Whisper.cpp

Description

whisper.cpp is a high-performance C/C++ inference implementation of OpenAI's Whisper automatic speech recognition model. It supports CUDA acceleration for fast multilingual speech recognition, translation, and language identification.

How to Build with SCALE

git clone https://github.com/ggerganov/whisper.cpp.git
cd whisper.cpp
cmake \
    -DCMAKE_BUILD_TYPE=Release \
    -DCMAKE_CUDA_COMPILER=nvcc \
    -DCMAKE_CUDA_ARCHITECTURES="$CUDAARCHS" \
    -DGGML_CUDA=ON \
    -Bbuild .
make -C build -j$(nproc)

Notes

• Built on top of ggml, a tensor library for machine learning
• All model sizes (tiny, base, small, medium, large) are supported
• CUDA acceleration significantly speeds up inference compared to CPU-only execution