Text to Speech
Paste text or open a .txt file and have it read aloud. Pick a voice, set the speed, and follow the sentence as it plays. Built-in voices keep your text on this device.
https://flxcode.com/text-to-speech/
Voices on this device only. Your text stays in this browser.
Speed, pitch, volume and voice changes apply from the next sentence.
Press Play to hear the text.
This browser has no speech synthesis, so this page can't read aloud. Current Chrome, Edge, Safari and Firefox have it.
No ads. No account. Nothing you type or record is uploaded. We count page events with a daily code. Check it.
Problem or idea? Email us
Quick start
- Type or paste text, or press Open a .txt file. Up to 20,000 characters.
- Pick a voice. The list shows built-in voices grouped by language.
- Set Speed, Pitch and Volume, then press Play. The sentence being read is highlighted below the buttons.
- Pause to stop for a moment, Resume to carry on, Stop to start over.
Built-in and online voices
The browser offers two kinds of voice. Built-in voices are installed on your device. They work offline and the text never leaves it. Online voices run on a server, so the browser sends them your text a sentence at a time. Chrome and Edge list some of both.
This page starts with built-in voices only. If your browser has online voices, a box appears: "Also list online voices". The label names the company when the voice names say it, for example Google, and says "your browser's maker" when they don't. Tick it and those voices join the list, marked online. Untick it and the page goes back to a built-in voice.
Whatever you pick, this page does not send your text anywhere itself. Only the speech engine of your browser gets it.
How the text is cut into sentences
The page splits your text before it speaks, so that the highlight can follow along and long texts don't stall. The splitter skips full stops that don't end a sentence: titles such as Dr., decimals, version numbers, web addresses and times like 5 p.m. It also starts a new piece at each line break and after the full-width marks 。!? used in Chinese and Japanese. This is the output for a small test text, made by the same code as the tool:
| Piece | Text | Characters |
|---|---|---|
| 1 | Dr. Smith paid $3.14 for 2 pens... | 34 |
| 2 | Then he left. | 13 |
| 3 | "Is it 5 p.m. already?" | 23 |
| 4 | she asked. | 10 |
| 5 | No, it was only noon. | 21 |
| 6 | Version 2.5.1 is out, see www.example.com. | 42 |
A sentence longer than 200 characters is cut at a comma, semicolon or colon, or at a space if there is none. A list of 45 short steps with commas, written as one sentence, becomes 4 pieces of 198, 191, 191, 127 characters. The built-in sample text is 7 pieces.
How long will it take
The estimate under the text box is words divided by 150 words per minute, divided by the speed. Voices differ, so treat it as a guide.
| Text | Words | At 1× | At 1.5× | At 2× |
|---|---|---|---|---|
| A short email | 120 | 48 s | 32 s | 24 s |
| A news article | 800 | 5 min | 4 min | 3 min |
| A long blog post | 2,000 | 13 min | 9 min | 7 min |
| The most this page takes | 3,509 | 23 min | 16 min | 12 min |
Numbers this uses
- 20,000 characters at most: this page's own limit, which keeps the sentence list quick. About 3,509 words.
- 200 characters per piece: our choice. Chrome on desktop is known to cut an utterance after about 15 seconds, and 200 characters is well under that at normal speed.
- 150 words per minute at speed 1: a typical pace for English narration. Used only for the time estimate.
- 5.7 characters per word including the space: typical English prose. Used to turn the character limit into words.
- Speed 0.5 to 2, pitch 0 to 2, volume 0 to 100%: the ranges of the browser's Web Speech API, which the sliders cover (speed is capped at 2 here, the API allows more).
- 6 seconds: if no sound starts within that time, the page reports that the voice isn't working.
