Student\"
6GOOGLE_TTS_URL= 'https://translate.google.com.tw/translate_tts?'
7payload = { 'ie': 'utf-8',
8'tk': '308912',
9'client': 't',
10'tl': 'en',
11'q': \u6587\u5b57,
12'total': 1,
13'idx': 0,
14'textlen': len(text) }
15try:
19hdr = {r = urllib.request.urlopen(req)
20
21
22byte= r.read()
23f.write(byte)
24byteNum= len(byte)
25except Exception as e:
26raise
27f.close()
\u5716\u4e94\u3001Communicate to Google \u7bc4\u4f8b\u7a0b\u5f0f\u78bc
", "text": "" }, "TABREF2": { "html": null, "num": null, "type_str": "table", "content": "
\u4ee5\u53ca\u5176 parameter \u6539\u6210\u5982\u8868\u4e8c\uff0c\u5373\u53ef\u5f97\u5230\u6b64\u539f\u6587\u7684\u7f85\u99ac\u62fc\u97f3\u3002
\u8868\u4e8c\u3001\u53d6\u5f97\u7f85\u99ac\u62fc\u97f3\u7684 parameters(\u4ee5\u4e2d\u6587\u70ba\u4f8b)
parameters\u503cparameters\u503c
ieUTF-8kc1
inputm1tk520254|125262
oeUTF-8dtbd
otf1dtex
trs1dtld
clientTdtmd
slZh-CNdtqca
hlZh-TWdtrw
rom1dtrm
srcrom1dtss
ssel0dtt
tsel0dtat
tl\u76ee\u6a19\u8a9e\u8a00(zh-TW)q\u6b32\u53d6\u5f97\u62fc\u97f3\u7684\u6587\u5b57
0:0:0.000000,0:0:5.760000
\u8a71\u8aaa\u5c71\u6771\u767b\u5dde\u5e9c\u6771\u9580\u5916\u6709\u4e00\u5ea7\u5927\u5c71\uff0c\u540d\u53eb\u84ec\u840a\u5c71\u3002
0:0:5.760000,0:0:9.144000
\u5c71\u4e0a\u6709\u500b\u95a3\u5b50\uff0c\u540d\u53eb\u84ec\u840a\u95a3\u3002
0:0:9.144000,0:0:13.968000
\u9019\u95a3\u9020\u5f97\u756b\u68df\u98db\u96f2\uff0c\u73e0\u7c3e\u6372\u96e8\uff0c\u5341\u5206\u58ef\u9e97\u3002
0:0:0.000000,0:0:5.760000
Hu\u00e0shu\u014d sh\u0101nd\u014dng d\u0113ng zh\u014du f\u01d4 d\u014dngm\u00e9n w\u00e0i y\u01d2u y\u012bzu\u00f2 d\u00e0sh\u0101n, m\u00edng ji\u00e0o p\u00e9ngl\u00e1i
sh\u0101n.
0:0:5.760000,0:0:9.144000
Sh\u0101nsh\u00e0ng y\u01d2u g\u00e8 g\u00e9 zi, m\u00edng ji\u00e0o p\u00e9ngl\u00e1i g\u00e9.
", "text": "Timed-text \u6a94\u6848\u7bc4\u4f8b \u82e5\u8f38\u5165\u6587\u672c\u4e0d\u70ba\u82f1\u6587\u6642\uff0c\u9700\u8981\u5f9e Google Translate \u53d6\u5f97\u5176\u539f\u6587\u7684\u7f85\u99ac\u62fc\u97f3\uff0c\u4e26\u4e14\u5c07 \u6b64\u7f85\u99ac\u62fc\u97f3\u53d6\u4ee3\u539f\u6587\uff0c\u5c07\u53e5\u5c64\u7d1a\u7684\u539f\u6587\u6587\u5b57\u6a94\u8f49\u6210\u53e5\u5c64\u7d1a\u7684\u7f85\u99ac\u62fc\u97f3\u6a94\uff0c\u800c\u82e5\u5229\u7528 Google Translate \u53d6\u5f97\u62fc\u97f3\uff0c\u5176\u4e5f\u6703\u5e6b\u6211\u5011\u505a\u65b7\u8a5e\u7684\u52d5\u4f5c\u3002 \u5229\u7528\u540c Communicate to Google \u7684\u65b9\u6cd5\uff0c\u53ea\u8981\u5c07 URL \u6539\u6210\uff0c http://translate.google.com.tw/translate_a/single \u5716\u4e03\u3001\u975e\u82f1\u6587\u6587\u5b57 sbv \u6a94\u8f49\u7f85\u99ac\u62fc\u97f3 sbv \u6a94 4. Audio Converter \u672c\u7bc0\u65e8\u5728\u8aaa\u660e\u5982\u4f55\u5c07 