Text to Speech – Natural Voices, Free MP3 Download
AIFree text to speech with natural-sounding voices that run inside your browser. Paste up to 20,000 characters, pick one of 11 American or British English voices, set the speed and listen while the audio is generated. Download it as MP3 or WAV. The Kokoro voice model works on your device, so your text is never sent to a speech service.
- Runs on your device
- Kokoro TTS
- Works offline
Updated
How to convert text to speech
Step 1: Paste or type your text in the box above. The counter shows how much of the 20,000-character limit you have used.
Step 2: Press the play button on a few voices to preview them, then choose one and set the Speed (0.5× to 2×).
Step 3: Choose MP3 or WAV and press Generate speech. The voice model downloads once after you confirm.
Step 4: Listen as each part is ready, then download the finished file.
Good to know
Formats
Nothing to upload — everything starts from this page and is saved to your device.
How it runs
- Kokoro voices
- Speech is generated by the open Kokoro model (Apache 2.0) running in your browser.
Which voice should you choose?
Start with Heart or Bella: the model’s authors grade them highest, and they suit narration and study notes. For a male voice, try Michael or Fenrir; for British English, Emma or George. Every voice has a preview button, so listen to two or three with your own first sentence.
| Voice | Accent | Voice type | Authors’ grade |
|---|---|---|---|
| Heart | US English | Female, warm | A |
| Bella | US English | Female | A− |
| Nicole | US English | Female, soft | B− |
| Sarah | US English | Female | C+ |
| Michael | US English | Male | C+ |
| Fenrir | US English | Male, deep | C+ |
| Puck | US English | Male, bright | C+ |
| Emma | UK English | Female | B− |
| Isabella | UK English | Female | C |
| George | UK English | Male | C |
| Fable | UK English | Male | C |
What is TTS?
TTS (text to speech) is software that reads written text aloud in a synthetic voice. Older systems stitched recorded sounds together and sounded robotic; neural models such as Kokoro, which this page uses, generate the voice from patterns learned from real speech and sound much closer to a person reading.
Kokoro has 82 million parameters, small enough to run in a browser tab. Text is split into sentences, the first short part is spoken almost at once, and the rest follows in parts of a few hundred characters with a short natural pause between them, so long text starts playing before it is finished.
Save text to speech as MP3 or WAV
Choose MP3 for sharing, uploading and phones, and WAV if you will edit the voice in an audio or video editor. Both are mono at 24 kHz, the rate the voice model produces. The file is named after the first words of your text, so several clips stay easy to tell apart.
| Speech length | MP3 (128 kbps) | WAV (16-bit) |
|---|---|---|
| 1 minute | 1.0 MB | 2.9 MB |
| 10 minutes | 9.6 MB | 28.8 MB |
| 30 minutes | 28.8 MB | 86.4 MB |
Can you use text to speech commercially?
The Kokoro model is released under the Apache 2.0 licence, which allows commercial use, and ProMedia adds no licence of its own to the audio you generate. What matters is the text: you need the right to use the words you turn into speech. This is general information, not legal advice; check the terms of any platform you publish on.
Make text to speech for YouTube videos
A generated voice-over works for tutorials, explainers and slideshows. A simple workflow:
- Write the script in short sentences; they are read more naturally than long ones.
- Generate the voice-over as WAV if you will edit it, or MP3 if not.
- Put it under your footage with add music to video, or turn it into a video with a still image using MP3 to MP4.
- Add captions from the same audio with the subtitle generator.
Why read text aloud with ProMedia
- Private. Drafts, notes and documents are read on your device; no text goes to a cloud voice service.
- Free. No sign-up, no monthly character quota, no watermark in the audio. The limit is 20,000 characters per run, so split longer text.
- Listen while it generates. Playback starts with the first sentence.
- Offline after one download. The voice model is 93 MB and stays cached in your browser.
Read aloud on iPhone, Android, Mac or Windows
Every system has a built-in reader for text on screen. Use those to listen once; use this page when you want a natural voice saved as a file.
iPhone and iPad
Settings → Accessibility → Spoken Content → Speak Selection reads highlighted text aloud. It cannot save audio; paste the text here in Safari for an MP3.
Android
Select to Speak in Accessibility settings reads what you tap. For a file, open this page in Chrome; the voice model runs on the phone.
Mac
Turn on Speak selection under Accessibility → Spoken Content, then press Option + Esc to hear selected text. This page adds MP3 and WAV export.
Windows
Edge’s Read aloud (Ctrl + Shift + U) reads web pages and PDFs. To keep the audio, paste the text here and download it.
What the voices cannot do yet
The voices speak English only, with American or British accents; other languages would need different open models with suitable licences. There is no voice cloning and no per-word control of emphasis or pitch, only the overall speed. Unusual names and abbreviations can be mispronounced: spell them the way they sound, for example “Ess Kyoo Ell” for SQL.
Tips & troubleshooting
A slightly slower speed, such as 0.9×, sounds more deliberate for study notes and accessibility; 1.1–1.2× suits quick social clips.
Commas and full stops control the pauses. Add a comma where you want the voice to breathe.
Turn Play while it generates off if you only want the file and not live playback.
Want to change how your own recording sounds instead? Try the voice changer.
Need the opposite direction? Audio to text turns speech back into editable text.
Read Aloud (Text to Speech) FAQ
How do I convert text to speech for free?
Paste your text above, pick a voice and speed, choose MP3 or WAV and press Generate speech. The voice model downloads once with your consent and runs in your browser, so it is free with no account, no character quota beyond 20,000 per run, and no watermark.
How do I save text to speech as MP3?
Select MP3 before you press Generate speech, then use the download button when it finishes. The MP3 is mono at 128 kbps, about a megabyte per minute of speech, and plays on any phone, computer or video editor.
Is text to speech free here?
Yes. There is no subscription and no paid voice tier. The audio is generated on your own device by an open-licence model, so ProMedia has no per-character cost to pass on. The only download is the voice model, once.
What is TTS?
TTS stands for text to speech: software that reads written text aloud in a synthetic voice. Modern neural TTS, such as the Kokoro model used here, sounds much closer to a human reader than the robotic voices of older screen readers.
Can I use text to speech commercially?
The Kokoro voice model is Apache 2.0 licensed, which permits commercial use, and ProMedia puts no licence on the audio you create. You still need the rights to the text itself, and platforms may have their own rules about synthetic voices.
Which languages are available?
English, in American and British accents, with 11 voices. Other languages are not available yet; they need open voice models with licences that allow this kind of use.
How long can the text be?
Up to 20,000 characters per run, roughly a long article. The text is spoken in parts, so the first sentence plays within moments while the rest is generated. For a whole book, split it into chapters and generate each one.