0 / 3,000 English voices only
The voice model is about 92 MB. It downloads once, when you press the button for the first time, and is then kept in your browser. A phone needs a good connection and enough free memory.
About the AI Voice Generator
This AI voice generator turns your text into a natural English voice and lets you download it as an MP3 or WAV file. Pick one of 28 voices, set the speed, press the button, and listen. Then save the file for a video, a podcast, a slide show, a school project or an audio message.
It is free and private. A small AI voice model runs inside your browser, on your own device. Your text is never sent to a server, and there is no sign-up and no watermark. The model is about 92 MB. It downloads once and your browser keeps it, so the next time it starts faster.
Features
- 28 natural voices — 20 American and 8 British, female and male
- Speed control — slower for learning or clear speech, faster for short clips
- MP3 and WAV download — MP3 is small, WAV is full quality
- Long text is fine — up to 3,000 characters, made sentence by sentence with a progress count
- Hear a sample first — one short sentence in the voice you picked
- Stop any time — you do not have to wait for a long text
- Private — the text never leaves your device
Why it is free and private
Most AI voice websites send your text to a server, run the voice there, and charge by the character. This page does the work on your own computer or phone instead. That is why there is no account, no limit per month and no upload.
The price is the first download. The voice model is about 92 MB, so use Wi-Fi if you can. After that, the model stays in your browser's storage.
What it can and cannot do
- English only. The voices speak English with an American or a British accent. If you type another language, the result will sound wrong. For other languages, try the Text to Speech tool, which uses the voices that are installed on your device.
- No voice cloning. You cannot copy your own voice or the voice of another person. You can only choose one of the ready-made voices.
- Numbers, dates and short forms. Write "twenty twenty-six" if the voice reads a number in a strange way. Short forms and unusual names may need a spelling that sounds right.
- Speed on your device. On a normal computer the speech is made about two to three times slower than it plays, so one minute of speech takes two to three minutes. A phone is slower, and an old phone may run out of memory. If that happens, use a shorter text.
Tips for a good result
- Write the text the way you would say it. Short sentences sound best.
- Use full stops and commas to make pauses. Start a new paragraph for a longer pause.
- Press Hear this voice to compare two or three voices before you make a long file.
- For learning or for clear speech, set the speed a little lower, about 0.85.
- Cut the file in parts or join parts with the Audio Trim and Audio Merger tools.
Frequently asked questions
Is this AI voice generator really free?
Yes. There is no sign-up, no credit system and no watermark. The voice runs on your own device, so there is no server cost for us to pass on to you.
Is my text uploaded?
No. The text is turned into speech in your browser. The only thing downloaded is the voice model, once, from a public file server. Your words are never sent anywhere.
Which voices and languages does it have?
There are 28 English voices: American female, American male, British female and British male. Other languages are not supported by this tool. Use Text to Speech for other languages. It uses the voices of your device and cannot save a file.
Can I use the voice in my videos, podcasts or for sales?
The voice model is released under the Apache-2.0 licence, which allows commercial use. Platforms may have their own rules about AI voices, for example a label for AI-made audio, so read the rules of the place where you publish.
Can it copy my voice or a famous voice?
No. This tool does not do voice cloning. You can only use the ready-made voices in the menu.
How long can the text be?
Up to 3,000 characters at a time, about 500 words. That is around three minutes of speech. For a longer text, make it in parts and join the files with the Audio Merger.
Why is the first start slow?
The first time, your browser downloads the voice model, about 92 MB. This happens once. After that the model comes from your browser's cache, and the page starts much faster.
Does it work on a phone?
It can, on a recent phone with a good connection and free memory. It is slower than on a laptop. If the page stops, try a shorter text or use a computer.
Related tools
- Text to Speech — read text aloud in many languages with your device's voices
- Audio to Text — the other way: turn a recording into text
- Audio Trim — cut the start or end of your voice file
- Audio Converter — change your file to another audio format
- Voice Recorder — record your own voice
Powered by the Kokoro-82M voice model (Apache-2.0) and kokoro-js (Apache-2.0).
How to use it
- Type or paste your text into the box. English only, up to 3,000 characters.
- Pick a voice and set the speed. Press Hear this voice to try it.
- Press Make the voice. The first time, wait while the model downloads.
- Listen to the result with the player.
- Press Download MP3 or Download WAV to save the file.