What works where
| Browser | Voices | Notes |
|---|---|---|
| Chrome, Edge (desktop) | System voices, plus online voices | Word-by-word highlight where the voice reports word boundaries. Long utterances can cut off, so the page sends sentences. |
| Safari (Mac, iPhone, iPad) | System voices, all built-in | Starts only after a tap. Stops when the screen locks or the app changes. Boundary events vary by voice. |
| Firefox | System voices | Needs a speech engine on the system. On Linux that means installing one. |
| Chrome on Android | System voices, plus online voices | The browser's pause isn't reliable, so Pause stops the sentence and Resume repeats it. |
If a voice has no word-boundary events, the highlight moves by sentence only. This table describes what the browsers document and what users report. We checked the page logic in desktop Chromium with a simulated speech engine, not on every device above.
If nothing plays
- Check the device volume and the Volume slider, and that the tab isn't muted.
- Pick another voice. A voice that is listed but not fully installed can fail silently, and the page then says so after 6 seconds.
- No voices at all: install one in the system settings, or on Linux install a speech package, then reload.
- On a phone, keep the screen on and the tab in front.
- Text with only symbols or emoji has nothing to say, so those lines are skipped.
Limits
- No audio file. A page can't record the browser's speech, so there is no MP3 or WAV download, and no screen-capture workaround.
- No settings are saved. Reload and the sliders go back to normal.
- Voice quality and language depend on your device. The page can't add voices.
- Pause repeats the sentence you were in, because the browser's own pause is unreliable on some systems.
To test the speakers or headphones, try the Speaker Test or the Tone Generator. To check the other direction, the Mic Test shows what your microphone picks up.
Nothing you type is uploaded by this page. We count events without cookies (details).
FAQ
Is my text uploaded anywhere?
Not by us. The page never sends your text to a server of ours, never puts it in the address bar and never stores it. The sound comes from the speech engine in your browser. The one exception is your choice: voices marked online, which are off until you switch them on, send the text to the browser maker to make the sound.
What is the difference between built-in and online voices?
A built-in voice is installed on your device and works offline. An online voice, such as the "Google" voices in Chrome, runs on a server: your browser sends it each sentence and gets sound back. Online voices often sound smoother, but your text leaves the device. This page lists built-in voices only until you tick the box.
Can I download the speech as an MP3 or WAV?
No. A web page can ask the browser to speak, but it can't record what the browser speaks. Tricks that record the whole screen or system sound are unreliable and not something this page does. If you need an audio file, use a desktop text to speech program with a save option.
Why does it stop partway through a long text?
Some browsers, Chrome on desktop in particular, cut an utterance off after about 15 seconds. The page avoids that by handing the browser one sentence at a time, up to 200 characters each, and starting the next when the last one ends. If it still stops, press Play and report which browser and voice you used.
Why are there so few voices, or none?
The list comes from your operating system and browser. Windows, macOS, Android and iOS each ship a different set, and Linux often has none until you install a speech package. Add more voices in the system settings (accessibility or speech), then reload the page. Chrome and Edge also offer online voices, which this page shows only if you ask.
Does it work on iPhone and Android?
Yes, with limits. Reading starts only after you tap Play. It usually stops when you lock the screen or switch apps, because the browser pauses the page. Android Chrome ignores the browser's own pause command, so Pause here cancels the current sentence and Resume says that sentence again from its start.
How do I change the speed and the voice?
Use the Speed slider, from 0.5 to 2 times normal, the Pitch slider and the Volume slider. The voice list is grouped by language. Changes apply from the next sentence, so you hear them within a few seconds. Not every voice supports every pitch, and some ignore it.
Can it read a PDF, a web page or a Word file?
Paste the text in, or open a plain .txt file. For a PDF or Word file, copy the text out first. The page takes up to 20,000 characters, about 3,500 words, and cuts the rest. For a longer text, load it in parts.
More from FLXCode
Tone GeneratorNew
Play any frequency from 1 Hz to 22 kHz. Sine, square, triangle or sawtooth.
Mic TestNew
Check your microphone: level, clipping, a quick recording, and what your browser does to the sound.
Speaker TestNew
Phase, sweep and range. A quick check for speakers and headphones.
Noise GeneratorNew
White, pink, brown and green noise with a sleep timer and fade-out. Plays with the screen off.