#8142 CUDA GPU acceleration unavailable. Verify NVIDIA GPU & CUDA installation (Help -> CUDA Help).

CUDA acceleration is unavailable primarily because the environment cannot detect a usable NVIDIA GPU or a compatible CUDA/cuDNN installation.

Solution steps

1. Confirm NVIDIA GPU exists
   Open CMD and run:

   nvidia-smi

   If it fails, open Device Manager → Display adapters. Only NVIDIA GPUs support CUDA; AMD/Intel GPUs are not supported.

2. Update NVIDIA driver
   CUDA 12.8+ requires a recent driver. Install the latest driver from NVIDIA's official site.

3. Install CUDA Toolkit 12.8 or newer
   Install CUDA Toolkit 12.8+ (12.8 or 13.x is compatible). Verify after install:

   nvcc -V

4. Install cuDNN 9.11 or newer
   Download cuDNN for your CUDA version and extract it.
5. Add CUDA/bin and cuDNN/lib to system PATH
   Ensure the bin directory of CUDA and the lib directory of cuDNN are present in the system environment variable PATH, then restart the computer and pyVideoTrans.
6. If no NVIDIA GPU is present
   CUDA acceleration cannot be enabled. Local ASR/TTS will fall back to CPU, which will be much slower. You can still use online API channels (e.g., OpenAI ASR, Edge-TTS) without CUDA.

Note: Your version is v4.15. Upgrading to the latest v4.16 is also recommended to rule out known bugs that may affect CUDA detection.

Please consult the related documentation:

• Configure CUDA and cuDNN on Windows