Text to speech MP3: turn any text into an audio file

Two different things go by the same name. One is having your device read text aloud: Edge, Chrome, Word, iPhone and Android all do it for free, and the sound only exists while the page is open. The other is converting text to an audio file you can keep, send, upload or drop into a video editor.

To hear something once at your desk, the reader you already have is enough, and our text to speech reader guide shows how to use it. To get an MP3, you need a tool that generates the audio and lets you download it.

• You want to listen away from a screen: driving, walking, phone locked.
• You're publishing it: a podcast, a video, a course or an ad.
• You're sending it to someone who doesn't have the text, or loading it into a phone system.
• You'll play it many times and don't want to reopen the same page every time.

Built-in readers are free and fine for listening on the spot. None of them, per their official help pages, lets you save the reading as a file.

Here's how it works in Voizum, right in your browser with nothing to install:

1. 1Open the generator and paste your text, or click Upload document. It accepts .txt, .md, .srt, .pdf, .docx, up to 15 MB, and the text is extracted in your own browser: the file itself isn't uploaded to any server.
2. 2Read through the text that lands in the box before you generate. Strip out what's harmless on paper but awkward out loud: page numbers, repeated headers, footnotes.
3. 3Pick one of the 300+ voices in the library and hit play to hear its sample. The language is detected automatically, but you can set it yourself.
4. 4Open Settings if you want to change speed, volume or the gap between sentences. The voice's play button already uses those settings, so you hear the change without generating anything.
5. 5Click Generate. You see the cost first: 60 credits per 1,000 characters, with a 100-credit minimum per audio.
6. 6When it's ready, play it and click Download. The MP3 is named after the voice, the date and the first words of your text, so it's easy to find in your downloads folder.

Anything you change under Settings is baked into the MP3. The speed control on the history player is separate: speed it up there and download, and the file still plays at the speed it was generated at.

On average, a minute of audio is about 1,000 characters including spaces. The voice moves that number: on the same text, a slow, calm voice can run noticeably longer than a brisk one. The calculator runs it on your own text.

It's a mono MP3 at 96 kbps. It opens in any player or video editor, and for speech that bitrate is plenty.

A single audio can hold up to 600,000 characters, about 10 hours of speech. If you upload a longer document, the text is cut at that limit and you're told so you can check the ending.

Even when a book fits, one MP3 per chapter usually works better:

• Picking up where you left off is easier: each chapter is its own track.
• If you want to change something, you regenerate one chapter, not the whole book.
• Smaller files move more easily between your computer and your phone.
• You can test the voice on chapter one before converting the rest.

If you have lots of short texts rather than one long one, like phone menu prompts or course lessons, the batch tool generates them in one go: up to 100 clips and 1,200,000 characters in total (about 20 hours). Each text becomes its own MP3, and at the end you download them all as a ZIP.

You pay per character with a single 100-credit minimum for the whole batch, not one per clip, so ten one-line prompts don't cost ten minimums. Use one voice for all of them or change it per clip.

For a conversation between two hosts or an interview, dialogue mode puts up to 5 different voices into a single MP3, in turn order.

Every audio shows up in your history, where you can play and download it for 4 days. After that the file is deleted and only its record stays, with the date, voice, length and credits but no sound.

Download it as soon as it's ready and always work from that copy.

Where having the file beats listening on screen:

• Podcasts: an episode narrated from your script, ready to upload. Apple Podcasts accepts MP3 and has its own technical audio requirements; if you export from an editor, set the final format there.
• Lessons and courses: audio lessons students can play anywhere, or the narration for a slide-based video course.
• Accessibility: an MP3 plays in any audio app, with no need to open a page or work a screen reader.
• Voicemail and phone systems: the greeting, business hours or menu options in a clear, consistent voice. The batch tool produces every prompt at once.
• Ads: a radio spot or the voiceover for a product video.
• Studying: your notes turned into audio to review on the commute or at the gym.

Sign in with Google and you get 600 free credits every month, about 10 minutes of audio, to generate and download MP3s with any voice in the library. They reset every month without rolling over, and no card is required.

When you need more, you buy credits with no subscription and they never expire: 60 credits per 1,000 characters, about 60 per minute of audio. You can use the audio you generate commercially too.