Google Translate TTS \u6240\u5f97\u7684 mp3 \u6a94\u6848\u8f49\u6210 CguAlign \u80fd\u63a5\u53d7\u7684 wav \u6a94\u6848\uff0c\u4f7f\u7528\u81ea\u7531\u8edf\u9ad4\u2500FFmpeg \u4f86\u5e6b\u52a9\u8f49\u6a94\uff0cFFmpeg \u53ef\u4ee5\u57f7\u884c\u97f3\u8a0a\u548c\u8996\u8a0a\u591a\u7a2e \u683c\u5f0f\u7684\u9304\u5f71\u3001\u8f49\u6a94\u3001\u4e32\u6d41\u529f\u80fd\uff0c\u56e0\u9700\u501f\u52a9 FFmpeg \u7684\u5e6b\u52a9\uff0c\u800c FFmpeg \u5c6c\u65bc\u5916\u90e8\u7a0b \u5f0f\uff0c\u5728 Python \u4e2d\u82e5\u9700\u8981\u547c\u53eb\u5916\u90e8\u7684\u7a0b\u5f0f\uff0c\u9700\u8981 import os \u6a21\u7d44\uff0c\u4e26\u4e14\u4f7f\u7528 os.system()" }, "TABREF4": { "html": null, "num": null, "type_str": "table", "content": "
1var audioFileName=\"YourAudioFile.wav\" \u5716\u5341\u4e94\u3001\u7dda\u4e0a\u5373\u6642\u7ffb\u8b6f\u7bc4\u4f8b
2 3document.write(\"<div><audio id='mainAudio' src='' controls=controls /></div>\"); document.getElementById(\"mainAudio\").src=audioFileName; \u56db\u3001\u7d50\u8ad6
4 5var playrate= mainAudio.playbackRate document.getElementById(\"playrate\").innerHTML =playrate; \u672c\u7814\u7a76\u7684\u76ee\u7684\u662f\u5229\u7528\u7d14\u6587\u5b57\u6a94\u8f49\u6210\u8a9e\u97f3\u6a94\u7684\u6280\u8853(Text-to-speech)\u7d50\u5408\u8a9e\u97f3\u8fa8\u8b58
6var time= mainAudio.currentTime; (Speech-recognition)\u4e2d\u7684\u97f3\u6587\u5c0d\u9f4a\u6280\u8853(Speech-text Synchronization)\u88fd\u4f5c\u80fd\u5920\u4ee5\u96fb\u8166\u8f14
7document.getElementById(\"audiotime\").innerHTML =time; \u52a9\u8a9e\u8a00\u5b78\u7fd2(Computer-assisted Language Learning)\u70ba\u76ee\u6a19\uff0c\u5e6b\u52a9\u8a9e\u8a00\u5b78\u7fd2\u8005\u80fd\u5920\u501f\u52a9\u6b64
\u5716\u5341\u56db\u3001\u8b80\u53d6\u97f3\u8a0a\u7684\u7bc4\u4f8b\u7a0b\u5f0f\u78bc \u7cfb\u7d71\u8f03\u8f15\u9b06\u5730\u5be6\u73fe\u8ddf\u8ff0\u5b78\u7fd2\u6cd5(Shadowing technique)\u7684\u4e00\u500b\u7cfb\u7d71\u3002
\u5728\u5716\u5341\u56db\u7684\u7bc4\u4f8b\u4e2d\uff0c\u5728\u7b2c 2 \u884c\u5148\u5275\u7acb\u4e00\u500b audio \u7269\u4ef6\uff0c\u4e26\u4e14\u5728\u7b2c 3 \u884c\u6307\u5b9a\u6b64\u7269\u4ef6\u6240 \u5728\u6b64\u7cfb\u7d71\u4e2d\uff0c\u4f7f\u7528\u8005\u80fd\u5920\u81ea\u7531\u5730\u53d6\u5f97\u4efb\u4f55\u60f3\u8ddf\u8ff0\u7684\u7d20\u6750\u7684\u6587\u672c\uff0c\u4ee5\u6b64\u6587\u672c\uff0c\u501f\u52a9\u672c\u7cfb\u7d71\uff0c
\u8981\u8b80\u53d6\u7684\u97f3\u8a0a\u6a94\u6848\uff0c\u7b2c 5 \u884c\u3001\u7b2c 7 \u884c\u80fd\u5920\u5f97\u5230\u6b64\u97f3\u8a0a\u7684\u64ad\u653e\u901f\u5ea6\u548c\u76ee\u524d\u6240\u64a5\u653e\u7684\u97f3 \u80fd\u5920\u5f9e Google Translate \u53d6\u5f97\u6b64\u6587\u672c\u7684 TTS \u8a9e\u97f3\u6a94\u53ca\u5176\u5df2\u5c0d\u9f4a\u7684\u5e36\u6709\u6642\u9593\u9ede\u7684\u6587\u672c
