20240202在WIN10下使用fast whisper缺少cudnn_ops_infer64_8.dll
2024/2/2 10:48

https://blog.csdn.net/feinifi/article/details/132548556
Could not locate cudnn_ops_infer64_8.dll. Please make sure it is in your library path!解决办法
安装cuDNN

c:\faster-whisper-webui>python cli.py --model large-v2 --vad silero-vad --language Japanese --output_dir c:\faster-whisper-webui\whisper_model c:\faster-whisper-webui\Downloads\test.mp4
Using faster-whisper for Whisper
[Auto parallel] Using GPU devices None and 8 CPU cores for VAD/transcription.
Creating whisper container for faster-whisper
Using parallel devices: None
Created Silerio model
Parallel VAD: Executing chunk from 0 to 74.072 on CPU device 0
Loaded Silerio model from cache.
Getting timestamps from audio file: c:\faster-whisper-webui\Downloads\test.mp4, start: 0, duration: 74.072
Processing VAD in chunk from 00:00.000 to 01:14.072
VAD processing took 6.546280500013381 seconds
Transcribing non-speech:
[{'end': 75.0716875, 'start': 0.0}]
Parallel VAD processing took 15.04769109992776 seconds
Device None (index 0) has 1 segments
(get_merged_timestamps) Using override timestamps of size 1
Processing timestamps:
[{'end': 75.0716875, 'start': 0.0}]
Running whisper from  00:00.000  to  01:15.072 , duration:  75.0716875 expanded:  0 prompt:  None language:  None
Loading faster whisper model large-v2 for device None
WARNING: fp16 option is ignored by faster-whisper - use compute_type instead.
Could not load library cudnn_ops_infer64_8.dll. Error code 126
Please make sure cudnn_ops_infer64_8.dll is in your library path!

百度：cudnn_ops_infer64_8.dll
https://blog.csdn.net/qq_30150579/article/details/128499694
Windows10下NVIDA CUDA,cuDNN和TensorRT安装教程

三. 安装cuDNN
去官网下载CUDNN，同前，注意版本号。下载地址：https://developer.nvidia.com/rdp/cudnn-download
【需要注册】
https://developer.nvidia.com/rdp/cudnn-download
cuDNN Download
NVIDIA cuDNN is a GPU-accelerated library of primitives for deep neural networks.

 I Agree To the Terms of the cuDNN Software License Agreement

Ethical AI
NVIDIA’s platforms and application frameworks enable developers to build a wide array of AI applications. Consider potential algorithmic bias when choosing or creating the models being deployed. Work with the model’s developer to ensure that it meets the requirements for the relevant industry and use case; that the necessary instruction and documentation are provided to understand error rates, confidence intervals, and results; and that the model is being used under the conditions and in the manner intended.

参考资料：
https://blog.csdn.net/qq_33959661/article/details/112971532
安装cudnn之后，tf训练报错缺少libcudnn_ops_infer.so.8解决方法

https://blog.csdn.net/Kelen99/article/details/121883743
Could not load library cudnn_ops_infer64_8.dll. Error code 126

https://www.yii666.com/blog/397588.html
记录-安装cuda与cudnn 及对应版本的tensorflow|pytorch tensorflow出现报错：Could not load library cudnn_cnn_infer64_8.dll. Error code 126

https://mp.weixin.qq.com/s?__biz=MzI1ODEzMDQ3OQ==&mid=2247487745&idx=1&sn=c8a083e688d2a8cdb0699a3f8ba81b1f&chksm=ea0d8641dd7a0f5784de711bb12f8c1788e70b30b2fe6ee39e158ed0ab9103db714a8e07124f&scene=27
Windows 上基于 TensorRT 的 YOLOV6 部署保姆级教程