Korean Native Audio for Anki

Fetch Korean pronunciation audio into configurable Anki note fields without locking Anki's interface during network requests.

Features

• Fetch the current editor/reviewer note from the Tools menu.
• Fetch directly from the note editor with the 🔊 Korean button or Ctrl+Shift+K.
• Generate Korean TTS directly from the note editor with the ✨ ElevenLabs button or Ctrl+Shift+E. This uses ElevenLabs Anna Kim with the tested clear-slow 0.78× shadowing preset.
• Fetch multiple selected Browser notes from the right-click menu.
• Configure ordered source-field fallbacks and destination fields per note shape.
• Optionally use official Korean Basic Dictionary (KRDICT) and Forvo API audio with your own keys.
• Use a predictable provider order: Naver → KRDICT → Forvo API → best-effort public Forvo extraction.
• Require an exact dictionary headword match instead of attaching one word from a sentence or a verb's basic form.
• Reject implausibly long Naver pronunciation files, which prevents a mislabeled upstream recording from being attached when its duration cannot fit the word.
• Only change a destination field after audio has been downloaded successfully.
• Run network work in the background and report specific batch skip reasons.

Compatibility

• Anki 2.1.50 or newer
• Windows, macOS, or Linux
• Both Qt 5 and Qt 6 Anki builds

The add-on uses Python's standard library and Anki's bundled APIs; it has no separately installed Python dependencies.

Configuration

The default mapping reads Korean and writes a [sound:...] tag to Sound. Open Tools → Korean Native Audio Settings… to use the native settings window. Tools → Add-ons → Korean Native Audio → Config opens the same window; neither route requires editing JSON.

Each mapping row contains:

• Read Korean from: one or more note field names, separated by commas. The first non-empty field is used.
• Save audio to: the note field that will receive the [sound:...] tag.

Rows are tried from top to bottom. Add, remove, or reorder rows to support different note types. New audio is appended when the destination already has content, so existing audio or notes are preserved. The same sound tag will not be added twice. Choose Replace existing content only when replacement is intentional.

The same window accepts an optional 32-character KRDICT Open API key, an optional Forvo API key, and an ElevenLabs API key. The ElevenLabs key is needed only for the dedicated Anna button. Keys are stored only in the local Anki add-on settings and are hidden in the window unless Show API keys is checked. The packaged defaults never contain a key. A regular or Premium Forvo website login is separate from Forvo API access; the add-on does not store a Forvo username or password.

Usage

• Note editor: click 🔊 Korean or press Ctrl+Shift+K
• ElevenLabs: click ✨ ElevenLabs or press Ctrl+Shift+E
• Current reviewer/editor note: Tools → Fetch Korean Audio for Current Note
• Current reviewer/editor note with Anna: Tools → Generate ElevenLabs Korean Audio for Current Note
• Configuration: Tools → Korean Native Audio Settings…
• Selected notes: open Browse, select notes, then right-click and choose the dictionary-fetch or ElevenLabs-generation action.

If all enabled automatic sources miss on a single note, the add-on offers to either generate Anna audio with ElevenLabs or open the Korean Forvo page for manual download. Forvo's page markup can change, so its automatic extractor is intentionally treated as a fallback.

The regular 🔊 Korean action is dictionary audio: surrounding punctuation such as ? is ignored for lookup, but full sentences and conjugated forms are skipped when the source only provides a different headword. The dedicated ✨ ElevenLabs action is text-to-speech and accepts both words and complete sentences. It uses eleven_multilingual_v2, Anna Kim voice ID uyVNoMrnUku1dZyVEXwD, speed 0.78, stability 0.90, similarity 0.90, and light pauses at long clause boundaries.

Your card template must include the configured destination field, such as {{Sound}}, for the audio to play.