\u8a0a\u6642\u9593\u9ede\uff0c\u6b64\u97f3\u8a0a\u6642\u9593\u9ede\u662f\u7528\u4f86\u505a\u97f3\u6587\u540c\u6b65\u975e\u5e38\u91cd\u8981\u7684\u8cc7\u8a0a\u3002 (Timed-text)\uff0c\u4e0d\u540c\u65bc\u4ee5\u5f80\u8f03\u5e38\u898b\u7684\u53e5\u5c64\u7d1a(Sentence-level)\u7684\u6587\u672c\uff0c\u672c\u7cfb\u7d71\u904b\u7528\u8a9e\u97f3\u8fa8\u8b58
\u6280\u8853\u80fd\u5920\u88fd\u4f5c\u51fa\u8a5e\u5c64\u7d1a(Word-level)\u7684\u6587\u672c\uff0c\u4ee5\u6b64\u5e36\u6709\u6642\u9593\u9ede\u7684\u6587\u672c\u85c9\u7531\u6211\u5011\u7684\u7db2\u7ad9\u700f
\u4ee5\u985e\u540c\u524d\u8ff0\u53d6\u5f97\u4e2d\u6587\u62fc\u97f3\u7684\u65b9\u6cd5\uff0c\u4e0d\u50c5\u80fd\u5920\u53d6\u5f97\u55ae\u5b57\u7684\u62fc\u97f3\uff0c\u540c\u6a23\u4e5f\u80fd\u5920\u53d6\u5f97\u55ae\u5b57 \u89bd\uff0c\u5373\u662f\u4e00\u672c\u97f3\u6587\u540c\u6b65\u7684\u96fb\u5b50\u6709\u8072\u66f8\uff0c\u5728\u6b64\u7db2\u7ad9\u4e0a\uff0c\u4e0d\u50c5\u53ef\u4ee5\u9032\u884c\u8ddf\u8ff0\u5b78\u7fd2\u6cd5\u5b78\u7fd2\uff0c\u4e5f
\u7684\u7ffb\u8b6f\uff0c\u6211\u5011\u53ef\u4ee5\u7528\u6b64\u529f\u80fd\u4f86\u5be6\u4f5c\u7dda\u4e0a\u67e5\u8a62\u5b57\u5178\u7684\u529f\u80fd\uff0c\u8207\u62fc\u97f3\u53d6\u5f97\u7684\u65b9\u6cd5\u4e0d\u540c\u7684 \u53ef\u4ee5\u505a\u4e00\u500b\u7dda\u4e0a\u67e5\u8a62\u5b57\u5178\u7684\u529f\u80fd\uff0c\u5176\u4e0d\u50c5\u53ef\u4ee5\u63d0\u4f9b\u97f3\u6587\u540c\u6b65\u7684\u96fb\u5b50\u66f8\u4f9b\u4f7f\u7528\u8005\u95b1\u8b80\u6587\u7ae0\uff0c
\u662f\uff0c\u7ffb\u8b6f\u529f\u80fd\u5fc5\u9808\u6307\u5b9a\u597d sl \u548c tl \u5169\u500b\u53c3\u6578\uff0c\u5176\u610f\u7fa9\u4ee3\u8868 source language \u4f86\u6e90\u8a9e\u8a00 \u4e5f\u53ef\u4ee5\u8b93\u4f7f\u7528\u8005\u85c9\u7531\u6717\u8aa6\u6587\u7ae0\u7684\u65b9\u5f0f\uff0c\u4e26\u900f\u904e\u8ddf\u8ff0\u5b78\u7fd2\u6cd5\u7684\u5be6\u4f5c\u548c\u5373\u6642\u7ffb\u8b6f\u7684\u6548\u679c\uff0c\u4ee5
\u548c target language \u76ee\u6a19\u8a9e\u8a00\u3002 \u9054\u5230\u81ea\u6211\u5167\u5316\u5b78\u7fd2\u53ca\u589e\u9032\u8a9e\u8a00\u80fd\u529b\u3002
", "text": "\u5716\u5341\u4e09\u3001\u5c07 lrc \u6a94\u8f49\u6210\u7db2\u9801\u4e0a\u7684\u6a19\u7c64\u8cc7\u8a0a \u4f7f\u7528 HTML5 \u65b0\u589e\u7684 audio \u6a19\u7c64\uff0c\u80fd\u5920\u5148\u5275\u5efa\u4e00\u500b\u64ad\u653e\u97f3\u8a0a\u7684\u7269\u4ef6\uff0c\u8ce6\u4e88\u6b64\u7269\u4ef6\u4e00 \u500b\u7368\u7279\u7684 ID\u2500mainAudio\uff0c\u518d\u5229\u7528 HTML DOM \u7684\u7269\u4ef6\uff0c\u80fd\u5920\u6307\u5b9a mainAudio \u8981\u8b80 \u53d6\u7684\u97f3\u8a0a\u6a94\u6848\u4ee5\u53ca\u6b64\u97f3\u8a0a\u7684\u4e00\u4e9b\u8cc7\u8a0a\u3002" } } } }