Tap the speaker button and hear the word pronounced by a neural voice of the language you are studying. The audio travels with you: the label [sound:…] in the CSV and the mp3s inside the ZIP.
A word you cannot pronounce is half a word learned.
When you have the vocabulary deck on screen, you tap 'Add audio' and the app voices each word with a neural Azure voice of the language you are studying. A speaker button appears next to each flashcard: you tap it and the word plays. The same goes for graded readers, where the audio is generated based on the glossary words.
The voice is not random: it is explicitly set by language in the app's catalog, with a different voice for English and for Spanish. This prevents the classic error of automatic readers, which pronounce an English word with Spanish phonetics or vice versa.
Only the word is voiced, never the example sentence. This is a deliberate choice: what you review on a flashcard is the term, and so the audio is short, played in one go, and takes up little space. There is a limit of 20 audio flashcards per deck and the synthesis occurs in batches of three, so a long deck won't get stuck waiting.
And the audio goes with you from the app. If you export the deck, the CSV carries the reference [sound:…] on the front of each flashcard that has voice, and the ZIP brings a media/ folder with the mp3s ready to copy into your collection.media of Anki: upon import, Anki plays the word without you having to record or download anything on your own.
Use it free without an account. If you want to save your result, you can sign up afterwards.
Save / Sign upNative voice, not a generic reader
the neural voice is set by language in the catalog, so the phonetics are those of the language you are studying.
Only the word, never the sentence
short audios that play with a tap and do not prolong the review.
Speaker button on every card and glossary entry
it plays instantly, within the app, without downloading anything.
The CSV is reconstructed with the [sound
…] tag of each voiced card, which is what Anki reads to play the sound.
Ventaja 5
The ZIP includes the mp3 files in a media/ folder and a README that tells you exactly where to copy them in Anki.
Also in graded readers
the glossary of the reading is voiced word by word with the same voice.
Up to 20 audio cards per deck and synthesis in batches of three
no endless waits or incomplete requests.
If you add audio again, only what is missing is voiced
what has already been generated is not repeated or charged again.
Build your deck or your graded reader, add native voice audio, and take it in the ZIP with the mp3s inside. Audio and downloads come with a free account. Standalone tool, not affiliated with Duolingo, Inc.
In the features that use AI, responses are generated by artificial intelligence and are not a substitute for professional advice.