CMU Sphinx

A veteran open source speech recognition toolkit providing speaker-independent large vocabulary continuous recognition.

CMU Sphinx (also known as PocketSphinx, Sphinx-4, and Pocketsphinx-Python) delivers a flexible suite of speech recognition tools developed at Carnegie Mellon University. It supports multiple languages including English, French, and Mandarin, utilizing Hidden Markov Models (HMMs) and n-gram statistical language models. The toolkit is designed for efficiency on low-resource platforms (PocketSphinx) and high-level research (Sphinx-4), making it a reliable choice for offline transcription and voice command integration without relying on cloud-based APIs.

https://cmusphinx.github.io/