Text-to-speech has quietly become very good. Modern "natural" and "enhanced" voices are far easier to listen to than the robotic voices of a few years ago, and they are useful well beyond accessibility: listening to long articles, checking pronunciation in a language you are learning, or proofreading your own writing by ear. Most people never change the default voice, though, and many do not know that extra languages must be installed separately.
How voices reach your browser
Websites and browser extensions usually speak through the Web Speech API, a standard browser feature. When a page asks the browser to speak, the browser offers a list of voices from two sources:
- Voices installed in your operating system. These work offline, and adding more of them (as described below) makes them available to every website and app.
- Voices provided by the browser itself. Chrome includes a set of online "Google" voices, and Microsoft Edge includes online "Natural" voices. These typically require an internet connection.
So if a reading tool only offers one robotic voice for your language, the fix is usually to install a better voice in your system settings and then restart the browser.
Windows 11
Add a voice for another language
- Open Settings → Time & language → Speech.
- Under Manage voices, select Add voices.
- Search for the language, tick it and select Add.
You can also install a full language pack under Time & language → Language & region; make sure the speech option is included.
Natural voices
Windows 11 also offers higher-quality natural voices, originally for its Narrator screen reader. They are managed under Settings → Accessibility → Narrator, where you can add natural voices and download them for offline use. Depending on your Windows version, apps may or may not be able to use them outside Narrator.
After installing voices, close and reopen your browser so it picks up the new list.
macOS
- Open System Settings → Accessibility → Spoken Content.
- Next to System voice, open the menu and choose Manage Voices.
- Browse by language and download the voices you want. Voices marked Enhanced or Premium are larger downloads but sound considerably more natural.
Once downloaded, the voices are available to the system and to browsers such as Safari and Chrome. You can preview each voice before downloading it.
Android
Most Android phones use Google's speech engine, although some manufacturers install their own.
- Open Settings and search for Text-to-speech output (usually under Accessibility, or under System → Languages).
- Choose your preferred engine, usually Speech Recognition & Synthesis from Google or Speech Services by Google.
- Tap the settings icon next to the engine, then Install voice data, and download the languages you need. Many languages offer several voice variants to choose from.
The exact menu names vary between manufacturers and Android versions, so searching the Settings app is usually the fastest route.
iPhone and iPad
Go to Settings → Accessibility → Spoken Content → Voices, pick a language and download an Enhanced or Premium voice. Turning on Speak Selection in the same menu adds a Speak button whenever you select text.
Getting the best results
- Match the voice to the language. An English voice reading French text sounds terrible. Tools that pick the voice automatically rely on the page's language setting, so choose the voice manually if the page is labelled incorrectly.
- Adjust the speed. Most people listen comfortably at 1.1x–1.4x once they are used to a voice; for language learning, slower is better.
- Prefer offline voices for privacy. Online voices send the text to a server to be spoken. Installed system voices process everything on your device.
- Restart after installing. Browsers usually load the voice list at startup.
Troubleshooting
- The new voice does not appear in the browser: fully quit and restart the browser (on Windows, check that no browser processes are still running in the background).
- Speech stops after a few sentences: some browser voices limit the length of a single request; reading tools usually split text into smaller chunks to avoid this.
- No sound at all: check that the tab is not muted (right-click the tab) and that the system output device is correct.
The bottom line
Five minutes in your system settings can replace a flat robotic voice with one that is genuinely pleasant to listen to, in every language you need. Install natural or enhanced voices, restart your browser, and every reading tool that uses the Web Speech API will benefit.