Clip Utils
Go Pro

Text to Speech: Free AI Voices, MP3 Download

Type or paste any text, pick one of 57 natural AI voices in 14 languages, from English and Hindi to Tamil, Arabic and Japanese, and download it as an MP3. It's free with no sign-up, and your text is never kept.

Your text never leaves your device. How?

The voice is made by an AI model that runs in your browser, so your text isn't sent anywhere. The model (about 92 MB) downloads once from Hugging Face and is then kept on your device. (On phones, it's made on our server instead, for speed, and isn't kept.)

0 / 10,000 characters

Language
Voice

✨ Better voicesClearer and more natural · Pro

Speed

Free, no sign-up, no watermark. The first time, the AI voice downloads once (about 92 MB); after that it starts in seconds, even offline.

How to turn text into speech

1

Type or paste your text

A script, a story, an article or a message, up to 10,000 characters at a time.

2

Pick a voice

Tap ▶ to hear any of the 57 voices first, then choose a speed.

3

Make the voice

On a phone, the voice is made on our server in seconds. On a computer, the AI voice downloads once (about 92 MB) and the speech is made in your browser, sentence by sentence.

4

Listen and download

Play it on the page and download it as an MP3 (or WAV), or turn it into a video with captions.

Hear the English voices

Each voice saying the same line. Pick the one you like in the tool above. “Hi! This is how I sound. Type anything, and I'll read it out for you.”

Alina · Woman

Wayne · Man

Heart · Woman

Bella · Woman

Nicole · Soft

Michael · Man

Fenrir · Man

Puck · Man

Emma · Woman

George · Man

What people use it for

Voice-overs for YouTube videos, Reels, Shorts and TikToks, especially faceless videos; narration for presentations and explainer videos; reading an article or study notes aloud to listen on the go; and audio for a message, a story or an announcement when you'd rather not record your own voice.

For a video, make the voice here, then open it in Audio to Video to add a background and captions that follow the words, or put the MP3 in CapCut, InShot or any video editor.

How the voice is made

Alina and Wayne are Murf's voices, made on our server in seconds on any device. The other voices come from Kokoro, an open-source AI speech model that sounds natural and is free to use, even commercially. On a computer it runs in your browser: your text is turned into the sounds of English, sentence by sentence, and the model speaks them in the voice you picked, with nothing uploaded. On phones, the same voice is made on our server in seconds (through DeepInfra), and your text isn't kept.

On a computer, the model (about 92 MB) downloads once from Hugging Face and is kept, so the next time it starts in seconds, even offline, and several parts of a long text are made at once.

Features

Natural voices

10 English voices (American and British, women and men, including a soft one for calm narration), 6 Hindi voices, and voices in Spanish, Brazilian Portuguese, French, German, Italian, Arabic, Japanese and Chinese. Hear each one before you choose.

MP3 download

Download the speech as an MP3 for any phone, editor or app, or as a WAV for editing. No watermark, no sign-up.

Private

On a computer, the voice is made in your browser. On phones, and for Hindi typed in English letters, your text goes to our server and is never kept.

Long texts

Up to 5 minutes of speech free at a time, as many times as you like; up to 10,000 characters (about 11 minutes) at once with Pro.

Better voices with Pro

For a voice-over that sounds clearer and more natural, Pro adds 18 better voices: in English, Hindi (Riya and Manoj), Spanish, Portuguese, French, Italian, Japanese and Chinese. Tap ▶ to hear each one free first.

Frequently asked questions

Is there a free text to speech with a British accent?

Yes. Emma and George speak British English, free, next to the American voices; the British Text to Speech page starts on Emma. Download the voice as an MP3, with no sign-up and no watermark.

Can I turn a long text or a blog post into audio?

Yes. Paste up to 10,000 characters at a time. The first 5 minutes of speech are free, as often as you like, so a short post is free whole; a longer one is read in full with ClipUtils Pro, or in parts of about 300 words each.

Which voice suits a faceless YouTube channel?

Michael and Fenrir are American men's voices and Heart and Bella American women's voices, all free; George and Emma are British. Play each with ▶ and pick the one that fits the channel, then keep it for every video so it becomes the channel's voice.

Is this text to speech really free?

Yes. There's no sign-up, no watermark and no monthly limit: up to 5 minutes of speech at a time, as often as you like. For a longer text, the first 5 minutes are free and all of it is part of ClipUtils Pro.

What are the better voices?

Clearer, more natural AI voices, made on our server in seconds: 18 of them in English, Hindi (Riya and Manoj), Spanish, Portuguese, French, Italian, Japanese and Chinese. Anyone can hear their samples in the voice list; voice-overs in them are part of ClipUtils Pro, up to 30 minutes a day. The usual voices stay free.

Can it speak in my own voice?

Yes: pick "Your own voice" in the voices, read one sentence out loud (about 10 seconds), and anything you type is spoken in your voice, in any of the 10 languages your own voice covers (Hindi in beta). You hear a sample and one short sentence free; voice-overs in your own voice are part of ClipUtils Pro. Only your own voice: the sentence you read says so, with a code that's checked.

Can I use the voice in my YouTube videos?

Yes. The audio you make with the free voices is yours to use in your videos, monetized ones included, and in ads and podcasts. No credit is needed.

Is my text uploaded?

On a computer, no: the AI model runs in your browser, so your text never leaves it (only the model, about 92 MB, is downloaded once from Hugging Face). On phones, your text is sent to our server to be spoken in seconds, and it isn't kept. So is Hindi typed in English letters, to be written in Hindi letters first.

Which languages does it speak?

Ten: English (American and British voices), Hindi, Spanish, Portuguese (Brazilian voices), French, German, Italian, Arabic, Japanese and Chinese (Mandarin). Pick the language above the voices. German, Arabic, Japanese and Chinese are made on our server, on computers too: 5 free minutes a day in Japanese and Chinese, 1 in German and Arabic, more with Pro. Hindi text is read in Hindi script (देवनागरी), with any English words in it read the English way.

Why does the first time take longer on a computer?

On a computer, the AI voice (about 92 MB) downloads the first time; after that it's kept and starts in seconds. On a phone there's nothing to download: the voice is made on our server in seconds.

How do I make a video with the voice?

After the voice is made, tap "Turn it into a video with captions": it opens in Audio to Video, where you pick a background or your own picture and the words appear on screen as they're spoken.

Popular conversions