This project is an Anki addon for pronunciation assessment. You can record your own voice when studying a flashcard and get an overview of what your pronunciation looks like!

When you are reviewing a flashcard and press Ctrl + W AnkiPA starts to record your voice and then send the data to Azure Speech Services to get an pronunciation assessment based on their service.

For Mac the command will be Cmd + W. It can be changed in the settings.

Open Anki and go to Tools -> Add-ons -> Get Add-ons... -> Place the code below and then click OK.

      86363097

See more details at AnkiWeb page. You can also clone this repository to your Anki addons folder if you prefer.

To use this addon you will need an Azure API key. You can create an free account at Azure Portal. After logging into your account, go to Speech services and create a new resource, selecting the region of the endpoint. Once created the resource you will be able to find your API key.

With that informations you need to open your Anki and go to Tools and then AnkiPA Settings. A window will appear and there you place your key and select the region and also choose the language that you will be evaluating your pronunciation.

In the settings besides the API configs you will find some other options:

• Card fields: a comma separated list of the fields to be used as text source for evaluating the pronunciation, prioritazed by order. If none of them be found the first one will be used. The field names are case sensitive.
• Timeout: how many seconds a pronunciation assessment can last
• Phoneme system: only for en-US/GB, it defines how the syllables will be shown
• Shortcut: your preferred shortcut to start recording your voice
• Enable sound effect: sounds based on pronunciation score
• Text extraction method: Choose how AnkiPA extracts text from your cards:
  • Auto (DOM first, then fields): Tries to extract visible text from the card DOM first, falls back to field extraction
  • DOM only: Only uses DOM extraction (useful for cards with dynamic/rotating content)
  • Fields only: Traditional field-based extraction (default behavior)
• CSS selectors for DOM extraction: Comma-separated CSS selectors to find text in the card DOM. Useful for custom card templates. Default: #sentences-inner .fr, .sentence.fr, .fr.sentence, [data-sentence], .example-sentence

AnkiPA relies on Azure Pronunciation Assessment, which supports multiple languages and different accents for English, Spanish and French.

The addon update the supported languages automatically, and the newly added ones will appear on the options when Microsoft releases them.

You can check the current supported languages here.

Also it has an updater that automatically will add new languages options when it's made available by Microsoft.

After clicking on Save button, in few seconds a popup screen will appear showing details about your pronunciation. It shows the percentage scores for accuracy, fluency and overall pronunciation , the amount of errors you committed and what words your pronounced correctly. When you hover your mouse in the words you can see details about the phonemes. The result screen looks like this: