Kokoro at a glance
Checked 15 Oct 2026Kokoro is an 82-million-parameter text-to-speech model (v1.0, January 2025) with 54 preset voices across 8 languages, run through the `kokoro` Python library. Code and weights are both Apache-2.0, so commercial use is allowed. It needs espeak-ng installed alongside. It does not clone voices from a sample and does no speech recognition. Hardware requirements are not documented.
RepoQuarry. “Kokoro.” Checked 15 Oct 2026. https://repoquarry.dev/projects/kokoroHow much of ElevenLabs does it replace?
1 of 4 jobs. Speech generation from 54 preset voices only; no voice cloning or custom voice design.
For voice cloning, speech recognition, voice agents, see other ElevenLabs alternatives. You may also need to build an interface, an API server and hosting yourself.
Facts, with sources
Every answer links to where we found it. We read the documentation; we did not install it.
With Apache-licensed weights, Kokoro can be deployed anywhere from production environments to personal projects.github.com · kokoro#readme ↗ 1d ago
This is an Apache-licensed model, and Kokoro has been deployed in numerous projects and commercial APIs.huggingface.co · Kokoro-82M ↗ 1d ago
This is an Apache-licensed model, and Kokoro has been deployed in numerous projects and commercial APIs.huggingface.co · Kokoro-82M ↗ 1d ago
On Mac M1/M2/M3/M4 devices, you can explicitly specify the environment variable PYTORCH_ENABLE_MPS_FALLBACK=1 to enable GPU acceleration.github.com · kokoro#windows-installation ↗ 1d ago
An inference library for Kokoro-82M. You can pip install kokoro.github.com · kokoro#readme ↗ 1d ago