← All articles · Guides · about 4 min read
Text to speech: the quiet tool for catching your own mistakes
Read the last paragraph you wrote out loud. Properly out loud, not in your head. You'll hit a sentence that trips you, a word you typed twice, a "the the" you would've sworn wasn't there. Your eyes read what you meant. Your ears read what you actually wrote. That gap is where every typo lives.
Text to speech (TTS) is just a way to hand that second reading to a voice. And it's the most underrated thing on this whole site — because almost everyone reaches for it to "listen to an article" and misses the better use: making the computer read your words back so you can finally hear them.
What it actually is (no jargon)
TTS turns written text into spoken audio. That's the whole idea. You paste words in, it speaks them, or saves them as a file you can play anywhere. The voices used to sound like a robot gargling marbles; the neural ones now sound like a person who's slightly bored but perfectly clear. Good enough for almost everything except fooling your mum on the phone.
The tool on this site gives you two ways to do it, and the choice matters more than people expect:
- Browser Local Read — uses the voices already on your device. Free, instant, and your text never leaves the page. Downside: you can listen, but you can't save a file, because browsers won't let a web page record their own speech engine.
- Cloud TTS (MP3) — sends the text to a speech service (Microsoft neural voices here) and returns an audio file you can preview and download. Better voices, and you get an MP3. Cost: the text has to be sent for that one read, then it's done.
Neither is "the real one." They're two tools for two jobs, and picking wrong just wastes your time.
The use almost nobody talks about: proofreading
Here's the stance worth taking: if you write anything — emails, posts, docs, a product description — TTS is the cheapest editor you'll ever hire. You skim your own draft and your brain autocorrects the mistakes because it knows what you meant. A voice has no idea what you meant. It reads exactly what's on the screen, filler words and all, and suddenly your "very unique" and your 60-word sentence from hell are impossible to ignore.
Try it once. Paste a paragraph you're proud of and listen. Count how many times you wince. That's the edit.
When you'll actually reach for it
Beyond catching your own slips, a few moments come up for most people:
- Long read on the move. A 2,000-word article you'd never sit down for becomes a 12-minute walk. Paste it, play it, go. Local read is perfect here — private, no file needed.
- Learning a language. Listening to a sentence in French or Japanese while reading along trains your ear faster than staring at the text. The reader handles English (US/UK), Korean, Japanese, French, German, Spanish, Chinese, Italian, Portuguese and Russian — pick the language, pick a voice, slow it down if needed.
- Pronunciation practice. Type a tricky word, hear it said right, repeat. Beats guessing from spelling (English, we're looking at you).
- A voiceover you need by tonight. Video intro, course narration, a social clip — type the script, generate the MP3, drop it in. This is the one job that needs Cloud TTS, because you can't edit audio you can't download.
- Tired eyes. Some evenings the screen is just glare. Listening instead of reading is a real option, not a luxury.
What it won't do (be honest with yourself)
TTS reads what's there. It will read your weak argument in a calm, confident voice and make it sound fine. So it's a proofreader of form, not of truth — it'll flag the repeated word, not the bad take. Use it for the first pass, then your brain for the second.
Also: a voice can't fix a 40-minute MP3 that should've been a 3-minute one. Keep the script tight before you generate, or you'll just have a long file of your own rambling.
Your text, and where it goes
People reasonably worry about pasting stuff into a web box. Straight answer for this tool:
- Local read: everything stays on your device. The text is never uploaded, full stop.
- Cloud TTS: the text goes to the speech service to make the audio, and that's it. It's not stored or trained on. (The site runs basic anonymous analytics to check pages work — the full picture is in the privacy policy.)
So a draft email? Local read, no second thought. A confidential contract you'd never email? Listen locally, don't ship it to the cloud voice. Same common sense you'd use with any file.
Quick answers
Can I save the audio?
Only with Cloud TTS — it gives you a downloadable MP3. Local read plays but can't export, because browsers block recording their own voice.
Which voice should I pick?
For proofreading your own writing, any clear voice works; a slightly slower rate (0.9×) makes awkward phrasing easier to hear. For a video, audition two or three and pick the one that doesn't put you to sleep.
Will it read other languages mixed in?
Mixed-language text is supported and read section by section, though a clean single language gives the smoothest result.
Is the MP3 good enough to publish?
For intros, explainers and social clips, yes. For an audiobook you're charging for, you'll want to trim and maybe add music — but the raw voice is a solid start.
So the next time a paragraph feels "off" but you can't say why, stop re-reading it. Paste it into the reader, hit play, and let a voice that doesn't know what you meant find the problem for you.