Text to Speech in Your Own Voice: Free Narration Tool (MP3, M4A, WAV)
Free text to speech narration tool: type a script and get a voice-over in 28 natural English voices or in your own voice from a 15-second sample. Runs in your browser, nothing uploaded. Download MP3, M4A or WAV, or add it to the MES Video Editor.
Type it, hear it: turn a script into a voice-over with a ready-made voice, or with your own voice from a short recording. Everything runs in this browser: your text and your voice never leave this computer.
28 English voices (American and British) from the Kokoro model (Apache-2.0). The first use downloads the voice model once; after that it works offline.
+ Add a voice
Record or upload 10 to 20 seconds of clear speech: one person, no music, a quiet room, speaking the way you want the narration to sound. Read this out loud, for example:
“Hello, and welcome back. Today we are going to work through a few examples step by step, so grab a pencil and some paper. If you get stuck, pause the video and try it yourself first.”
Uses Chatterbox (Resemble AI, MIT licence), English only. It copies the sound of the voice from your sample; each take comes out a little different, so a line you don't like can be made again.
Your narration
How the narration tool works
Type or paste a script and press Make the narration. The text is cut into sentences, each one is spoken by a voice model running in your browser, and the pieces are joined with the pauses you choose. Listen to each line on its own, make any line again, then download the whole thing as MP3, M4A or WAV, or drop it straight into the MES Video Editor.
Ready-made voices
28 English voices (American and British, men and women) from the open Kokoro-82M model. The first time, the model (about 90 MB) is downloaded and kept by the browser; it then works offline on any computer, and on most phones too (more slowly).
Your own voice
Record or upload 10 to 20 seconds of your speech once, save it, and every script after that can be read in a voice that sounds like you. This uses the open Chatterbox model by Resemble AI (MIT licence). It is big (about 1.5 GB, downloaded once) and needs a computer with a graphics chip the browser can use (WebGPU: Chrome, Edge, Brave or Opera on Windows, Mac or ChromeOS). Without one it still works, but very slowly.
Questions
- Is my voice uploaded anywhere?
- No. The recording is kept in this browser (IndexedDB) and the voice model runs here. Delete a voice with its × button.
- How do I get a better copy of my voice?
- Use a clean, close recording with no echo or music, 10 to 20 seconds long, spoken in the style you want. Try a few samples and keep the best one.
- Can I use someone else's voice?
- Only with their permission. Copying a person's voice without consent to mislead people is wrong and in many places illegal.
- Which languages work?
- English. Other languages are read with English pronunciation.
- How do I add pauses?
- A blank line starts a new paragraph (a longer pause). Type
[pause]for 1 second or[pause 2.5]for 2.5 seconds anywhere in the text.