High-Performance ASR in C++: transcribe.cpp Library

🛠 transcribe.cpp library for high-performance speech recognition

A cross-platform C++ library for ASR has been introduced, supporting over 16 model families (including Whisper, Parakeet, Canary, and Moonshine) via the ggml runtime and GGUF format. The project provides hardware acceleration on GPUs (CUDA, Vulkan, Metal) and includes ready-to-use bindings for Python, JS, Rust, and Swift.

🌍 A specialized engine based on ggml simplifies the deployment of high-quality speech recognition on edge devices and local servers without the need for heavy dependencies.

👤 Developers can now run advanced models locally with GPU support across various operating systems using familiar programming languages.

Source 1: https://github.com/handy-computer/transcribe.cpp

Sources