whisper.cpp

whisper.cpp is a C/C++ port of OpenAI's Whisper model, designed for high-performance inference of automatic speech recognition (ASR). It offers a lightweight implementation, making it suitable for integration across diverse platforms. The core implementation is contained in whisper.h and whisper.cpp, utilizing the ggml machine learning library. Key features include Apple Silicon optimization (ARM NEON, Accelerate framework, Metal, Core ML), AVX/VSX intrinsics support, mixed F16/F32 precision, and integer quantization. It supports CPU-only inference, efficient GPU support (NVIDIA, OpenVINO, Ascend NPU, Moore Threads GPU), and a C-style API. Platforms supported include Mac OS, iOS, Android, Java, Linux, WebAssembly, Windows, Raspberry Pi, and Docker. Zero memory allocations at runtime are also a focus, enhancing efficiency.