Text to speech: turn any text, link or document into an MP3
Long reports, articles and study notes are easier to get through when you can listen on a walk or in the car. Paste the text, a link or a file, and this prompt has your todo.is agent read it aloud in a natural voice, clean out things that sound odd when spoken, and send you an MP3.
The prompt
- Read this aloud as an MP3: [TEXT, LINK OR ATTACHED FILE]. Language: [LANGUAGE]. Voice: [VOICE]. Speed: [SPEED]. Before recording, make it sound natural when spoken: skip menus, ads, footnote numbers and URLs, read headings as short announcements, and spell out abbreviations. Say these words like this: [PRONUNCIATIONS]. If it's longer than 15 minutes, split it into parts with clear names. Send the MP3 to me here.
What to change
- [TEXT, LINK OR ATTACHED FILE]: Paste the text, a link to an article, or "the attached PDF" (Word, PDF and PowerPoint work).
- [LANGUAGE]: E.g. "English", "French", "Arabic". It can also translate first: "translate to German, then read it".
- [VOICE]: E.g. "warm female, British accent", "deep calm male", "neutral newsreader".
- [SPEED]: "normal", "a little slower, I'm learning the language", or "1.2x".
- [PRONUNCIATIONS]: Names or terms, e.g. "Siobhan = shi-VAWN, SQL = sequel". Or write "none".
Example result
- Audio: "The quiet cost of remote meetings" (article link)
- English · warm female voice, British accent · normal speed · about 12 minutes
- File delivered
- • remote_work_article_audio.mp3: 12 min 08 s, 128 kbps, 11 MB
- What your agent changed so it sounds right
- • Skipped the cookie banner, newsletter sign-up box, "related articles" list and 9 inline links
- • Removed footnote numbers like [3] and read the one useful footnote at the end of its section
- • Read headings as short announcements: "Part two. Why video calls tire us out."
- • Expanded abbreviations: "e.g." became "for example", "approx." became "about", "Q3" became "the third quarter"
- • Read the table as sentences: "Teams with no-meeting Fridays reported fewer late-night emails."
- • Said "Siobhan" as shi-VAWN, as you asked
- Chapter markers (in the order you'll hear them)
- • 00:00 Introduction
- • 02:10 Why video calls tire us out
- • 05:45 The meetings nobody needed
- • 09:20 What three teams changed
- • 11:30 The takeaway
- Other ways to use it
- • Study notes: "read my attached biology notes slowly, and pause after each definition"
- • Language practice: "read this Spanish text at 80% speed, then again at normal speed"
- • Every morning: "at 7:30 each weekday, read me the top 3 stories from these two sites as one MP3 on Telegram"
- • Accessibility: listen to long emails or letters instead of reading on a small screen
How to do it with todo.is
- Copy the prompt and paste in your text or link, or attach the document.
- Pick the language, voice and speed, and add any names it should say a certain way.
- Send it from the Today screen in todo.is or to your agent on WhatsApp or Telegram.
- Download the MP3 or play it straight from the chat. Ask for another voice, slower speed or a new part any time.
Tips for a better result
- Give pronunciations for names, brands and acronyms up front. They are the most common thing a voice gets wrong.
- For study, ask for a slower speed and a short pause after each heading so you can take notes.
- For very long documents, ask for one MP3 per chapter. It's easier to pick up where you left off.
- Make it recurring: a daily briefing read aloud at the same time each morning becomes a private podcast.
text to speech: FAQ
- Does AI text to speech sound natural? Modern AI voices sound close to a human reader, with natural pauses and intonation. Unusual names and abbreviations are the main weak spot, so add pronunciation notes for them.
- Which languages can it read? Many languages, including English, Spanish, French, German, Portuguese and Arabic. It can also translate your text first and then read the translation.
- Can it read a PDF or a web page aloud? Yes. Attach a PDF, Word or PowerPoint file, or paste a public link, and your agent pulls out the main text before recording. Scanned PDFs are read with OCR first.
- Can I use the audio in my videos? For voice-overs, check the terms of the voice service for commercial use. For timed narration that fits a video, the AI voice-over generator prompt works better.
JavaScript is required to use the todo.is app.