parakeet.cpp

parakeet.cpp is an open-source C++ library for running NVIDIA's Parakeet automatic speech recognition (ASR) models locally, enabling fast and efficient offline transcription. It is designed for developers who want to integrate high-quality speech-to-text capabilities into their applications without relying on cloud services. The project follows in the footsteps of whisper.cpp, bringing efficient local inference to Parakeet-family models. It targets performance-conscious use cases where low latency and privacy matter.

• Local inference: Run Parakeet ASR models entirely on-device, with no internet connection required, ensuring data privacy.
• C++ implementation: Built in C++ for high performance and easy integration into native applications and pipelines.
• Parakeet model support: Supports NVIDIA's Parakeet family of ASR models, known for state-of-the-art English transcription accuracy.
• Lightweight design: Minimal dependencies and a small footprint make it suitable for embedding in resource-constrained environments.
• Open source: Freely available on GitHub under an open-source license, welcoming community contributions and forks.
• Developer-friendly: Provides a straightforward API for integrating speech recognition into C++ projects, with example usage in the repository.

To get started, clone the repository from GitHub, follow the build instructions in the README to compile the library, download a compatible Parakeet model checkpoint, and run the provided example to transcribe audio files locally.