{"id":1224,"date":"2026-09-07T20:56:25","date_gmt":"2026-09-07T12:56:25","guid":{"rendered":"http:\/\/www.tantan.ink\/index.php\/2026\/09\/07\/vibevoice-microsoft-tts-codes-removed-asr-kept-open\/"},"modified":"2026-09-07T20:56:25","modified_gmt":"2026-09-07T12:56:25","slug":"vibevoice-microsoft-tts-codes-removed-asr-kept-open","status":"publish","type":"post","link":"https:\/\/www.tantan.ink\/index.php\/2026\/09\/07\/vibevoice-microsoft-tts-codes-removed-asr-kept-open\/","title":{"rendered":"\u5fae\u8f6f\u8bed\u97f3\u5f00\u6e90\uff1aTTS \u64a4\u56de\u4ee3\u7801\uff0cASR \u4e00\u8def\u5f00\u5230\u5e95"},"content":{"rendered":"<p>\u5fae\u8f6f 2025 \u5e74 8 \u6708\u5e95\u5f00\u6e90\u7684 VibeVoice\uff0c9 \u6708 3 \u65e5\u8fd8\u5728\u53d1\u65b0\u7248\u672c\uff0c\u4f46\u8fd9\u662f\u4e24\u4e2a\u6545\u4e8b\u3002TTS \u53d1\u5e03 11 \u5929\u5c31\u88ab\u5fae\u8f6f\u81ea\u5df1\u64a4\u56de\u4ee3\u7801\uff0c\u4ed3\u5e93\u91cc\u7684 Quick Try \u6807\u7b7e\u76f4\u63a5\u6807 Disabled\uff1b\u540c\u4e00\u4ed3\u5e93\u7684 ASR \u7ebf\u5374\u4e00\u8def\u5f00\u5230 9 \u6708\u521d\uff0cCPU \u91cf\u5316\u7248\u3001Transformers \u96c6\u6210\u3001\u6d41\u5f0f\u8bc6\u522b\u4e00\u6c14\u5475\u6210\u3002\u540c\u4e00\u5c4b\u6a90\u4e0b\u4e24\u4e2a\u65b9\u5411\uff0c\u5f00\u6e90\u7684\u59ff\u52bf\u5dee\u592a\u591a\u3002<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.tantan.ink\/wp-content\/uploads\/2026\/09\/00-architecture.png\" alt=\"00-architecture.png\" \/><\/p>\n<p>\u8fd9\u5f20\u662f\u5b98\u65b9\u7ed9\u7684 VibeVoice \u67b6\u6784\u56fe\u3002\u4e00\u4e2a LLM\uff08Qwen2.5 \u5bb6\u65cf\uff09\u505a\u8bed\u4e49\u7406\u89e3\uff0c\u52a0\u4e00\u4e2a\u58f0\u5b66\u548c\u8bed\u4e49 tokenizer \u628a\u97f3\u9891\u538b\u5230 7.5Hz \u6781\u4f4e\u5e27\u7387\uff0c\u518d\u7528\u4e00\u4e2a diffusion head \u51fa\u58f0\u5b66\u7ec6\u8282\u3002\u6574\u6761\u7ebf\u4ece\u9aa8\u67b6\u5230\u6548\u679c\u90fd\u6bd4\u8f83\u5e72\u51c0\uff0c\u662f\u6709\u70b9\u610f\u601d\u7684\u5de5\u4f5c\u3002<\/p>\n<p>\u8bf4\u8bf4 TTS \u8fd9\u6761\u7ebf\u5e72\u8fc7\u4ec0\u4e48\u30022025-08-25 \u5fae\u8f6f\u5f00\u6e90 <strong>VibeVoice-TTS 1.5B<\/strong>\uff1a\u5355\u904d\u5408\u6210\u6700\u957f 90 \u5206\u949f\u97f3\u9891\u3001\u652f\u6301\u6700\u591a 4 \u4e2a\u8bf4\u8bdd\u4eba\uff0c\u8bba\u6587\u88ab ICLR 2026 \u6536\u4e3a Oral\u3002\u6a21\u578b\u57fa\u4e8e Qwen2.5-1.5B\uff0c\u6302\u4e00\u4e2a 7.5Hz \u7684\u8fde\u7eed\u8bed\u97f3 tokenizer\u3002\u4ece\u6307\u6807\u4e0a\u770b\u5f88\u786c\u3002<\/p>\n<p>\u7136\u540e 11 \u5929\u540e\uff0c2025-09-05\uff0c\u4ed3\u5e93 News \u533a\u66f4\u65b0\u4e86\u4e00\u6761\uff1a<\/p>\n<blockquote>\n<p>After release, we discovered instances where the tool was used in ways inconsistent with the stated intent. Since responsible use of AI is one of Microsoft&#8217;s guiding principles, <strong>we have removed the VibeVoice-TTS code from this repository<\/strong>.<\/p>\n<\/blockquote>\n<p>\u8bad\u7ec3\u4ee3\u7801\u3001\u63a8\u7406\u4ee3\u7801\u5168\u64a4\u3002HF \u4e0a\u7684 1.5B \u6743\u91cd\u8fd8\u5728\uff0c\u4f46\u5b98\u65b9\u4e0d\u518d\u7ed9\u63a8\u7406\u8def\u5f84\uff0cREADME \u8868\u683c\u91cc TTS-1.5B \u7684 Quick Try \u5217\u5199 &#8220;Disabled&#8221;\u3002\u6574\u4e2a TTS \u4ed3\u5e93\u53d8\u6210\u4e00\u4e2a\u53d1\u4e0d\u51fa\u53bb\u7684\u793c\u7269\u3002<\/p>\n<p>\u4e0b\u9762\u662f ASR \u8fd9\u6761\u7ebf\u7684\u65f6\u95f4\u7ebf\uff1a<\/p>\n<ul>\n<li>2026-01-21\uff1a\u5f00\u6e90 <strong>VibeVoice-ASR<\/strong>\uff0c60 \u5206\u949f\u5355\u904d\u3001\u8bf4\u8bdd\u4eba\u8bc6\u522b + \u65f6\u95f4\u6233 + \u81ea\u5b9a\u4e49\u70ed\u8bcd\uff0c\u539f\u751f\u652f\u6301 50+ \u8bed\u8a00\uff0c\u5fae\u8c03\u4ee3\u7801\u4e00\u8d77\u653e\u51fa\u3002<\/li>\n<li>2026-03-06\uff1a\u8fdb HuggingFace Transformers\uff0c\u5e38\u89c4 pipeline \u80fd\u76f4\u63a5\u7528\u3002<\/li>\n<li>2026-07-23\uff1a<strong>VibeVoice-ASR-BitNet<\/strong>\uff0c\u8fb9\u7f18 CPU \u63a8\u7406\u5f15\u64ce\u3002\u5f02\u6784\u91cf\u5316\u628a 4.62GB \u6a21\u578b\u538b\u5230 1.58GB\uff0c3 \u7ebf\u7a0b CPU \u5b9e\u65f6 RTF&lt;1\uff0c\u4e0d\u7528 GPU\u3002<\/li>\n<li>2026-09-03\uff1a<strong>VibeVoice-ASR-Streaming<\/strong>\uff0c\u6d41\u5f0f ASR\uff0c\u97f3\u9891\u8fd8\u5728\u9001\u8fdb\u6765\u7684\u540c\u65f6\u6309\u6bb5\u8f93\u51fa\u6587\u5b57\uff0c\u81ea\u5b9a\u4e49\u70ed\u8bcd\uff0c10 \u79cd\u8bed\u8a00\u3002<\/li>\n<\/ul>\n<p>\u4e00\u6574\u5e74\u91cc ASR \u66f4\u4e86\u56db\u6b21\uff0c\u6bcf\u6b21\u90fd\u7ed9\u771f\u4ee3\u7801\uff0c\u8fde CPU \u5b9e\u65f6\u8dd1\u8fd9\u79cd\u8fb9\u7f18\u573a\u666f\u90fd\u4e13\u95e8\u505a\u4e86\u4e00\u4e2a\u53d8\u4f53\u3002\u4e2d\u95f4\u8fd8\u987a\u624b\u628a vLLM \u63a8\u7406\u3001Transformers \u96c6\u6210\u3001\u6d41\u5f0f\u8bc6\u522b\u5168\u505a\u9f50\u4e86\u3002<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.tantan.ink\/wp-content\/uploads\/2026\/09\/01-github.png\" alt=\"01-github.png\" \/><\/p>\n<p>\u8fd9\u5f20\u662f VibeVoice GitHub \u4ed3\u5e93\u9875\u7684\u73b0\u72b6\u300253.9k stars\u30016.1k forks\u3001MIT license \u4e00\u5207\u6b63\u5e38\uff0c\u4f46 TTS \u90a3\u4e2a 1.5B \u6a21\u578b\u7684 Quick Try \u5217\u5c31\u662f\u4e00\u7247\u7070\u5e95\u3002<\/p>\n<p>\u540c\u4e00\u516c\u53f8\u3001\u540c\u4e00\u4e2a\u4ed3\u5e93\u3001\u4e24\u79cd\u8282\u594f\uff0c\u8fd9\u79cd\u5206\u5316\u672c\u8eab\u5c31\u662f\u4e2a\u4fe1\u53f7\u3002TTS \u662f\u6df1\u5ea6\u4f2a\u9020\u3001\u8bc8\u9a97\u3001\u5192\u5145\u7535\u8bdd\u7684\u91cd\u707e\u533a\uff0cASR \u4e3b\u8981\u5c31\u662f\u4f1a\u8bae\u8f6c\u5199\u3001\u5b57\u5e55\u3001\u68c0\u7d22\uff0c\u600e\u4e48\u770b\u90fd\u662f TTS \u98ce\u9669\u66f4\u9ad8\u3002\u5fae\u8f6f\u9009\u4e86&#8221;\u64a4 TTS\u3001\u7559 ASR&#8221;\u7684\u8def\u7ebf\uff0c\u770b\u5f97\u51fa\u6765\u98ce\u9669\u504f\u597d\u5728\u54ea\u513f\u3002<\/p>\n<p>\u4f46\u64a4\u4ee3\u7801\u8fd9\u4e8b\uff0c\u600e\u4e48\u8bf4\u5462\uff0c\u66f4\u50cf\u662f\u5835\u9519\u4e86\u51fa\u53e3\u3002<strong>HF \u4e0a\u7684 1.5B \u6743\u91cd\u5168\u7f51\u53ef\u4e0b<\/strong>\uff0c\u771f\u60f3\u505a\u574f\u4e8b\u7684\u4eba\u4e0d\u7f3a\u63a8\u7406\u8def\u5f84\u3002\u64a4\u4ed3\u5e93\u4ee3\u7801\u53ea\u8ba9\u6b63\u7ecf\u7684\u672c\u5730\u73a9\u5bb6\u3001\u7814\u7a76\u8005\u3001\u5e94\u7528\u5f00\u53d1\u8005\u66f4\u96be\u7528\u2014\u2014\u6211\u4eec\u65e2\u6ca1\u6cd5\u672c\u5730\u4f53\u9a8c 90 \u5206\u949f\u957f TTS\uff0c\u4e5f\u6ca1\u6cd5\u62ff\u4ee3\u7801\u53bb\u7814\u7a76 tokenizer \u7684\u7ec6\u8282\uff0c\u6700\u591a\u53ea\u80fd\u4e0b\u4e2a\u6743\u91cd\u5e72\u77aa\u773c\u3002<\/p>\n<p>\u6211\u6bd4\u8f83\u671f\u5f85\u7684\u662f\u53e6\u4e00\u5934\u3002<strong>VibeVoice-ASR-BitNet \u7684 1.58GB CPU \u5b9e\u65f6\u7248\u672c<\/strong>\uff0c\u5bf9\u672c\u5730\u515a\u592a\u53cb\u597d\u4e86\u3002\u7406\u8bba\u4e0a\u585e\u8fdb