Skip to main content
ponderloop

Text to Speech Reader: Read Any Text Aloud in Your Browser

Paste or type text, choose a voice and press Play: your browser reads it aloud using the speech voices installed on your device. The example passage (46 words) takes about 18 s at normal speed, which this page takes as 150 words per minute at rate 1.0×; the estimate scales with the speed you set. The word being spoken is highlighted where the browser reports it, and the sound comes from your browser’s speech engine, so it cannot be saved as an MP3 file here.

Updated Runs in your browser: nothing you enter is uploadedHow this tool works

This page never uploads or saves your text. Your browser’s speech engine speaks it; network voices (marked in the list) may send it to the browser maker’s service.

Text to read

46 words · about 18 s at 1.0× (150 words per minute)

Voice and speed

1.0×
1.0
100%

Changes apply from the next sentence. Long text is read one sentence at a time.

Reading panel

Audio cannot be downloaded: the browser plays speech aloud but gives the page no access to the sound, so there is no MP3 or WAV export.

How long text takes to read aloud at each speed

Words ÷ (150 × rate) minutes. 150 words per minute is a typical presentation pace at rate 1.0×; a real voice varies with its language, punctuation and pauses, so treat these as estimates.
RateWords per minute500 words1,000 words5,000 words
0.5×756 min 40 s13 min 20 s1 h 6 min
0.8×1134 min 27 s8 min 53 s44 min 27 s
1.0×1503 min 20 s6 min 40 s33 min 20 s
1.3×1882 min 40 s5 min 20 s26 min 40 s
1.5×2252 min 13 s4 min 27 s22 min 13 s
2.0×3001 min 40 s3 min 20 s16 min 40 s

Voice settings in this reader

The Web Speech API accepts wider ranges (rate 0.1 to 10, pitch 0 to 2, volume 0 to 1); the sliders here stop where speech stays understandable. Voices differ in how much of each setting they honour.
SettingLowestDefaultHighestApplies
Rate (speed)0.5×1.0×2.0×From the next sentence
Pitch012From the next sentence
Volume011From the next sentence

How to have text read aloud in your browser

  1. Enter your text. Type or paste it into the box. The estimate under the box shows the word count and how long it takes to read aloud at the current speed.
  2. Choose a voice. Pick one from the list, grouped by language. Voices marked “network” may send your text to the browser maker’s speech service; the others run on your device and are chosen first.
  3. Set speed, pitch and volume. Speed runs from 0.5× to 2.0×. Changes take effect from the next sentence.
  4. Press Play. Use Pause and Resume to hold your place, and Stop to end. The word being spoken is highlighted in the reading panel when your browser and voice report word positions.

Frequently asked questions

How does the text to speech reader work?
It uses the Web Speech API built into your browser (the speechSynthesis interface). The page hands the browser a sentence at a time and the browser’s own speech engine, with the voices installed on your system or offered by your browser, produces the sound. The page itself contains no voices and does no speech processing.
Can I download the audio as an MP3?
No. The Web Speech API plays speech through the speakers but gives a web page no access to the audio, so this page cannot record or export it. To save speech as a file you need a tool that generates audio itself, or a screen or system audio recorder, which this page does not provide.
Is my text sent anywhere?
This page never uploads or stores your text. Whether the speech engine does depends on the voice. A voice marked “network” is a remote service (its localService flag is false), and your browser may send the text to the browser maker’s service to turn it into audio. A voice without that mark runs on your device. The reader picks a local voice by default whenever one is installed.
Why is the voice list empty at first, or different from another browser?
Browsers load their voices asynchronously, and Chrome fills the list only after the page is open (it announces this with a voiceschanged event, which this page listens for). Each browser and operating system offers its own voices, so the list on a phone, a Mac and a Windows PC will differ.
Why does the page read one sentence at a time?
Chrome and other Chromium browsers are widely reported to stop speaking a single long utterance part-way, often after about 15 seconds, without any error. Splitting the text into sentences of at most 180 characters and starting each from the end of the last avoids it. The example above is 3 sentences. The side effect is that speed, pitch, volume and voice changes apply from the next sentence.
Why is no word highlighted?
Highlighting relies on the boundary event, which reports the position of each word as it is spoken. Not every browser and voice sends it: network voices and some mobile browsers do not. When no word positions arrive, the reader highlights the sentence being spoken instead and says so.
How is the reading time worked out?
Words ÷ (150 × rate) minutes, so the example’s 46 words take 18 s at 1.0× and 9 s at 2.0×. It is an estimate: the actual voice, its language and the pauses at punctuation change the real time.
Does it work for right-to-left languages and other scripts?
The text box and the reading panel take their direction from the first letter you type, so Hebrew and Arabic read right to left. Whether the text is spoken correctly depends on the voice: pick one whose language matches the text.
What if my browser has no speech synthesis?
The controls are replaced by a notice and nothing is spoken. The word count and reading-time estimate keep working. Current versions of Chrome, Edge, Safari and Firefox support speech synthesis; some in-app browsers do not.

How it works

Speaking uses the SpeechSynthesis interface of the Web Speech API: each sentence is a SpeechSynthesisUtterance with the chosen voice, rate, pitch and volume, spoken with speechSynthesis.speak(), and the next one starts from the previous utterance’s end event. The text is split on line breaks and sentence ends (Intl.Segmenter where available), and any sentence longer than 180 characters is cut at the last space or comma before that length. The voice list comes from speechSynthesis.getVoices() and is rebuilt on every voiceschanged event; voices are grouped by the primary language subtag and sorted local-first. The first selection is a local voice (localService is true) matching your browser languages, then the browser’s default voice; a network voice is chosen by default only when the device has no local voice at all. The highlight uses the boundary event’s charIndex (and charLength, where given), shifted by the sentence’s offset into your text, and is drawn with text nodes and a mark element, never as HTML. Reading time is the word count divided by 150 words per minute multiplied by the rate; 150 is a typical presentation pace, not a property of any voice. Pause, Resume and Stop call speechSynthesis.pause(), resume() and cancel(). Nothing is recorded: the API does not expose audio to the page. Everything runs in your browser and your text is never sent by this page.

Sources

All text & filesAll toolsHome & DIY calculators