Free tool · No sign-up

Free Text to Speech: Natural AI Voices, MP3 Download

Last updated By Avalon

Turn text into natural-sounding speech and download it as MP3 or WAV.

    Your text never leaves your device. The voice is made by an AI model that runs in your browser, so nothing is uploaded to a server. The audio is AI-generated; please say so where you publish it.

    What this tool does

    To turn text into speech for free, paste your text above, pick a voice and press Create speech: you can play the result right away and download it as an MP3, with no sign-up. The voices come from Kokoro, an open AI voice model, and run in your browser, so your text stays on your device. It suits voice-overs for short videos, narration for slides, proofreading by ear and listening to your own writing.

    How to use it

    1. Type or paste your text into the box, or open a .txt file.
    2. Choose a Voice and a Speed.
    3. Choose MP3 for a small file that plays everywhere, or WAV for video editing.
    4. Press Create speech. The first time, the voice model downloads once.
    5. Press Play to listen, then Download.

    Tips for natural speech

    • Write numbers, dates and prices the normal way: "$19.99", "October 4" and "9:30" are read out in full.
    • Punctuation shapes the rhythm: a comma adds a short pause and a new paragraph a longer one.
    • For a video voice-over, keep sentences short and leave an empty line where the scene changes.
    • If a name sounds wrong, spell it the way it is said.
    • Need captions for the finished video? Use our free subtitle maker.

    Credits

    The voices are Kokoro-82M by hexgrad (Apache 2.0). Pronunciations use the dictionaries from misaki (Apache 2.0). The model runs with ONNX Runtime Web (MIT), and MP3 files are written with lamejs (LGPL). Full notices: third-party licenses.

    Free, Pro and Pro+

    FreeProPro+
    Text per audio file1,000 characters20,000 characters20,000 characters in your browser, 40,000 characters with fast server voices
    One audio file per paragraph, as a ZIP—YesYes
    Saved presets—YesYes
    Result qualityThe same in your browser on every planThe same in your browser, higher accuracy on our server
    Server tools paid with creditsWith a credit packWith a credit pack300 credits every month

    Pro+ credits pay for server tools: higher-accuracy speech to text, higher-accuracy auto subtitles and fast server voices. A credit pack (150 credits, valid for 12 months) works on any plan. Compare plans.

    Frequently asked questions

    Can I use the audio in YouTube videos or for commercial work?

    Yes. The voice model, Kokoro-82M, is released under the Apache 2.0 license, which allows commercial use, and we do not add any terms of our own to the audio you make. The voice is AI-generated: say so where your platform asks you to label AI content.

    Is my text uploaded anywhere?

    No. The voice model runs inside your browser, on your own device. Your text is not sent to our servers or anyone else. The only exception is the optional Pro+ fast server voices, which you switch on yourself.

    Why does the first run take longer?

    The first time, your browser downloads the voice model and its pronunciation dictionary, about 97 MB. They are kept in your browser, so later visits start right away.

    How long does it take?

    It depends on your device. On a recent laptop the audio is usually made faster than it plays; phones and older computers are slower. The progress bar shows how far along it is, and you can stop at any time.

    Which languages does it speak?

    English, with American voices. Other languages are not supported yet.

    Why does a word sound wrong?

    Pronunciations come from a dictionary of about 180,000 English words. Names and new words that are not in it are read by spelling rules, which can miss. Writing the word the way it sounds, for example Shiv-awn for Siobhan, usually fixes it.

    What are the limits?

    The free version voices up to 1,000 characters at a time, with no daily cap. Avalon Pro voices up to 20,000 characters at a time, can save one audio file per paragraph in a ZIP, and keeps your voice and speed as saved presets. Avalon Pro+ adds fast server voices for texts up to 40,000 characters, paid with credits.

    Get each new experiment by email

    One short email at the end of each month: the experiments we ran, the numbers, and the steps you can copy. Unsubscribe any time.