NAS\u3001\u585e\u8fdb\u5c0f\u4e3b\u673a\u90fd\u4e0d\u9700\u8981\u663e\u5361\uff0c60 \u5206\u949f\u5355\u904d\u5e26\u8bf4\u8bdd\u4eba\u5206\u79bb\u7684\u8f6c\u5199\u5728 CPU \u4e0a\u5b9e\u65f6\u8dd1\u51fa\u6765\uff0c\u8fd9\u4e2a\u4e8b\u513f\u5728\u4ee5\u524d\u8981\u4e48\u9760\u4e91 API\u3001\u8981\u4e48\u672c\u5730\u8dd1 Whisper \u7136\u540e\u81ea\u5df1\u5806 diarization\uff0c\u73b0\u5728\u80fd\u76f4\u63a5\u6362\u4e0a\u53bb\u3002<\/p>\n<p>\u66f4\u8f7b\u7684 Realtime-0.5B \u6d41\u5f0f TTS\uff08~300ms \u9996\u97f3\u5ef6\u8fdf\u3001\u6700\u957f 10 \u5206\u949f\uff09\u5fae\u8f6f\u5012\u662f\u7559\u4e86\u4e0b\u6765\u3002\u731c\u6d4b\u662f\u56e0\u4e3a 0.5B \u4f53\u91cf\u4e0b\u6ee5\u7528\u95e8\u69db\u66f4\u9ad8\u3001\u5bf9\u751f\u6210\u5185\u5bb9\u4e5f\u52a0\u4e86\u6c34\u5370\u4e4b\u7c7b\u7684\u63aa\u65bd\uff0c\u76f8\u5bf9\u5b89\u5168\u3002\u8fd9\u4e00\u652f\u53ef\u4ee5\u88c5\u4e0a\u73a9\uff0c\u4f46\u662f 90 \u5206\u949f\u957f\u6587\u90a3\u6761\u7ebf\u57fa\u672c\u5e9f\u4e86\u3002<\/p>\n<p>\u5b98\u65b9\u6548\u679c\u56fe\u91cc\u7ed9\u7684\u5bf9\u6bd4\u662f\u8fd9\u6837\uff1a<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.tantan.ink\/wp-content\/uploads\/2026\/09\/02-tts-results.png\" alt=\"02-tts-results.png\" \/><\/p>\n<p>90 \u5206\u949f\u957f\u6587\u30014 \u8bf4\u8bdd\u4eba\u3001\u81ea\u7136\u7684\u8f6e\u6362\u548c\u60c5\u7eea\u8fde\u8d2f\uff0c\u4ece\u611f\u5b98\u4e0a\u770b\u6bd4\u4e00\u822c\u7684\u5f00\u6e90 TTS \u5f3a\u4e00\u622a\u3002\u95ee\u9898\u662f\uff0c\u8fd9\u4e2a\u56fe\u5bf9\u5e94\u7684 1.5B \u6a21\u578b\u73b0\u5728\u6ca1\u4ee3\u7801\u53ef\u7528\uff0c\u793e\u533a\u6709\u7b2c\u4e09\u65b9\u9006\u5411\u6216\u91cd\u5efa\u65b9\u6848\u524d\u90fd\u8dd1\u4e0d\u8d77\u6765\u3002<\/p>\n<p>\u6211\u7684\u6001\u5ea6\u5f88\u7b80\u5355\uff1a\u4e0e\u5176\u64a4\u56de\uff0c\u4e0d\u5982\u5728\u63a8\u7406\u5c42\u52a0\u6c34\u5370\u548c\u514d\u8d23\u58f0\u660e\uff0c\u50cf Realtime-0.5B \u90a3\u6837\u4fdd\u4f4f\u4e00\u4e2a\u53d7\u9650\u53ef\u7528\u7684\u7248\u672c\u3002&#8221;\u53ea\u5f00\u6e90\u6743\u91cd\u3001\u4e0d\u5f00\u6e90\u4ee3\u7801&#8221;\u8fd9\u79cd\u534a\u622a\u5f00\u6e90\uff0c\u5bf9\u751f\u6001\u7684\u4f24\u5bb3\u6bd4\u4e0d\u5f00\u6e90\u8fd8\u5927\u2014\u2014\u6743\u91cd\u5168\u5728 HF \u4e0a\u6302\u7740\uff0c\u672c\u5730\u771f\u8981\u7528\u8fd8\u662f\u5f97\u81ea\u5df1\u60f3\u529e\u6cd5\uff0c\u628a&#8221;\u5f00\u6e90&#8221;\u5f53\u4e86\u4e00\u56de\u8425\u9500\u3002<\/p>\n<p>ASR \u7ebf\u5012\u662f\u8d8a\u505a\u8d8a\u624e\u5b9e\uff0c\u503c\u5f97\u8ddf\u8e2a\u3002\u5982\u679c\u672c\u5730\u8dd1\u4f1a\u8bae\u8f6c\u5199\u5acc Whisper \u7684\u8bf4\u8bdd\u4eba\u5206\u79bb\u592a\u5f31\uff0c\u7b49 BitNet \u7248\u76f4\u63a5\u6362\u4e0a\u53bb\u57fa\u672c\u6ca1\u6210\u672c\u3002<\/p>\n<hr>\n<p>\u8d44\u6599\u6765\u6e90\uff1a<\/p>\n<ul>\n<li><a href=\"https:\/\/github.com\/microsoft\/VibeVoice\">microsoft\/VibeVoice<\/a>\uff08MIT\uff0c53.9k stars\uff0c9\/3 \u4ecd\u5728\u66f4\u65b0\uff09<\/li>\n<li><a href=\"https:\/\/arxiv.org\/abs\/2508.19205\">VibeVoice-TTS \u6280\u672f\u62a5\u544a<\/a><\/li>\n<li><a href=\"https:\/\/arxiv.org\/abs\/2601.18184\">VibeVoice-ASR \u6280\u672f\u62a5\u544a<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/microsoft\/VibeVoice-ASR-Streaming-7B\">ASR-Streaming-7B \u6743\u91cd<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/microsoft\/VibeASR.cpp\">VibeVoice-ASR-BitNet\uff08CPU \u63a8\u7406\u5f15\u64ce\uff09<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>\u5fae\u8f6f 2025 \u5e74 8 \u6708\u5e95\u5f00\u6e90 VibeVoice\uff0c9 \u6708 3 \u65e5\u8fd8\u5728\u53d1\u65b0\u7248\u672c\uff0c\u4f46\u8fd9\u662f\u4e24\u4e2a\u6545\u4e8b\u2014\u2014TTS \u53d1\u5e03 11 \u5929\u64a4\u56de\u4ee3\u7801\uff0cASR \u4e00\u5e74\u56db\u6b21\u66f4\u65b0\u3002<\/p>\n","protected":false},"author":2,"featured_media":1221,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[],"class_list":["post-1224","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/posts\/1224","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/comments?post=1224"}],"version-history":[{"count":0,"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/posts\/1224\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/media\/1221"}],"wp:attachment":[{"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/media?parent=1224"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/categories?post=1224"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.tantan.ink\/index.php\/wp-json\/wp\/v2\/tags?post=1224"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}