How browser text to speech works
Every modern browser exposes window.speechSynthesis, part of the Web Speech API. You hand it a SpeechSynthesisUtterance — your text plus a voice, rate, and pitch — and the browser asks the operating system's speech engine to play it. The page never touches raw audio samples; it only starts, pauses, and stops playback. That is why no browser TTS tool can offer a download button: the audio exists only as sound coming out of your speakers.
The voice list arrives asynchronously. On the first call speechSynthesis.getVoices() can return an empty array, and the real list shows up a moment later through the voiceschanged event. This tool listens for that event and repopulates the picker when it fires.
Which voices you get depends on your OS and browser
| Platform | Voices you typically see |
|---|---|
| macOS / iOS | The system voices (Samantha, Daniel, and the enhanced Siri voices). More languages and higher-quality voices can be downloaded in System Settings → Accessibility → Spoken Content. |
| Windows | Microsoft David, Zira, Mark, plus any voices added through Time & Language → Speech. Edge adds a large set of online "natural" voices. |
| Chrome / Edge | The OS voices, plus bundled voices labelled "Google" for major languages. The Google ones are generated in the cloud. |
| Android | Voices from the Google or Samsung text-to-speech engine set as the system default. |
| Linux | Nothing by default. Install speech-dispatcher and a synthesiser such as espeak-ng to get voices. |
Because the list is device-specific, a voice you pick on one machine may not exist on another. There is no universal voice name.
Accessibility uses
Having text read aloud helps people with dyslexia, low vision, or reading fatigue get through articles, email, and study material without staring at a screen. It doubles as a proofreading trick: hearing your own writing read back catches clumsy sentences and missing words that the eye skips. Language learners use it to check pronunciation and pacing, and anyone can turn a long read into something to listen to while commuting or cooking. Lowering the rate and pitch a little tends to make dense technical text easier to follow.
Questions
Can I download the audio as an MP3?
No. The Web Speech API plays speech but exposes no file or stream to record, so no browser TTS page can export audio. Use a server-side service such as Amazon Polly or Google Cloud Text-to-Speech if you need a file — that is out of scope here.
Why are there no voices in Chrome sometimes?
The list loads asynchronously and can start empty until the voiceschanged event fires. On Linux, Chrome has no voices unless speech-dispatcher and a voice package are installed. A reload, or one click on the page before pressing play, usually fixes it.
Which languages are supported?
Whatever your OS and browser ship voices for. Add languages in your system settings, then use the language filter above the voice picker.
Is my text sent anywhere?
For local system voices the text is synthesised on your device and never leaves it. A few voices are cloud-backed — the ones Chrome labels "Google" and some "online" voices — and those send the text to the vendor. Pick a non-online voice to keep everything local.
Why does speech stop after about 15 seconds?
A Chrome bug cuts a single utterance off after roughly 15 seconds. This tool splits long text at sentence boundaries and queues the chunks, so articles play through to the end.