搭建fast-whisper 环境时报错 Unable to load any of {libcudnn_ops.so.9.1.0, libcudnn_ops.so.9.1, libcudnn_ops.so.9, libcudnn_ops.so}
<p>fast-whisper 官网地址：https://github.com/SYSTRAN/faster-whisper</p><p>搭建环境时，按照官方的创建环境要求执行的</p>
<ol>
<li>使用<code>conda create -n fast_whisper python=3.9</code>创建虚拟环境</li>
<li>执行<code>pip install faster-whisper</code> 安装库</li>
<li><code>pip install nvidia-cublas-cu12 nvidia-cudnn-cu12==8.*</code></li>
<li>降低numpy版本，fast-whisper只能基于numpy 1.* 版本</li>
</ol>
<p>关于linux上创建虚拟环境可以参考 https://zhuanlan.zhihu.com/p/440548295</p>
<p>在运行示例的时候报错如下：<br>
<img src="https://img2024.cnblogs.com/blog/3491310/202411/3491310-20241114133145514-1728194421.png" alt="image" loading="lazy"></p>
<p>代码如下：</p>
<pre><code class="language-python">from faster_whisper import WhisperModel
model_size = "large-v3"
# Run on GPU with FP16
model = WhisperModel(model_size, device="cuda", compute_type="float16")
# or run on GPU with INT8
# model = WhisperModel(model_size, device="cuda", compute_type="int8_float16")
# or run on CPU with INT8
# model = WhisperModel(model_size, device="cpu", compute_type="int8")
segments, info = model.transcribe("test_wavs/whisper_test.wav", beam_size=5)
print("Detected language '%s' with probability %f" % (info.language, info.language_probability))
for segment in segments:
print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
</code></pre>
<p>原因应该是没有安装cudnn相关库。</p>
<h2 id="方法一">方法一：</h2>
<h3 id="下载cudnn安装包">下载cudnn安装包</h3>
<p>去官网下载了对应的cudnn安装包，网址如下：https://developer.nvidia.com/rdp/cudnn-archive#a-collapse805-111<br>
<img src="https://img2024.cnblogs.com/blog/3491310/202411/3491310-20241114133706770-489673813.png" alt="image" loading="lazy"><br>
选择对应的linux x86_64版本下载。</p>
<h3 id="解压安装">解压安装</h3>
<p>参考https://zhuanlan.zhihu.com/p/701577195</p>
<h2 id="方法二">方法二：</h2>
<p>直接在创建的虚拟环境中执行<code>conda install cudnn</code><br>
然后再运行即可运行成功。</p>
<p>我这边直接在虚拟环境中执行方法二后，程序即成功运行了，故此没有再继续验证方法一。</p><br><br>
来源：https://www.cnblogs.com/kevinarcsin001/p/18545837
頁:
[1]