{"id":32264,"date":"2025-07-09T09:04:32","date_gmt":"2025-07-09T01:04:32","guid":{"rendered":"https:\/\/www.kdjingpai.com\/?p=32264"},"modified":"2025-07-21T13:44:57","modified_gmt":"2025-07-21T05:44:57","slug":"cosyvoice-2","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/pt\/cosyvoice-2\/","title":{"rendered":"CosyVoice\uff1a\u963f\u91cc\u5f00\u6e90\u7684\u591a\u8bed\u8a00\u514b\u9686\u4e0e\u751f\u6210\u5de5\u5177"},"content":{"rendered":"<p><a href=\"https:\/\/www.kdjingpai.com\/de\/cosyvoice\/\">CosyVoice<\/a> \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u591a\u8bed\u8a00\u8bed\u97f3\u751f\u6210\u6a21\u578b\uff0c\u4e13\u6ce8\u4e8e\u9ad8\u8d28\u91cf\u7684\u6587\u672c\u8f6c\u8bed\u97f3\uff08TTS\uff09\u6280\u672f\u3002\u5b83\u652f\u6301\u591a\u79cd\u8bed\u8a00\u7684\u8bed\u97f3\u5408\u6210\uff0c\u63d0\u4f9b\u96f6\u6837\u672c\u8bed\u97f3\u751f\u6210\u3001\u8de8\u8bed\u8a00\u8bed\u97f3\u514b\u9686\u548c\u7ec6\u7c92\u5ea6\u60c5\u611f\u63a7\u5236\u7b49\u529f\u80fd\u3002Cos\u2013<\/p>\n<p>yVoice 2.0 \u76f8\u6bd4\u4e0a\u4e00\u7248\u672c\uff0c\u663e\u8457\u964d\u4f4e\u4e8630%\u523050%\u7684\u53d1\u97f3\u8bef\u5dee\uff0c\u97f3\u8d28\u548c\u97f5\u5f8b\u81ea\u7136\u5ea6\u5927\u5e45\u63d0\u5347\uff0cMOS\u8bc4\u5206\u4ece5.4\u63d0\u9ad8\u52305.53\u3002\u5b83\u901a\u8fc7\u7b80\u5316\u7684\u6a21\u578b\u67b6\u6784\u548c\u4f18\u5316\u7b97\u6cd5\uff0c\u5b9e\u73b0\u4f4e\u5ef6\u8fdf\u7684\u6d41\u5f0f\u548c\u975e\u6d41\u5f0f\u8bed\u97f3\u5408\u6210\uff0c\u9002\u7528\u4e8e\u5b9e\u65f6\u4ea4\u4e92\u573a\u666f\u3002\u9879\u76ee\u63d0\u4f9b\u5b8c\u6574\u7684\u63a8\u7406\u3001\u8bad\u7ec3\u548c\u90e8\u7f72\u652f\u6301\uff0c\u5f00\u53d1\u8005\u53ef\u8f7b\u677e\u4e0a\u624b\uff0c\u9002\u5408\u8bed\u97f3\u52a9\u624b\u3001\u914d\u97f3\u548c\u591a\u8bed\u8a00\u5185\u5bb9\u521b\u4f5c\u7b49\u5e94\u7528\u3002<\/p>\n<p>\u5f00\u653e\u670d\u52a1\uff1a<a href=\"https:\/\/www.kdjingpai.com\/cosyvoice\/\">CosyVoice\uff1a\u963f\u91cc\u63a8\u51fa\u76843\u79d2\u6025\u901f\u8bed\u97f3\u514b\u9686\u5f00\u6e90\u9879\u76ee\uff0c\u652f\u6301\u60c5\u611f\u63a7\u5236\u6807\u7b7e<\/a><\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li>\u96f6\u6837\u672c\u8bed\u97f3\u751f\u6210\uff1a\u57fa\u4e8e\u77ed\u97f3\u9891\u6837\u672c\u751f\u6210\u4e0e\u76ee\u6807\u58f0\u97f3\u76f8\u4f3c\u7684\u8bed\u97f3\uff0c\u65e0\u9700\u989d\u5916\u8bad\u7ec3\u3002<\/li>\n<li>\u8de8\u8bed\u8a00\u8bed\u97f3\u5408\u6210\uff1a\u652f\u6301\u591a\u8bed\u8a00\u8bed\u97f