The most natural-sounding text-to-speech voices on a Mac

1. Apple's Premium and Enhanced voices (free)

1. Open voice settings. Go to System Settings → Accessibility → Read & Speak (called Spoken Content on macOS 14 Sonoma and 15 Sequoia).
2. Open Manage Voices. Click the System voice menu and choose Manage Voices.
3. Download a Premium or Enhanced voice. Find your language, then download a voice marked Premium or Enhanced. They're larger downloads than the default compact voices.
4. Select it. Choose the new voice as the System voice, and click Play Sample to compare.

These voices work offline and in any app that uses the system voices, including Speak Selection and Aloud. Siri's own voices are generally not available to third-party apps.

2. Kokoro: a neural voice that runs on your Mac

Kokoro is an open-weight neural text-to-speech model (82 million parameters, Apache 2.0 licence). Neural voices like it have more natural rhythm and intonation than traditional voices, and because it's small it can run on a Mac rather than a server.

In Aloud, Kokoro is a one-time 327 MB download on Macs with Apple silicon and macOS 15 or later. It offers eight English voices, American and British, works offline, and keeps the word-by-word highlight. Other Macs use Apple voices. Other Mac apps also use Kokoro, for example voice-generation tools aimed at making audio for videos.

3. Cloud voices (ElevenLabs, Cartesia and others)

The most expressive voices today run in the cloud. The trade-offs: your text is sent to the provider, you need an internet connection, and you pay per use, which is why apps built on them use subscriptions and allowances.

Aloud lets you bring your own ElevenLabs or Cartesia API key. The key stays in your Mac's Keychain, text is sent only when you choose one of their voices, and the provider bills you directly.

Which should you use?

Voice preference is personal. Try a Premium voice first; it costs nothing. Kokoro is worth hearing if you listen for long stretches.