Kokoro review

Q: How do I install Kokoro?

Run pip install kokoro. The kokoro package needs Python 3.10 or newer, below 3.13. The install section above has the other options from the project's README.

Kokoro is a Python inference library for Kokoro-82M, an open-weight text-to-speech model with 82 million parameters.

Updated Nov 1, 2026. Review written Oct 22, 2026 from the project's own sources; usage data refreshed every Monday.

TL;DR

Kokoro is a Python inference library for the Kokoro-82M open-weight TTS model. It is for developers who want to generate speech in code, from production services to personal projects. The README describes the model as lightweight and states that its weights are Apache-licensed.

9.8kGitHub stars
+174new stars, week of Oct 25
150.7kdownloads a week, +50% vs 4-week avg
#2in Voice and realtime this week

Kokoro at a glance

Category: Voice and realtime
Maintained by: hexgrad (individual, @hexgrad on GitHub)
License: Apache 2.0
Language: JavaScript, Python, CSS
Repository: github.com/hexgrad/kokoro
PyPI package: kokoro 0.9.4
Latest release: kokoro 0.9.4 on PyPI, Apr 5, 2025
First published: Jan 10, 2025

What Kokoro does

Kokoro converts text to speech through a KPipeline object and yields audio chunks at a 24000 Hz sample rate.
Kokoro supports American and British English, Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese and Mandarin Chinese through language codes.
Kokoro lets you choose a voice by name, such as af_heart, or load a voice tensor from a .pt file.
Kokoro accepts a speed setting and a split_pattern that controls how input text is split into chunks.
Kokoro returns graphemes and phonemes alongside each audio chunk, and it accepts inline pronunciation hints in brackets.
Kokoro uses the misaki G2P library for grapheme-to-phoneme conversion, and it falls back to espeak-ng for out-of-dictionary English words.

How it works

Kokoro runs as a Python library. You install it with pip, create a KPipeline with a language code, and pass it text and a voice. The pipeline returns a generator of graphemes, phonemes and audio, which you can play or save with soundfile. It relies on misaki for G2P and espeak-ng for fallback. It can run in Google Colab.

How to install and start using Kokoro

With pip

pip install kokoro

espeak-ng on Debian/Ubuntu

apt-get -qq -y install espeak-ng > /dev/null 2>&1

Japanese support

pip install misaki[ja]

Requirements

Python 3.10 or later and earlier than 3.13 (version 0.9.4 on PyPI).
espeak-ng, used for English fallback and some non-English languages. On Windows, install it with the .msi installer from the espeak-ng releases page.
misaki[ja] for Japanese and misaki[zh] for Mandarin Chinese.
On Apple Silicon Macs, set PYTORCH_ENABLE_MPS_FALLBACK=1 to enable GPU acceleration.

First steps

Install kokoro and soundfile with pip, then install espeak-ng.
Create a pipeline with KPipeline(lang_code='a') for American English, and make sure the language code matches the voice.
Call the pipeline with your text and voice='af_heart', then loop over the generator to get audio chunks.
Save each chunk with sf.write as a WAV file at 24000 Hz.

Who Kokoro is for

Python developers who want to add local text-to-speech to scripts or applications.
Projects that need open-weight TTS models that can be deployed in production or personal projects.
Developers who need speech in English and several other languages, including Japanese, Hindi and Mandarin Chinese.
Developers who want to experiment with TTS quickly in Google Colab notebooks.

Limitations to know

Kokoro depends on espeak-ng, which is a separate system install on each platform.
Kokoro supports a fixed set of languages, and Japanese and Chinese need extra misaki packages.
Kokoro requires Python 3.10 to 3.12. Python 3.13 is not supported.
Kokoro does not use Apple Silicon GPU acceleration by default. It needs an environment variable set.

Is Kokoro growing? Usage data

In the week of Oct 25–31, 2026, Kokoro ranked #65 on the Devtoolwatch Board and #2 in Voice and realtime, with a momentum score of 72 out of 100. Its main package had 150.7k downloads that week, +50% against its own 4-week average.

New GitHub stars per week for the last 12 weeks, from GitHub's stargazer history (today's stargazers by the week they starred).

Show the weekly numbers

New GitHub stars per week for Kokoro
Week of	New stars
Aug 9, 2026	105
Aug 16, 2026	90
Aug 23, 2026	87
Aug 30, 2026	106
Sep 6, 2026	106
Sep 13, 2026	96
Sep 20, 2026	129
Sep 27, 2026	114
Oct 4, 2026	107
Oct 11, 2026	172
Oct 18, 2026	248
Oct 25, 2026	174

The Board ranks tools by four signals each week: new stars, new stars for the repo's size, added downloads and download growth beyond the market. How the ranking works.

