pyttsx3

An offline, cross-platform text-to-speech library for Python that hooks directly into your operating system's native voice engines.

When you need text-to-speech without cloud latency, API keys, or internet dependencies, pyttsx3 is the industry standard (it runs entirely offline). The library acts as a universal wrapper around native OS engines: SAPI5 on Windows, NSSpeechSynthesizer on macOS, and eSpeak on Linux. With a clean, three-line initialization sequence (import, init, and say), developers can easily adjust speaking rates, toggle between male and female system voices, and export synthesized speech directly to MP3 files.

https://github.com/nateshmbhat/pyttsx3