PF4Public@users.noreply.github.com ggml-org/whisper.cpp https://github.com/ggml-org/whisper.cpp/issues whisper.cpp is a port of OpenAI's Whisper automatic speech recognition (ASR) model to plain C/C++, built on top of the vendored ggml tensor library. It provides the C APIs libwhisper and libparakeet (Parakeet TDT), the command line tools whisper-cli, whisper-bench, whisper-quantize, whisper-server, whisper-vad-speech-segments, parakeet-cli and parakeet-quantize, and optional GPU backends (CUDA, Vulkan, OpenCL) in addition to the CPU backend. Enable the NVIDIA CUDA backend (GGML_CUDA) Link the examples against FFmpeg to decode more audio formats Use OpenBLAS as the BLAS backend (GGML_BLAS) Enable the OpenCL backend (GGML_OPENCL) Use OpenMP for threading (GGML_OPENMP) Enable the RPC backend to distribute layers across machines (GGML_RPC) Build the SDL2-based examples whisper-stream, whisper-command, whisper-lsp and wchess Link against the installed sci-ml/ggml instead of the bundled copy; the CPU ISA set and the backends are then selected by ggml's own USE flags Enable the Vulkan backend (GGML_VULKAN)