{"id":5900,"date":"2024-09-09T17:19:07","date_gmt":"2024-09-09T09:19:07","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=5900"},"modified":"2025-01-06T20:57:27","modified_gmt":"2025-01-06T12:57:27","slug":"crisperwhisper","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/ja\/crisperwhisper\/","title":{"rendered":"CrisperWhisper\uff1a\u7cbe\u786e\u7684\u9010\u5b57\u8bed\u97f3\u8f6c\u5f55\u5de5\u5177"},"content":{"rendered":"<p>CrisperWhisper \u662f\u4e00\u4e2a\u57fa\u4e8e OpenAI Whisper \u7684\u9ad8\u7ea7\u8bed\u97f3\u8bc6\u522b\u5de5\u5177\uff0c\u4e13\u6ce8\u4e8e\u5feb\u901f\u3001\u7cbe\u786e\u548c\u9010\u5b57\u7684\u8bed\u97f3\u8f6c\u5f55\u3002\u5b83\u80fd\u591f\u63d0\u4f9b\u7cbe\u786e\u7684\u5355\u8bcd\u7ea7\u65f6\u95f4\u6233\uff0c\u751a\u81f3\u5728\u8bed\u97f3\u586b\u5145\u548c\u505c\u987f\u7684\u60c5\u51b5\u4e0b\u4e5f\u80fd\u51c6\u786e\u8bb0\u5f55\u3002CrisperWhisper \u901a\u8fc7\u8c03\u6574\u6807\u8bb0\u5668\u548c\u81ea\u5b9a\u4e49\u6ce8\u610f\u529b\u635f\u5931\u6765\u63d0\u9ad8\u65f6\u95f4\u6233\u7684\u7cbe\u5ea6\uff0c\u5e76\u51cf\u5c11\u8f6c\u5f55\u5e7b\u89c9\uff0c\u786e\u4fdd\u6bcf\u4e2a\u53d1\u97f3\u7684\u5355\u8bcd\u90fd\u88ab\u51c6\u786e\u8bb0\u5f55\u3002<\/p>\n<p>&nbsp;<\/p>\n<blockquote>\n<h3>\u8bba\u6587\u7e3d\u7d50<\/h3>\n<p>CrisperWhisper \u662f\u4e00\u6b3e\u6539\u9032\u7248\u7684 Whisper \u8a9e\u97f3\u8b58\u5225\u6a21\u578b\uff0c\u901a\u904e\u8abf\u6574 tokenizer \u548c\u4f7f\u7528\u52d5\u614b\u6642\u9593\u626d\u66f2\uff08DTW\uff09\u7b97\u6cd5\uff0c\u5be6\u73fe\u4e86\u66f4\u7cbe\u78ba\u7684\u8a5e\u7d1a\u6642\u9593\u6233\u8a18\uff0c\u4e26\u80fd\u5920\u63d0\u4f9b\u66f4\u7d30\u7dfb\u7684\u8a9e\u97f3\u8f49\u9304\uff0c\u589e\u5f37\u5c0d\u8a9e\u97f3\u4e2d\u66ab\u505c\u548c\u586b\u8a5e\u4e8b\u4ef6\u7684\u6aa2\u6e2c\u80fd\u529b\uff0c\u4e26\u964d\u4f4e\u865b\u5e7b\uff08hallucinations\uff09\u7684\u7522\u751f\u3002<\/p>\n<h4>\u6458\u8981<\/h4>\n<p>CrisperWhisper \u662f\u57fa\u65bc Whisper \u6a21\u578b\u7684\u4e00\u9805\u6539\u9032\uff0c\u65e8\u5728\u63d0\u4f9b\u66f4\u7cbe\u78ba\u7684\u8a5e\u7d1a\u6642\u9593\u6233\u8a18\u548c\u66f4\u7d30\u7dfb\u7684\u8a9e\u97f3\u8f49\u9304\u3002\u8a72\u6a21\u578b\u901a\u904e\u8abf\u6574 Whisper \u7684 tokenizer\uff0c\u4f7f\u5f97 DTW \u7b97\u6cd5\u80fd\u5920\u66f4\u6e96\u78ba\u5730\u5c0d\u9f4a\u97f3\u983b\u7247\u6bb5\u8207\u8a5e\u8a9e\uff0c\u5f9e\u800c\u63d0\u9ad8\u4e86\u6642\u9593\u6233\u8a18\u7684\u7cbe\u78ba\u5ea6\u3002\u9019\u9805\u6280\u8853\u5c0d\u65bc\u6355\u6349\u6240\u6709\u767c\u97f3\u7684\u8a9e\u97f3\u8f49\u9304\u7279\u5225\u6709\u7528\uff0c\u9019\u5c0d\u65bc\u81e8\u5e8a\u8a55\u4f30\u8a9e\u97f3\u3001\u5206\u6790\u8a9e\u8a00\u898f\u5283\u904e\u7a0b\u4ee5\u53ca\u8b58\u5225\u8a8d\u77e5\u8ca0\u8377\u6307\u6a19\u975e\u5e38\u91cd\u8981\u3002CrisperWhisper \u9084\u901a\u904e\u8a13\u7df4\u548c\u5c0d\u6297\u566a\u97f3\u7684\u65b9\u6cd5\uff0c\u63d0\u9ad8\u4e86\u5c0d\u55ae\u4e00\u8072\u97f3\u6e90\u7684\u95dc\u6ce8\u5ea6\u548c\u566a\u97f3\u9069\u61c9\u6027\uff0c\u4e26\u4e14\u5728\u591a\u500b\u57fa\u6e96\u6578\u64da\u96c6\u4e0a\u9032\u884c\u4e86\u6e2c\u8a66\uff0c\u5c55\u793a\u4e86\u5176\u5728\u8a9e\u97f3\u8b58\u5225\u3001\u8a5e\u6bb5\u5206\u5272\u3001\u586b\u8a5e\u4e8b\u4ef6\u6aa2\u6e2c\u4ee5\u53ca\u964d\u4f4e\u865b\u5e7b\u5167\u5bb9\u65b9\u9762\u7684\u512a\u7570\u6027\u80fd\u3002\u6b64\u5916\uff0c\u8a72\u6a21\u578b\u7684\u4ee3\u78bc\u548c\u4e00\u500b\u5177\u6709\u6e96\u78ba\u8a5e\u7d1a\u6642\u9593\u6233\u8a18\u7684\u5408\u6210\u6578\u64da\u96c6\u5df2\u7d93\u958b\u6e90\u3002<\/p>\n<h3>\u89c0\u9ede<\/h3>\n<ul>\n<li><strong>\u6539\u9032\u7684 Tokenizer<\/strong>\uff1aCrisperWhisper \u901a\u904e\u79fb\u9664 tokenizer \u4e2d\u7684\u591a\u9918\u7a7a\u683c\uff0c\u4e26\u5c0d\u7279\u5b9a\u8a5e\u8a9e\uff08\u5982 &#8220;uh&#8221; \u548c &#8220;um&#8221;\uff09\u9032\u884c\u91cd\u65b0\u6a19\u8a18\uff0c\u4f7f\u5f97 DTW \u7b97\u6cd5\u80fd\u5920\u66f4\u6e96\u78ba\u5730\u5c0d\u9f4a\u97f3\u983b\u7247\u6bb5\u8207\u8a5e\u8a9e\uff0c\u5f9e\u800c\u63d0\u9ad8\u4e86\u6642\u9593\u6233\u8a18\u7684\u7cbe\u78ba\u5ea6\u3002<\/li>\n<li><strong>\u5c0d\u6297\u566a\u97f3\u7684\u6280\u8853<\/strong>\uff1a\u6a21\u578b\u901a\u904e\u8a13\u7df4\u6642\u52a0\u5165\u566a\u97f3\u548c\u591a\u8072\u97f3\u6e90\u7684\u6578\u64da\uff0c\u63d0\u9ad8\u4e86\u5c0d\u566a\u97f3\u7684\u9069\u61c9\u6027\uff0c\u4e26\u4e14\u901a\u904e\u5f15\u5165\u7a7a\u767d\u7684\u8a13\u7df4\u6a23\u672c\u4f86\u6e1b\u5c11\u865b\u5e7b\u5167\u5bb9\u7684\u7522\u751f\u3002<\/li>\n<li><strong>\u512a\u7570\u7684\u6027\u80fd<\/strong>\uff1aCrisperWhisper \u5728\u591a\u500b\u57fa\u6e96\u6578\u64da\u96c6\u4e0a\u9032\u884c\u4e86\u6e2c\u8a66\uff0c\u5305\u62ec AMI Meeting Corpus\u3001TED-LIUM \u548c LibriSpeech\uff0c\u4e26\u5728\u9019\u4e9b\u6578\u64da\u96c6\u4e0a\u5c55\u793a\u4e86\u512a\u7570\u7684\u8a5e\u7d1a\u6642\u9593\u6233\u8a18\u548c\u8a9e\u97f3\u8b58\u5225\u6027\u80fd\u3002<\/li>\n<li><strong>\u958b\u6e90\u4ee3\u78bc\u548c\u6578\u64da\u96c6<\/strong>\uff1a\u8a72\u6a21\u578b\u7684\u4ee3\u78bc\u548c\u4e00\u500b\u5408\u6210\u7684\u8a9e\u97f3\u6578\u64da\u96c6\u5df2\u7d93\u958b\u6e90\uff0c\u9019\u6709\u52a9\u65bc\u7814\u7a76\u4eba\u54e1\u548c\u958b\u767c\u8005\u9032\u4e00\u6b65\u7814\u7a76\u548c\u6539\u9032\u8a9e\u97f3\u8b58\u5225\u6280\u8853\u3002<\/li>\n<li><strong>\u865b\u5e7b\u5167\u5bb9\u7684\u964d\u4f4e<\/strong>\uff1aCrisperWhisper \u901a\u904e\u7cbe\u78ba\u7684\u6642\u9593\u6233\u8a18\u548c\u5c0d\u865b\u5e7b\u5167\u5bb9\u7684\u7279\u5b9a\u8655\u7406\uff0c\u6709\u6548\u964d\u4f4e\u4e86\u865b\u5e7b\u5167\u5bb9\u7684\u7522\u751f\uff0c\u9019\u5c0d\u65bc\u63d0\u9ad8\u8a9e\u97f3\u8b58\u5225\u7cfb\u7d71\u7684\u53ef\u9760\u6027\u7279\u5225\u91cd\u8981\u3002<\/li>\n<\/ul>\n<\/blockquote>\n<p>&nbsp;<\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li><strong>\u7cbe\u786e\u7684\u5355\u8bcd\u7ea7\u65f6\u95f4\u6233<\/strong>\uff1a\u5373\u4f7f\u5728\u8bed\u97f3\u586b\u5145\u548c\u505c\u987f\u7684\u60c5\u51b5\u4e0b\uff0c\u4e5f\u80fd\u63d0\u4f9b\u7cbe\u786e\u7684\u65f6\u95f4\u6233\u3002<\/li>\n<li><strong>\u9010\u5b57\u8f6c\u5f55<\/strong>\uff1a\u9010\u5b57\u8bb0\u5f55\u6bcf\u4e2a\u53d1\u97f3\u7684\u5355\u8bcd\uff0c\u5305\u62ec\u586b\u5145\u8bcd\u5982\u201c\u55ef\u201d\u548c\u201c\u554a\u201d\u3002<\/li>\n<li><strong>\u586b\u5145\u8bcd\u68c0\u6d4b<\/strong>\uff1a\u68c0\u6d4b\u5e76\u51c6\u786e\u8f6c\u5f55\u586b\u5145\u8bcd\u3002<\/li>\n<li><strong>\u5e7b\u89c9\u51cf\u8f7b<\/strong>\uff1a\u51cf\u5c11\u8f6c\u5f55\u5e7b\u89c9\uff0c\u63d0\u9ad8\u51c6\u786e\u6027\u3002<\/li>\n<li><strong>\u5f00\u6e90\u4ee3\u7801<\/strong>\uff1a\u4ee3\u7801\u516c\u5f00\uff0c\u4fbf\u4e8e\u7528\u6237\u67e5\u770b\u548c\u4f7f\u7528\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<p>&nbsp;<\/p>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<ol>\n<li><strong>\u73af\u5883\u51c6\u5907<\/strong>\uff1a\n<ul>\n<li>\u786e\u4fdd\u5df2\u5b89\u88c5 Python 3.7 \u53ca\u4ee5\u4e0a\u7248\u672c\u3002<\/li>\n<li>\u5b89\u88c5\u5fc5\u8981\u7684\u4f9d\u8d56\u5e93\uff1a<code>pip install -r requirements.txt<\/code>\u3002<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u4e0b\u8f7d\u4ee3\u7801<\/strong>\uff1a\n<ul>\n<li>\u514b\u9686 GitHub \u4ed3\u5e93\uff1a<code>git clone https:\/\/github.com\/nyrahealth\/CrisperWhisper.git<\/code>\u3002<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u8fd0\u884c\u5e94\u7528<\/strong>\uff1a\n<ul>\n<li>\u8fdb\u5165\u9879\u76ee\u76ee\u5f55\uff1a<code>cd CrisperWhisper<\/code>\u3002<\/li>\n<li>\u8fd0\u884c\u5e94\u7528\uff1a<code>python app.py<\/code>\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<h3>\u4f7f\u7528\u6307\u5357<\/h3>\n<ol>\n<li><strong>\u57fa\u672c\u4f7f\u7528<\/strong>\uff1a\n<ul>\n<li>\u6253\u5f00\u5e94\u7528\u540e\uff0c\u4e0a\u4f20\u9700\u8981\u8f6c\u5f55\u7684\u97f3\u9891\u6587\u4ef6\u3002<\/li>\n<li>\u9009\u62e9\u8f6c\u5f55\u6a21\u5f0f\uff08\u9010\u5b57\u8f6c\u5f55\u6216\u6807\u51c6\u8f6c\u5f55\uff09\u3002<\/li>\n<li>\u70b9\u51fb\u201c\u5f00\u59cb\u8f6c\u5f55\u201d\u6309\u94ae\uff0c\u7b49\u5f85\u8f6c\u5f55\u5b8c\u6210\u3002<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u9ad8\u7ea7\u529f\u80fd<\/strong>\uff1a\n<ul>\n<li><strong>\u65f6\u95f4\u6233\u8c03\u6574<\/strong>\uff1a\u5728\u8bbe\u7f6e\u4e2d\u53ef\u4ee5\u8c03\u6574\u65f6\u95f4\u6233\u7684\u7cbe\u5ea6\u3002<\/li>\n<li><strong>\u586b\u5145\u8bcd\u68c0\u6d4b<\/strong>\uff1a\u5f00\u542f\u6216\u5173\u95ed\u586b\u5145\u8bcd\u68c0\u6d4b\u529f\u80fd\u3002<\/li>\n<li><strong>\u5bfc\u51fa\u7ed3\u679c<\/strong>\uff1a\u8f6c\u5f55\u5b8c\u6210\u540e\uff0c\u53ef\u4ee5\u5c06\u7ed3\u679c\u5bfc\u51fa\u4e3a\u6587\u672c\u6587\u4ef6\u6216\u5176\u4ed6\u683c\u5f0f\u3002<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u5e38\u89c1\u95ee\u9898<\/strong>\uff1a\n<ul>\n<li><strong>\u8f6c\u5f55\u4e0d\u51c6\u786e<\/strong>\uff1a\u786e\u4fdd\u97f3\u9891\u8d28\u91cf\u826f\u597d\uff0c\u907f\u514d\u80cc\u666f\u566a\u97f3\u3002<\/li>\n<li><strong>\u65f6\u95f4\u6233\u4e0d\u7cbe\u786e<\/strong>\uff1a\u5c1d\u8bd5\u8c03\u6574\u65f6\u95f4\u6233\u8bbe\u7f6e\uff0c\u6216\u4f7f\u7528\u66f4\u9ad8\u8d28\u91cf\u7684\u97f3\u9891\u6587\u4ef6\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n<h3>\u793a\u4f8b<\/h3>\n<ol>\n