3\u751f\u6210\uff0c\u4fdd\u6301\u97f3\u8272\u4e00\u81f4\u6027\uff0c\u9002\u7528\u4e8e\u5168\u7403\u5185\u5bb9\u521b\u4f5c\u3002<\/li>\n<li>\u7ec6\u7c92\u5ea6\u60c5\u611f\u63a7\u5236\uff1a\u652f\u6301\u6dfb\u52a0\u7b11\u58f0\u3001\u505c\u987f\u7b49\u60c5\u611f\u8868\u8fbe\uff0c\u751f\u6210\u66f4\u81ea\u7136\u7684\u8bed\u97f3\u3002<\/li>\n<li>\u65b9\u8a00\u548c\u53e3\u97f3\u8c03\u6574\uff1a\u53ef\u751f\u6210\u7279\u5b9a\u65b9\u8a00\uff08\u5982\u56db\u5ddd\u8bdd\uff09\u6216\u53e3\u97f3\u7684\u8bed\u97f3\uff0c\u589e\u5f3a\u672c\u5730\u5316\u4f53\u9a8c\u3002<\/li>\n<li>\u6d41\u5f0f\u8bed\u97f3\u5408\u6210\uff1a\u652f\u6301\u4f4e\u5ef6\u8fdf\u6d41\u5f0f\u8f93\u51fa\uff0c\u9996\u5305\u5ef6\u8fdf\u4f4e\u81f3150ms\uff0c\u9002\u5408\u5b9e\u65f6\u5e94\u7528\u3002<\/li>\n<li>\u6a21\u578b\u8bad\u7ec3\u652f\u6301\uff1a\u63d0\u4f9b\u9884\u8bad\u7ec3\u6a21\u578b\u548c\u4ece\u5934\u8bad\u7ec3\u7684\u5b8c\u6574\u6d41\u7a0b\uff0c\u6ee1\u8db3\u5f00\u53d1\u8005\u9700\u6c42\u3002<\/li>\n<li>\u9ad8\u97f3\u8d28\u8f93\u51fa\uff1a\u4f18\u5316\u97f3\u8d28\u548c\u97f5\u5f8b\uff0cMOS\u8bc4\u5206\u8fbe\u52305.53\uff0c\u63a5\u8fd1\u5546\u4e1a\u5316\u6c34\u5e73\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>\u5b89\u88c5\u6d41\u7a0b<\/h3>\n<p>\u8981\u4f7f\u7528 CosyVoice\uff0c\u9996\u5148\u9700\u8981\u5b89\u88c5\u5fc5\u8981\u7684\u73af\u5883\u548c\u4f9d\u8d56\u3002\u4ee5\u4e0b\u662f\u8be6\u7ec6\u7684\u5b89\u88c5\u6b65\u9aa4\uff0c\u786e\u4fdd\u7528\u6237\u80fd\u591f\u5feb\u901f\u914d\u7f6e\u5e76\u8fd0\u884c\u9879\u76ee\u3002<\/p>\n<ol>\n<li><strong>\u5b89\u88c5 Conda<\/strong><br \/>\nCosyVoice \u63a8\u8350\u4f7f\u7528 Conda \u7ba1\u7406\u73af\u5883\u3002\u8bf7\u8bbf\u95ee\u00a0<code>https:\/\/docs.conda.io\/en\/latest\/miniconda.html<\/code>\u00a0\u4e0b\u8f7d\u5e76\u5b89\u88c5 Miniconda\u3002\u7136\u540e\u521b\u5efa\u5e76\u6fc0\u6d3b\u865a\u62df\u73af\u5883\uff1a<\/p>\n<pre><code>conda create -n cosyvoice python=3.10\r\nconda activate cosyvoice\r\n<\/code><\/pre>\n<\/li>\n<\/ol>\n<pre><code>\r\n<\/code><\/pre>\n<ol start=\"2\">\n<li><strong>\u5b89\u88c5\u4f9d\u8d56<\/strong><br \/>\n\u4f7f\u7528 pip \u5b89\u88c5\u9879\u76ee\u6240\u9700\u7684 Python \u5305\u3002\u7531\u4e8e\u7f51\u7edc\u539f\u56e0\uff0c\u5efa\u8bae\u4f7f\u7528\u56fd\u5185\u955c\u50cf\u6e90\uff08\u5982\u963f\u91cc\u4e91\uff09\uff1a<\/p>\n<pre><code>pip install -r requirements.txt -i https:\/\/mirrors.aliyun.com\/pypi\/simple\/ --trusted-host=mirrors.aliyun.com\r\n<\/code><\/pre>\n<p>\u5b89\u88c5\u00a0<code>pynini<\/code>\uff08\u7528\u4e8e\u6587\u672c\u5904\u7406\uff09\uff1a<\/p>\n<pre><code>conda install -y -c conda-forge pynini==2.1.5\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u89e3\u51b3 sox \u517c\u5bb9\u6027\u95ee\u9898<\/strong><br \/>\n\u5982\u679c\u9047\u5230 sox \u76f8\u5173\u95ee\u9898\uff0c\u6839\u636e\u64cd\u4f5c\u7cfb\u7edf\u5b89\u88c5 sox\uff1a<\/p>\n<ul>\n<li><strong>Ubuntu<\/strong>\uff1a\n<pre><code>sudo apt-get install sox libsox-dev\r\n<\/code><\/pre>\n<\/li>\n<li><strong>CentOS<\/strong>\uff1a\n<pre><code>sudo yum install sox sox-devel\r\n<\/code><\/pre>\n<\/li>\n<\/ul>\n<\/li>\n<li><strong>\u4e0b\u8f7d\u9884\u8bad\u7ec3\u6a21\u578b<\/strong><br \/>\nCosyVoice \u63d0\u4f9b\u591a\u4e2a\u9884\u8bad\u7ec3\u6a21\u578b\uff08\u5982\u00a0<code>CosyVoice2-0.5B<\/code>\u3001<code>CosyVoice-300M<\/code>\u3001<code>CosyVoice-300M-SFT<\/code>\u00a0\u548c\u00a0<code>CosyVoice-300M-Instruct<\/code>\uff09\u3002\u7528\u6237\u53ef\u4ece GitHub \u4ed3\u5e93\u4e0b\u8f7d\u8fd9\u4e9b\u6a21\u578b\u5e76\u653e\u7f6e\u5728\u00a0<code>pretrained_models<\/code>\u00a0\u76ee\u5f55\u4e0b\u3002\u53e6\u9700\u4e0b\u8f7d\u00a0<code>CosyVoice-ttsfrd<\/code>\u00a0\u8d44\u6e90\u4ee5\u652f\u6301\u6587\u672c\u524d\u7aef\u5904\u7406\u3002<\/li>\n<li><strong>\u514b\u9686\u9879\u76ee\u4ee3\u7801<\/strong><br \/>\n\u4f7f\u7528 Git \u514b\u9686 CosyVoice \u4ed3\u5e93\uff0c\u5e76\u786e\u4fdd\u5b50\u6a21\u5757\u6b63\u786e\u52a0\u8f7d\uff1a<\/p>\n<pre><code>git clone --recursive https:\/\/github.com\/FunAudioLLM\/CosyVoice.git\r\ncd CosyVoice\r\ngit submodule update --init --recursive\r\n<\/code><\/pre>\n<\/li>\n<\/ol>\n<h3>\u4f7f\u7528\u65b9\u6cd5<\/h3>\n<p>CosyVoice \u63d0\u4f9b\u591a\u79cd\u529f\u80fd\u6a21\u5757\uff0c\u5305\u62ec\u96f6\u6837\u672c\u8bed\u97f3\u751f\u6210\u3001\u8de8\u8bed\u8a00\u5408\u6210\u548c\u6307\u4ee4\u5f0f\u8bed\u97f3\u751f\u6210\u3002\u4ee5\u4e0b\u662f\u5177\u4f53\u64cd\u4f5c\u6d41\u7a0b\u3002<\/p>\n<h4>\u96f6\u6837\u672c\u8bed\u97f3\u751f\u6210<\/h4>\n<p>\u96f6\u6837\u672c\u8bed\u97f3\u751f\u6210\u5141\u8bb8\u7528\u6237\u57fa\u4e8e\u77ed\u97f3\u9891\u6837\u672c\u751f\u6210\u76ee\u6807\u97f3\u8272\u7684\u8bed\u97f3\u3002\u4f8b\u5982\uff0c\u751f\u6210\u4e00\u6bb5\u5e26\u6709\u7279\u5b9a\u97f3\u8272\u7684\u8bed\u97f3\uff1a<\/p>\n<pre><code>from cosyvoice import CosyVoice2\r\nimport torchaudio\r\ncosyvoice = CosyVoice2('pretrained_models\/CosyVoice2-0.5B', load_jit=False, load_trt=False, fp16=False)\r\nprompt_speech_16k = torchaudio.load('.