mmwillet2/Kokoro_GGUF

• hexgrad/Kokoro-82M

Purpose

The purpose of this repository is to store various TTS.cpp compatible GGUF encoded model files for the Kokoro TTS model.

Model Types

Currently there are two types of model each with five levels of quantization. The GGUF model files containing espeak are configured to expect and use espeak for phonemization and those containing noespeak support tts native phonemization. The GGUF model files with no quantization suffix (i.e. Kokoroespeak.gguf and Kokoronoespeak.gguf) use only 32bit floating point precision and the GGUF model files with Q4, Q5, Q8, and F16 suffixes support Q40 quantization, Q50 quantization, Q8_0 quantization, and 16 bit floating point precision respectively.

Kokoro

This page only contains the GGUF encoded model files of the original Kokoro model. For the original model please see the repository here.