Kokoro alternatives

Other voice and realtime AI tools on the Board in the week of Oct 25–31, 2026, by momentum:

Voice and realtime ranked by momentum, the week of Oct 25–31, 2026
Rank	Tool	Stars	Downloads/wk
#1	faster-whisperfaster-whisper is a Python reimplementation of OpenAI's Whisper speech-to-text model built on the CTranslate2 inference engine for…	26.4k	4.4M
#3	ChatterboxChatterbox is a family of open-source text-to-speech and voice cloning models by Resemble AI, with English and multilingual variants.	27.5k	73.2k
#4	VoiceMemInfrastructure for the next generation of voice agents, designed to provide universal memory. It is divided into a left brain and a…	2.8k	–
#5	PipecatPipecat is an open-source Python framework for building real-time voice and multimodal conversational agents, maintained by Daily and…	17k	348.5k
#6	WhisperXWhisperX is a Python speech recognition tool built on Whisper that adds word-level timestamps and speaker diarization.	25k	158.6k
#7	LiveKit AgentsLiveKit Agents is an open-source Python framework for building realtime voice, video, and multimodal AI agents that join LiveKit rooms…	15k	2.4M

All voice and realtime AI tools, ranked

Kokoro FAQ

Is Kokoro free and open source?

Yes. Kokoro's source code is on GitHub (hexgrad/kokoro) under the Apache 2.0 license, so you can use it for free.

How do I install Kokoro?

Run pip install kokoro. The kokoro package needs Python 3.10 or newer, below 3.13. The install section above has the other options from the project's README.

Is Kokoro still maintained?

Activity has slowed: the hexgrad/kokoro repository was last updated on Aug 6, 2025, and the latest release, kokoro 0.9.4 on PyPI, came out on Apr 5, 2025.

How popular is Kokoro?

Kokoro has 9.8k GitHub stars as of Nov 1, 2026. It gained 174 new stars in the week of Oct 25–31, 2026, against a 12-week average of 128 a week. Its main package had 150.7k downloads that week, +50% against its own 4-week average. That made it #2 among voice and realtime AI tools on the Devtoolwatch Board.

Kokoro vs Chatterbox: how do they compare?

Kokoro and Chatterbox are both tracked by Devtoolwatch as voice and realtime AI tools. Kokoro has 9.8k GitHub stars and Chatterbox has 27.5k. In the week of Oct 25–31, 2026, Kokoro's main package had 150.7k downloads and Chatterbox's had 73.2k. Chatterbox is a family of open-source text-to-speech and voice cloning models by Resemble AI, with English and multilingual variants. Chatterbox review.

What are the best alternatives to Kokoro?

Other voice and realtime AI tools on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: faster-whisper (#1), Chatterbox (#3), VoiceMem (#4), Pipecat (#5), WhisperX (#6). The alternatives table above has their numbers.

How do I use Kokoro TTS in Python?

Kokoro is used by creating a KPipeline with a language code, such as 'a' for American English. You call the pipeline with text and a voice like 'af_heart', then iterate over the generator. Each item gives graphemes, phonemes and 24000 Hz audio, which you can save with soundfile.

Can I run Kokoro locally?

Kokoro runs locally as a Python library. Install it with pip, install espeak-ng for your OS, and run your script. On Apple Silicon Macs, you can set PYTORCH_ENABLE_MPS_FALLBACK=1 to enable GPU acceleration.

What languages does Kokoro support?

Kokoro supports American English, British English, Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese and Mandarin Chinese. Japanese needs misaki[ja] and Chinese needs misaki[zh].

What is Kokoro-82M?

Kokoro-82M is an open-weight TTS model with 82 million parameters. Its weights are Apache-licensed. The kokoro library is the inference library for this model.

Is Kokoro a Japanese word?

Kokoro is a Japanese word. The README says it translates to "heart" or "spirit."

Can I use a custom voice file with Kokoro?

Kokoro can load a voice tensor directly with torch.load from a .pt file. You then pass the tensor to the pipeline as the voice argument instead of a voice name.

Sources

Kokoro README badge

Maintainers can show the tool's weekly Board rank in their README:

[![Devtoolwatch](https://index.devtoolwatch.com/badge/kokoro.svg)](https://index.devtoolwatch.com/tools/kokoro/)

How this page is made: Devtoolwatch writes each review from the project's README, releases, package metadata and website, with an AI model, then checks every claim against the source it cites and drops what the sources don't support. The numbers come from Devtoolwatch's daily GitHub snapshots, GitHub's stargazer history, and the npm and PyPI download APIs. Spotted an error? Email [email protected].