<li><strong>\u9010\u5b57\u8f6c\u5f55\u793a\u4f8b<\/strong>\uff1a\n<pre><code>\u539f\u97f3\u9891\uff1a\u55ef\uff0c\u6211\u89c9\u5f97\u8fd9\u4e2a\u9879\u76ee\u975e\u5e38\u6709\u8da3\u3002\r\n\u8f6c\u5f55\u7ed3\u679c\uff1a\u55ef\uff0c\u6211\u89c9\u5f97\u8fd9\u4e2a\u9879\u76ee\u975e\u5e38\u6709\u8da3\u3002\r\n\u65f6\u95f4\u6233\uff1a[0:00:01] \u55ef\uff0c[0:00:02] \u6211\uff0c[0:00:03] \u89c9\u5f97\uff0c[0:00:04] \u8fd9\u4e2a\uff0c[0:00:05] \u9879\u76ee\uff0c[0:00:06] \u975e\u5e38\uff0c[0:00:07] \u6709\u8da3\u3002\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u586b\u5145\u8bcd\u68c0\u6d4b\u793a\u4f8b<\/strong>\uff1a\n<pre><code>\u539f\u97f3\u9891\uff1a\u55ef\uff0c\u6211\u89c9\u5f97\u8fd9\u4e2a\u9879\u76ee\u975e\u5e38\u6709\u8da3\u3002\r\n\u8f6c\u5f55\u7ed3\u679c\uff1a\u55ef\uff0c\u6211\u89c9\u5f97\u8fd9\u4e2a\u9879\u76ee\u975e\u5e38\u6709\u8da3\u3002\r\n\u586b\u5145\u8bcd\uff1a[0:00:01] \u55ef\r\n<\/code><\/pre>\n<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>CrisperWhisper \u662f\u4e00\u4e2a\u57fa\u4e8e OpenAI Whisper \u7684\u9ad8\u7ea7\u8bed\u97f3\u8bc6\u522b\u5de5\u5177\uff0c\u4e13\u6ce8\u4e8e\u5feb\u901f\u3001\u7cbe\u786e\u548c\u9010\u5b57\u7684\u8bed\u97f3\u8f6c\u5f55\u3002\u5b83\u80fd\u591f\u63d0\u4f9b\u7cbe\u786e\u7684\u5355\u8bcd\u7ea7\u65f6\u95f4\u6233\uff0c\u751a\u81f3\u5728\u8bed\u97f3\u586b\u5145\u548c\u505c\u987f\u7684\u60c5\u51b5\u4e0b\u4e5f\u80fd\u51c6\u786e\u8bb0\u5f55\u3002CrisperWhisper \u901a\u8fc7\u8c03\u6574\u6807\u8bb0&#8230;<\/p>\n","protected":false},"author":1,"featured_media":60982,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[230,216],"class_list":["post-5900","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tool","tag-aikaiyuanxiangmu","tag-aiyuyinzhuanwenben"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts\/5900","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/comments?post=5900"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts\/5900\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/media\/60982"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/media?parent=5900"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/categories?post=5900"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/tags?post=5900"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}