Help
The questions this tool actually raises.
Why are some accents missing?
Device voices come from your operating system, so which ones you have depends on your OS and browser. Apple devices, Android and Microsoft Edge tend to carry the most. Chrome on Windows is thin on Australian and Indian; Firefox on Linux often has only one robotic voice. American, British and Australian are also covered by our server voices, so everyone gets those three. Indian, Canadian and Irish exist only on your device — the server engine has no model for them, and we will not substitute a neighbouring accent and mislabel it.
How do I install more device voices?
iPhone or iPad: Settings → Accessibility → Spoken Content → Voices → English. Android: Settings → System → Languages & input → Text-to-speech output → Google TTS → Install voice data. macOS: System Settings → Accessibility → Spoken Content → System Voice → Manage Voices. Windows: Settings → Time & language → Speech → Manage voices. Reload this page afterwards.
What's the difference between device voices and server voices?
Device voices are the speech voices already installed in your operating system, reached through the browser. They are instant, work offline, and nothing you type leaves your machine. Server voices are synthesised by a neural engine (Deepgram Aura-2 on Cloudflare Workers AI) and cover American, British and Australian only. Every voice in the list is labelled with its source, so you always know which one you are hearing.
Why does the first play of a server voice take a moment?
A server voice has to be synthesised and downloaded on first use, which takes a second or two. Every play after that is near-instant. Device voices have no such delay because nothing is fetched.
Can I download the audio?
Not for device voices — browsers give no way to capture what the speech engine plays. Server voices are ordinary audio files, so your browser can save them.
Does it work offline?
Device voices marked as offline do. Turn on "Offline voices only" in Settings to hide everything that needs a connection, including all server voices.
How is male and female decided?
Server voices are recorded as one or the other. Device voices are trickier: the browser exposes no gender field at all, so it is inferred from the voice name. When that cannot be determined, the voice is shown with no gender label rather than a guess.
Still stuck?
If a voice is listed but silent, check your device isn't muted and that the page has been allowed to play audio. On iPhone, the first play must follow a tap — the browser blocks speech that starts on its own.