Text to Speech

This tool makes your browser read text aloud. It uses the speech synthesis interface built into your browser and operating system (speechSynthesis in the Web Speech API): you pick a voice, adjust the rate and pitch, and press a button to play. Its boundaries first: it can only play the speech. It cannot save it as an MP3 or WAV file, its pause and resume button appears only when the browser supports it, and it has no voices of its own. Which voices you can use and how good they sound depends entirely on your browser, your operating system and the voice packs installed on it.

How to use it

1. Type or paste what you want read into 'Text to read aloud'. When the page opens it contains a short sample sentence that follows the interface language.
2. Choose a sound in the 'Voice' drop-down. Options appear as 'voice name (language code)', such as a voice tagged en-US. The page selects by default the first voice that matches the interface language (Chinese voices in the Chinese interface, English voices in the English interface). Right after opening, the list may say 'Loading…' for a moment.
3. Drag the 'Rate' slider to change speed from 0.5x to 2x in steps of 0.1; the label shows the current value. Drag the 'Pitch' slider to change pitch from 0 to 2 in steps of 0.1. Both default to 1.
4. Press 'Start reading'. If the text is empty or only whitespace, nothing happens. Once speech begins the button changes to 'Read again' and 'Pause' and 'Stop' buttons appear next to it.
5. 'Read again' cancels the current playback and starts over from the beginning. 'Pause' halts the speech where it is and the same button turns into 'Resume', which continues from that point. 'Stop' ends the speech immediately, even when paused.

A worked example

Say you are preparing an English talk and want to hear whether your sentences flow. Paste the script into the box, pick a voice tagged en-US or en-GB, lower the rate to 0.8x to judge the rhythm, then raise it to 1.2x to find long sentences that trip you up. When you reach a spot you want to rewrite, press 'Stop', edit, and press 'Start reading' again. Or, if you are learning Chinese, enter a sentence, choose a zh-CN voice and listen at 0.6x to 0.8x to hear the pronunciation. The tool gives no scores or measurements; it just lets you listen. This example describes a use case and involves no test data.

Supported features and known limits

• Your browser must support speechSynthesis. If it does not, the page hides the controls and shows a message suggesting the latest Chrome, Edge or Safari.
• All voices come from your device. The list differs between systems and browsers and may be empty or arrive late. If there is no Chinese voice, the usual reason is that the system lacks the voice pack, which you can add in the system settings.
• The default voice follows the interface language only; it does not switch to match the language of the text you type. If your text is in a different language from the selected voice (for example English text in the Chinese interface), choose a matching voice yourself, or the result may sound odd or be skipped.
• Browsers sometimes reload the voice list in the background. A voice you chose manually is kept when the list refreshes.
• Whether the pitch slider has any effect depends on the chosen voice; some voices ignore it. The effect of the rate setting also depends on the speech engine.
• Pause/resume relies on the browser's speechSynthesis.pause() and resume(); if the browser lacks them, the 'Pause' button is not shown. Some browsers and voices (certain online voices, for example) handle pausing unreliably, so pausing may not work or resuming may restart from the beginning.
• You cannot export or record audio. Use a dedicated voice-over or recording tool if you need a file.
• In some browsers a very long text may stop partway or behave unreliably, which comes from differences in browser speech engines. Enter long material in sections.
• Whether speech continues when you switch tabs or lock a phone depends on browser policy.
• Some browsers include network-based voices (for example, online voices supplied by browser vendors). The page's own code makes no network request, but whether the voice you pick is an online one is up to the browser, and an online voice may pass your text to the vendor's service for synthesis. If that matters to you, choose a local system voice.

Why does the drop-down keep saying 'Loading…'?

In some browsers the voice list loads asynchronously. The page listens for the browser's update event and fills the list once it is ready. If it stays empty for a long time, the device may have no voices, the browser may restrict the feature, or speech synthesis may be unsupported. Try refreshing, or use another browser or device.

Why do I hear nothing when I press the button?

Check that the text box has content, the system volume is not muted and the voice matches the language of your text. Also make sure the browser tab itself is not muted. Some browsers need a user interaction before audio will play, and pressing the button counts as one. If there is still no sound, try another voice or another browser.

Why does it sound so different on different devices?

The voice is produced by your device's speech engine, not by this website. The same text read on Windows, macOS, Android or iOS uses each system's built-in or vendor-supplied voices, so timbre, phrasing and the handling of ambiguous words all differ.

Can it read mixed Chinese and English text?

That depends on the voice. A Chinese voice meeting English words may pronounce them, spell them out in a Chinese manner or skip them. An English voice usually cannot read Chinese characters at all. For mixed content, split it into parts and use a voice that matches each language.

Privacy and processing

The page itself does not send your text to this site or to any server, and it does not store it; refreshing restores the sample sentence. The speaking is done by the browser's speechSynthesis. As noted above, whether a particular browser voice sends text to its vendor's service is decided by the browser and the voice you choose, outside this page's control. For sensitive material, prefer a local voice.