Text to Speech

Hear your text read aloud by the voices already installed on your device.

How this text to speech tool works

The page hands your text to the speech engine that is already part of your browser and operating system, through the Web Speech API's SpeechSynthesis interface. Nothing is sent to our server. The voice menu lists whatever speechSynthesis.getVoices() returns, which MDN describes as all the voices available on the current device, so the names you see depend on where you are sitting: Microsoft voices on Windows, Apple voices on a Mac or iPhone, Google voices on Android.

Press Play and the text is read from the beginning through your speakers or headphones; Stop cancels it. In our test on Windows 11 with a Chromium browser the menu offered three local Microsoft voices (David, Mark and Zira), all for US English, and playback started within a second. A different machine will show a different list, and that is normal.

What the controls do

The sliders set properties of the SpeechSynthesisUtterance object that the page creates for your text. The tool uses a narrower range than the API allows because the extremes are unpleasant to listen to.

Control In this tool Range the API allows Default
VoiceEvery voice your device reportsAny voice from getVoices()First voice in the list
Speed0.5x to 2x, in steps of 0.10.1 to 10 (rate)1
Pitch0.5 to 2, in steps of 0.10 to 2 (pitch)1

MDN notes that some engines and voices constrain the rate and pitch they accept, so the same slider position can sound different from one voice to the next. Change the sliders before pressing Play; they apply to the next utterance, not the one already speaking.

Why the voice list differs between devices

  • The operating system supplies most voices: they come with its language packs. Installing another language or speech pack in your system settings adds voices here too, after a reload.
  • Some browsers add remote voices: MDN's localService flag separates voices synthesised on your device from those a browser fetches from a remote service. The menu here does not show the flag, but names containing "Online" or "Google" are usually remote.
  • A voice speaks one language: the language code in brackets after each name, such as en-US or de-DE, tells you which. English text read by a German voice comes out with a German accent and odd stress, so match the code to your text.
  • The list can load late: if the menu is empty when the page opens, wait a moment or reload; browsers fill the list asynchronously and the page refreshes it when the voiceschanged event fires.

What this tool cannot do

  • Save an audio file: the API plays speech, it does not hand back MP3 or WAV data. To keep a recording you need a screen or audio recorder, or a cloud speech service that returns a file.
  • Read SSML: the page passes plain text. Tags from the W3C Speech Synthesis Markup Language would be spoken as text, not interpreted.
  • Finish very long passages reliably: some engines stop part-way through a long text. If that happens, paste one section at a time.
  • Fix pronunciation: there is no custom dictionary. Spell a tricky name the way it sounds in the text box instead.

Practical uses

Hearing text catches what reading skims over: a missing word, a sentence that runs out of breath, a name that is spelled right but sounds wrong. Paste a draft email or article and listen once before sending. Language learners can hear a phrase with a voice in the target language, and anyone recording a voiceover can time a script by playing it at speed 1.

Two companion tools fit naturally: the Image to Text (OCR) converter turns a photographed page into text you can paste here and listen to, and the Word Counter tells you how long the passage is before you press Play.

Frequently Asked Questions

Why do available voices differ between browsers?

Each browser exposes the speech engine of the operating system it runs on, and some add remote voices of their own. The same laptop can therefore show different lists in two browsers, and a phone shows yet another. Installing system language packs is the reliable way to add voices.

Can I save the speech as an audio file?

Not with this tool. The Web Speech API renders speech straight to the audio output and offers no method to export it. Use a recorder, or a cloud text to speech service that returns audio files.

Does text to speech work offline?

Yes, with local voices. Once the page has loaded, a voice synthesised on your device needs no connection. Remote voices (MDN's localService is false for them) need the internet because the browser fetches the audio from its vendor's service.

Is my text sent to your server?

No. The page passes the text to the speech engine in your browser and never posts it to tools.edeskcloud.com. If you choose a remote voice, your browser, not this site, sends the text to the service that produces that voice.