\/asset\/zero_shot_prompt.wav')[0]\r\nfor i, j in enumerate(cosyvoice.inference_zero_shot(\r\n'\u6536\u5230\u597d\u53cb\u4ece\u8fdc\u65b9\u5bc4\u6765\u7684\u751f\u65e5\u793c\u7269\uff0c\u90a3\u4efd\u610f\u5916\u7684\u60ca\u559c\u4e0e\u6df1\u6df1\u7684\u795d\u798f\u8ba9\u6211\u5fc3\u4e2d\u5145\u6ee1\u4e86\u751c\u871c\u7684\u5feb\u4e50\uff0c\u7b11\u5bb9\u5982\u82b1\u513f\u822c\u7efd\u653e\u3002',\r\n'\u5e0c\u671b\u4f60\u4ee5\u540e\u80fd\u591f\u505a\u7684\u6bd4\u6211\u8fd8\u597d\u5466\u3002',\r\nprompt_speech_16k,\r\nstream=False)):\r\ntorchaudio.save(f'zero_shot_{i}.wav', j['tts_speech'], cosyvoice.sample_rate)\r\n<\/code><\/pre>\n<ul>\n<li><strong>\u64cd\u4f5c\u6b65\u9aa4<\/strong>\uff1a\n<ol>\n<li>\u51c6\u5907\u4e00\u6bb5 16kHz \u7684\u63d0\u793a\u97f3\u9891\uff08\u5982\u00a0<code>zero_shot_prompt.wav<\/code>\uff09\u3002<\/li>\n<li>\u8c03\u7528\u00a0<code>inference_zero_shot<\/code>\u00a0\u65b9\u6cd5\uff0c\u4f20\u5165\u76ee\u6807\u6587\u672c\u3001\u63d0\u793a\u6587\u672c\u548c\u97f3\u9891\u3002<\/li>\n<li>\u4fdd\u5b58\u751f\u6210\u7684\u8bed\u97f3\u6587\u4ef6\u3002<\/li>\n<\/ol>\n<\/li>\n<li><strong>\u6ce8\u610f<\/strong>\uff1a\u82e5\u9700\u590d\u73b0\u5b98\u7f51\u7ed3\u679c\uff0c\u8bf7\u8bbe\u7f6e\u00a0<code>text_frontend=False<\/code>\u3002<\/li>\n<\/ul>\n<h4>\u8de8\u8bed\u8a00\u8bed\u97f3\u5408\u6210<\/h4>\n<p>\u8de8\u8bed\u8a00\u5408\u6210\u652f\u6301\u751f\u6210\u591a\u8bed\u8a00\u8bed\u97f3\uff0c\u5e76\u4fdd\u6301\u97f3\u8272\u4e00\u81f4\u3002\u4f8b\u5982\uff0c\u751f\u6210\u5e26\u6709\u7b11\u58f0\u7684\u8bed\u97f3\uff1a<\/p>\n<pre><code>for i, j in enumerate(cosyvoice.inference_cross_lingual(\r\n'\u5728\u4ed6\u8bb2\u8ff0\u90a3\u4e2a\u8352\u8bde\u6545\u4e8b\u7684\u8fc7\u7a0b\u4e2d\uff0c\u4ed6\u7a81\u7136[laughter]\u505c\u4e0b\u6765\uff0c\u56e0\u4e3a\u4ed6\u81ea\u5df1\u4e5f\u88ab\u9017\u7b11\u4e86[laughter]\u3002',\r\nprompt_speech_16k,\r\nstream=False)):\r\ntorchaudio.save(f'fine_grained_control_{i}.wav', j['tts_speech'], cosyvoice.sample_rate)\r\n<\/code><\/pre>\n<ul>\n<li><strong>\u64cd\u4f5c\u6b65\u9aa4<\/strong>\uff1a\n<ol>\n<li>\u51c6\u5907\u63d0\u793a\u97f3\u9891\u3002<\/li>\n<li>\u5728\u6587\u672c\u4e2d\u6dfb\u52a0\u60c5\u611f\u6807\u7b7e\uff08\u5982\u00a0<code>[laughter]<\/code>\uff09\uff0c\u5177\u4f53\u652f\u6301\u7684\u6807\u7b7e\u89c1\u00a0<code>cosyvoice\/tokenizer\/tokenizer.py#L248<\/code>\u3002<\/li>\n<li>\u4fdd\u5b58\u751f\u6210\u7684\u8bed\u97f3\u6587\u4ef6\u3002<\/li>\n<\/ol>\n<\/li>\n<\/ul>\n<h4>\u6307\u4ee4\u5f0f\u8bed\u97f3\u751f\u6210<\/h4>\n<p>\u6307\u4ee4\u5f0f\u751f\u6210\u652f\u6301\u6307\u5b9a\u65b9\u8a00\u6216\u98ce\u683c\u3002\u4f8b\u5982\uff0c\u751f\u6210\u56db\u5ddd\u8bdd\u8bed\u97f3\uff1a<\/p>\n<pre><code>for i, j in enumerate(cosyvoice.inference_instruct2(\r\n'\u6536\u5230\u597d\u53cb\u4ece\u8fdc\u65b9\u5bc4\u6765\u7684\u751f\u65e5\u793c\u7269\uff0c\u90a3\u4efd\u610f\u5916\u7684\u60ca\u559c\u4e0e\u6df1\u6df1\u7684\u795d\u798f\u8ba9\u6211\u5fc3\u4e2d\u5145\u6ee1\u4e86\u751c\u871c\u7684\u5feb\u4e50\uff0c\u7b11\u5bb9\u5982\u82b1\u513f\u822c\u7efd\u653e\u3002',\r\n'\u7528\u56db\u5ddd\u8bdd\u8bf4\u8fd9\u53e5\u8bdd',\r\nprompt_speech_16k,\r\nstream=False)):\r\ntorchaudio.save(f'instruct_{i}.wav', j['tts_speech'], cosyvoice.sample_rate)\r\n<\/code><\/pre>\n<ul>\n<li><strong>\u64cd\u4f5c\u6b65\u9aa4<\/strong>\uff1a\n<ol>\n<li>\u63d0\u4f9b\u76ee\u6807\u6587\u672c\u548c\u6307\u4ee4\uff08\u5982\u201c\u7528\u56db\u5ddd\u8bdd\u8bf4\u201d\uff09\u3002<\/li>\n<li>\u4f7f\u7528\u63d0\u793a\u97f3\u9891\u751f\u6210\u8bed\u97f3\u3002<\/li>\n<li>\u4fdd\u5b58\u8f93\u51fa\u6587\u4ef6\u3002<\/li>\n<\/ol>\n<\/li>\n<\/ul>\n<h4>\u6d41\u5f0f\u5408\u6210<\/h4>\n<p>\u6d41\u5f0f\u5408\u6210\u9002\u5408\u5b9e\u65f6\u5e94\u7528\uff0c\u9996\u5305\u5ef6\u8fdf\u4f4e\u81f3150ms\u3002\u8bbe\u7f6e\u00a0<code>stream=True<\/code>\u00a0\u5373\u53ef\u542f\u7528\u6d41\u5f0f\u8f93\u51fa\uff0c\u5177\u4f53\u4ee3\u7801\u53c2\u8003\u4e0a\u8ff0\u793a\u4f8b\u3002<\/p>\n<h3>\u5176\u4ed6\u6ce8\u610f\u4e8b\u9879<\/h3>\n<ul>\n<li><strong>\u6a21\u578b\u9009\u62e9<\/strong>\uff1a\u63a8\u8350\u4f7f\u7528\u00a0<code>CosyVoice2-0.5B<\/code>\u00a0\u6a21\u578b\u4ee5\u83b7\u5f97\u6700\u4f73\u97f3\u8d28\u548c\u6027\u80fd\u3002<\/li>\n<li><strong>\u73af\u5883\u914d\u7f6e<\/strong>\uff1a\u786e\u4fdd Python \u7248\u672c\u4e3a 3.10\uff0c\u4f9d\u8d56\u5b89\u88c5\u65e0\u8bef\u3002<\/li>\n<li><strong>\u8c03\u8bd5<\/strong>\uff1a\u82e5\u514b\u9686\u5b50\u6a21\u5757\u5931\u8d25\uff0c\u591a\u6b21\u8fd0\u884c\u00a0<code>git submodule update --init --recursive<\/code>\u00a0\u76f4\u5230\u6210\u529f\u3002<\/li>\n<li><strong>\u6587\u6863\u53c2\u8003<\/strong>\uff1a\u66f4\u591a\u7ec6\u8282\u89c1\u00a0<code>cosyvoice\/tokenizer\/tokenizer.py<\/code>\u00a0\u548c\u5b98\u65b9\u6587\u6863\u00a0<code>https:\/\/funaudiollm.github.io\/cosyvoice2<\/code>\u3002<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>\u5e94\u7528\u573a\u666f<\/h2>\n<ol>\n<li><strong>\u8bed\u97f3\u52a9\u624b\u5f00\u53d1<\/strong><br \/>\nCosyVoice \u7684\u4f4e\u5ef6\u8fdf\u6d41\u5f0f\u5408\u6210\u548c\u591a\u8bed\u8a00\u652f\u6301\u975e\u5e38\u9002\u5408\u5f00\u53d1\u667a\u80fd\u8bed\u97f3\u52a9\u624b\u3002\u5f00\u53d1\u8005\u53ef\u5229\u7528\u96f6\u6837\u672c\u751f\u6210\u5feb\u901f\u5b9a\u5236\u4e2a\u6027\u5316\u97f3\u8272\uff0c\u751f\u6210\u81ea\u7136\u6d41\u7545\u7684\u8bed\u97f3\u56de\u590d\uff0c\u63d0\u5347\u7528\u6237\u4ea4\u4e92\u4f53\u9a8c\u3002<\/li>\n<li><strong>\u591a\u8bed\u8a00\u5185\u5bb9\u521b\u4f5c<\/strong><br \/>\n\u5185\u5bb9\u521b\u4f5c\u8005\u53ef\u4f7f\u7528\u8de8\u8bed\u8a00\u5408\u6210\u529f\u80fd\uff0c\u5feb\u901f\u751f\u6210\u591a\u8bed\u8a00\u914d\u97f3\u3002\u4f8b\u5982\uff0c\u4e3a\u89c6\u9891\u6216\u64ad\u5ba2\u751f\u6210\u82f1\u8bed\u3001\u6c49\u8bed\u6216\u5176\u4ed6\u8bed\u8a00\u7684\u8bed\u97f3\uff0c\u4fdd\u6301\u97f3\u8272\u4e00\u81f4\uff0c\u964d\u4f4e\u5236\u4f5c\u6210\u672c\u3002<\/li>\n<li><strong>\u6559\u80b2\u4e0e\u8bed\u8a00\u5b66\u4e60<\/strong><br \/>\nCosyVoice \u652f\u6301\u65b9\u8a00\u548c\u60c5\u611f\u63a7\u5236\uff0c\u53ef\u7528\u4e8e\u8bed\u8a00\u5b66\u4e60\u5e94\u7528\uff0c\u751f\u6210\u7279\u5b9a\u53e3\u97f3\u6216\u60c5\u611f\u7684\u8bed\u97f3\uff0c\u5e2e\u52a9\u5b66\u4e60\u8005\u7ec3\u4e60\u542c\u529b\u548c\u53d1\u97f3\u3002<\/li>\n<li><strong>\u5f71\u89c6\u4e0e\u6e38\u620f\u914d\u97f3<\/strong><br \/>\n\u5f71\u89c6\u548c\u6e38\u620f\u5f00\u53d1\u8005\u53ef\u5229\u7528\u7ec6\u7c92\u5ea6\u60c5\u611f\u63a7\u5236\u751f\u6210\u5e26\u6709\u7b11\u58f0\u3001\u505c\u987f\u7b49\u6548\u679c\u7684\u8bed\u97f3\uff0c\u589e\u5f3a\u89d2\u8272\u8868\u73b0\u529b\uff0c\u63d0\u5347\u4f5c\u54c1\u6c89\u6d78\u611f\u3002<\/li>\n<\/ol>\n<p>&nbsp;<\/p>\n<h2>QA<\/h2>\n<ol>\n<li><strong>CosyVoice \u652f\u6301\u54ea\u4e9b\u8bed\u8a00\uff1f<\/strong><br \/>\nCosyVoice \u652f\u6301\u591a\u79cd\u8bed\u8a00\u7684\u8bed\u97f3\u5408\u6210\uff0c\u5177\u4f53\u8bed\u8a00\u5217\u8868\u53ef\u5728\u5b98\u65b9\u6587\u6863\u6216\u00a0<code>cosyvoice\/tokenizer\/tokenizer.py<\/code>\u00a0\u4e2d\u67e5\u770b\uff0c\u8986\u76d6\u5e38\u89c1\u8bed\u8a00\u548c\u90e8\u5206\u65b9\u8a00\u3002<\/li>\n<li><strong>\u5982\u4f55\u964d\u4f4e\u8bed\u97f3\u751f\u6210\u7684\u5ef6\u8fdf\uff1f<\/strong><br \/>\n\u4f7f\u7528\u6d41\u5f0f\u5408\u6210\u6a21\u5f0f\uff08<code>stream=True<\/code>\uff09\u53ef\u5c06\u9996\u5305\u5ef6\u8fdf\u964d\u81f3150ms\u3002\u786e\u4fdd\u4f7f\u7528\u9ad8\u6027\u80fd\u786c\u4ef6\u548c\u4f18\u5316\u540e\u7684\u6a21\u578b\uff08\u5982\u00a0<code>CosyVoice2-0.5B<\/code>\uff09\u3002<\/li>\n<li><strong>\u662f\u5426\u9700\u8981\u8bad\u7ec3\u6a21\u578b\uff1f<\/strong><br \/>\n\u4e0d\u9700\u8981\u3002CosyVoice \u63d0\u4f9b\u9884\u8bad\u7ec3\u6a21\u578b\uff0c\u76f4\u63a5\u4e0b\u8f7d\u5373\u53ef\u4f7f\u7528\u3002\u82e5\u9700\u5b9a\u5236\uff0c\u53ef\u53c2\u8003\u5b98\u65b9\u6587\u6863\u4ece\u5934\u8bad\u7ec3\u3002<\/li>\n<li><strong>\u5982\u4f55\u6dfb\u52a0\u81ea\u5b9a\u4e49\u97f3\u8272\uff1f<\/strong><br \/>\n\u4f7f\u7528\u00a0<code>inference_zero_shot<\/code>\u00a0\u65b9\u6cd5\uff0c\u4f20\u5165\u76ee\u6807\u97f3\u9891\u6837\u672c\u548c\u63d0\u793a\u6587\u672c\u5373\u53ef\u751f\u6210\u81ea\u5b9a\u4e49\u97f3\u8272\u3002\u4fdd\u5b58\u97f3\u8272\u4fe1\u606f\u53ef\u8c03\u7528\u00a0<code>cosyvoice.add_zero_shot_spk<\/code>\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>CosyVoice \u662f\u4e00\u4e2a\u5f00\u6e90\u7684\u591a\u8bed\u8a00\u8bed\u97f3\u751f\u6210\u6a21\u578b\uff0c\u4e13\u6ce8\u4e8e\u9ad8\u8d28\u91cf\u7684\u6587\u672c\u8f6c\u8bed\u97f3\uff08TTS\uff09\u6280\u672f\u3002\u5b83\u652f\u6301\u591a\u79cd\u8bed\u8a00\u7684\u8bed\u97f3\u5408\u6210\uff0c\u63d0\u4f9b\u96f6\u6837\u672c\u8bed\u97f3\u751f\u6210\u3001\u8de8\u8bed\u8a00\u8bed\u97f3\u514b\u9686\u548c\u7ec6\u7c92\u5ea6\u60c5\u611f\u63a7\u5236\u7b49\u529f\u80fd\u3002Cos\u2013 yVoice 2.0 \u76f8\u6bd4\u4e0a\u4e00\u7248\u672c\uff0c\u663e\u8457\u964d\u4f4e\u4e8630%\u5230&#8230;<\/p>\n","protected":false},"author":1,"featured_media":32782,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20,392,400],"tags":[230,215,237],"class_list":["post-32264","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tool","category-models","category-speech-model","tag-aikaiyuanxiangmu","tag-aiwenbenzhuanyuyin","tag-aiyuyinkelong"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/posts\/32264","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/comments?post=32264"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/posts\/32264\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/media\/32782"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/media?parent=32264"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/categories?post=32264"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/pt\/wp-json\/wp\/v2\/tags?post=32264"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}