About our audio

Spoken examples are synthesized speech, not recordings of a person.

Which voice

Most words and sentences, including games and picture flashcards, are spoken by a Microsoft Azure Speech neural voice, Andrew (en-US-AndrewNeural). It speaks General American English. Guided-path conversations also use Cora (en-US-CoraMultilingualNeural) for questions, with Andrew for reply choices. Microsoft's terms for its AI services ask publishers to say when a voice is synthetic, and this page and the footer of every page do that.

Fixed clips, not live speech

Each clip is generated once, saved as a file and played from this site. Nothing you do on the site creates speech, and nothing you type or click is sent to Microsoft.

Checked before it is used

Before publishing a new clip, we listen for the right word, stress and accent. A synthetic voice can still make mistakes. If you hear a word that sounds wrong, that is useful to know. Pronunciation varies between accents, and a different accent is not an error.

Only these disclosed voices

The site uses Andrew and Cora as described above. It never substitutes your browser's built-in voice. If a recording cannot be loaded, the page says so and you can press Play again.