Glyphcall Voice

Kokoro TTS online, free

Kokoro-82M is one of the best small open text-to-speech models. Try it here without Python, a GPU or an account; the model is downloaded once and cached by your browser.

The first time, your browser fetches the voice (about 90 MB) and keeps it. After that it is ready on this device, and playback begins after the first sentence. Voices in English, Spanish, French, Italian, Portuguese and Hindi.

Whole PDF or ebook? The free PDF and ebook to audio maker reads it chapter by chapter. Want it offline? Install the free app.
Studio voices on our servers: Voice Pro, $5 a month for 100,000 characters (Get Voice Pro: $5 a month), or $7 a month for voices and transcription together (Get both for $7). Pay by card, Apple Pay, Google Pay or Link.
For developers: the same job is an API: up to 2,000 characters as one MP3 per call, from one cent per call, no account and no monthly fee. OpenAI and ElevenLabs request formats work as they are. Text to speech API · Buy credits
Also in your tools: n8n node · Make.com app · Obsidian plugin · Apify: audio and video transcriber · Apify: podcast transcriber

What Kokoro-82M is

Kokoro is an open-weight text-to-speech model with 82 million parameters, released under the Apache-2.0 licence. It is small enough to run on an ordinary laptop or phone, yet its voices sound natural, with steady pacing and few of the robotic artefacts of older small models.

How this page runs Kokoro

Kokoro voices

All 41 Kokoro v1.0 voices for English, Spanish, French, Italian, Brazilian Portuguese and Hindi are in the list, best-rated first: 20 American and 8 British English voices, plus 13 more. A few good starting points:

Speed runs from 0.6x to 1.6x. Kokoro reads punctuation: commas give short pauses, full stops longer ones, and a new paragraph a clear break.

How to use Kokoro TTS online

  1. Paste or type your text in the box above.
  2. Pick a voice and, if you like, change the speed.
  3. Press Speak. The first run downloads the model; after that, speech starts within a few seconds.
  4. Listen, then press Download MP3 or Download WAV.

Tips for better Kokoro output

Kokoro as an API

Need Kokoro-quality speech from your own code, a phone, or a server with no browser? The hosted text-to-speech API takes OpenAI-shaped requests (POST /v1/audio/speech), returns MP3, costs from one cent per call of up to 2,000 characters, and needs no account or monthly plan.

Questions

Is it really free? Yes. Your own computer makes the speech, so there is no account, free every month and no watermark.

Which model makes the voices? The open Kokoro-82M model (Apache-2.0 licence), run on your own computer.

Is my text uploaded? No. The text stays on your computer. Only the voice model is downloaded, once.

Which languages? English (American and British, 28 voices), Spanish, French, Italian, Brazilian Portuguese and Hindi: 41 voices in all. Other languages are not offered yet.

Do I need Python or a GPU to run Kokoro? No. This page runs Kokoro-82M in the browser with WebAssembly. A recent Chrome, Edge, Firefox or Safari on a computer from the last few years is enough.

Does Kokoro work offline? Yes, after the first run. The model is cached by your browser, so you can disconnect and keep making speech.

Which Kokoro languages work here? English (American and British, 28 voices), Spanish, French, Italian, Brazilian Portuguese and Hindi: 41 voices in all. Other languages are not offered yet. Japanese and Chinese need a different phonemizer and are not offered.

How long can the text be? There is no fixed cap. Very long texts take longer, since your own computer does the work; for whole books use the audiobook maker, which saves each chapter as it goes.

More free tools: Convert text to MP3 · Free voice generator from text · Read any text aloud · Free voice-over for your videos · Free text to speech, no sign-up · Free AI voice generator · Free online text reader · TTS online · Transcribe audio · SRT generator