Kokoro

A small (~82M parameter) hexgrad/Kokoro text-to-speech model that converts input text into spoken audio. It is optimized for efficiency and low cost, with pricing based on the length of the input text.