Turn text into natural speech online

Type or paste up to 5,000 characters, choose a Spanish or English voice and create speech you can hear immediately. The selected model is downloaded once, then your text is processed on your device and never uploaded.

Free
No install, no account
Files never leave your computer
English and Spanish

How to turn text into speech

Everything you need is above: no account and no editor session. Start with a short paragraph to compare voices, then generate the full text.

Type or paste your text in Text. You can use up to 5,000 characters, including the pause tags described below.
Choose a Voice, set Speed from 0.75× to 1.5×, and select WAV or MP3 as the Format.
Click Generate speech. On first use, the tool downloads the model it needs; the progress bar shows the size and Cancel stops the download.
Listen to the preview and click Download to save the generated audio.

Voices, model sizes and pauses

Kokoro and Piper both run on your device. Models are requested only when you generate, and completed downloads are kept in this browser for later visits.

Kokoro offers three Spanish voices and four US English voices. Its shared model is roughly 90 MB, plus a runtime of about 27 MB the first time any compatible model is used.
Piper provides additional voices from Spain, Mexico and the US. Each voice has its own download of roughly 28 to 64 MB, plus the shared runtime if it is not present.
Write [pause] or [pausa] for half a second of silence, or set a length with [pause 2s] or [pausa 300ms]. A blank line between paragraphs adds 0.8 seconds.
Other square-bracket tags are skipped rather than read aloud, so you can remove them before generating if the result is not what you intended.

WAV or MP3, quality and practical limits

WAV is the default and keeps the generated samples without another lossy encoding step. MP3 is encoded at 192 kbps and is usually more convenient for sharing.

For editing, mixing or further processing, choose WAV. For a smaller file to send or publish, choose MP3.
Long passages take more time and memory to synthesize. Split a very long script into sections if your device struggles, and keep each generated file before replacing the text.
Pronunciation, emphasis and pacing depend on the selected model. Listen carefully to names, abbreviations, numbers and language changes before using the result.

Need to shape the audio afterwards?

Open the full editor to trim the speech, adjust its volume, remove noise, combine it with music or add it to a multitrack project. You can then export the finished audio in the format your project needs.

Frequently asked questions

How much text can I convert at once?

The text box accepts up to 5,000 characters. That is a practical limit for this quick tool, not a promise that every device can generate the longest text quickly; shorter sections are easier to manage.

Which format should I download, WAV or MP3?

Choose WAV when you plan to edit the voice or want to avoid MP3 quality loss. Choose MP3 at 192 kbps when a smaller, widely supported file is more useful.

Why is there a download before the first generation?

The speech model is not included in the page. It is fetched on demand and cached in your browser, along with a shared runtime when needed. Later generations reuse the files already present.

Can I use both Spanish and English voices?

Yes. The voice list includes Spanish and US English options from Kokoro and Piper. Select the language that matches each text for the best pronunciation; one generation uses one selected voice.

What should I check before using the generated audio?

Listen to the whole result and review pronunciation, numbers, pauses and names. Synthetic voices can make mistakes, so correct the text or try another voice and generate it again when necessary.