Fact · Speech recognition
Running faster-whisper on a GPU requires NVIDIA libraries to be installed first.
Checked against its source on October 17, 2026
In the source's own words
“GPU execution requires the following NVIDIA libraries to be installed:”
faster-whisper: README (GitHub, SYSTRAN), SYSTRAN, GitHub (SYSTRAN/faster-whisper), 2026, GPU.
Open the original- What it measures
- The README lists cuBLAS for CUDA 12 and cuDNN 9 for CUDA 12, and says the latest ctranslate2 releases support only CUDA 12 and cuDNN 9.
- What it does not show
- The README covers NVIDIA GPUs only. On a Mac, faster-whisper runs on the CPU.