anki-voiced

Generate Anki flashcard decks with AI-voiced audio.

Quick Start

pip install anki-voiced       # Install the tool
anki-voiced init              # Create sample vocabulary.csv and deck.toml
anki-voiced create vocabulary.csv  # Generate deck with AI audio

Installation

# Using uv (recommended)
uv tool install anki-voiced

# Using pip
pip install anki-voiced

# Using pipx
pipx install anki-voiced

To uninstall:

uv tool uninstall anki-voiced
# or
pip uninstall anki-voiced

System Requirements

• Python 3.10+
• espeak-ng for phonemization:

  # Ubuntu/Debian
  sudo apt install espeak-ng
  
  # macOS
  brew install espeak-ng
  

Japanese Language Support

For Japanese TTS, download the UniDic dictionary after installation:

python -m unidic download

This downloads ~500MB of dictionary data required for Japanese text processing.

Usage

Create a deck from CSV

# Quick single deck
anki-voiced create sentences.csv -o my-deck.apkg --lang japanese --voice male

# With config file for multi-tier decks
anki-voiced create --config deck.toml

# Read from stdin
cat data.csv | anki-voiced create - -o deck.apkg

Initialize a new project

anki-voiced init --lang japanese

Creates:

• vocabulary.csv - Sample vocabulary file
• deck.toml - Configuration file for multi-tier decks

List available voices

anki-voiced voices
anki-voiced voices --lang japanese

Combine multiple decks

anki-voiced join tier1.apkg tier2.apkg -o combined.apkg -n "Master Deck"

Commands

create

Generate an Anki deck with AI-voiced audio from a CSV file.

Usage: anki-voiced create [OPTIONS] DATA_FILE

Options:
  -o, --output PATH       Output .apkg file [default: DATA_FILE.apkg]
  -n, --name TEXT         Deck name shown in Anki [default: filename]
  -l, --lang LANG         Language: english, japanese, french, portuguese
  -v, --voice VOICE       Voice: male, female, or specific voice name
  -t, --template NAME     Card template: basic, double-card, cloze
  -c, --config FILE       TOML config (overrides other options)
      --dry-run           Show what would be generated
      --force             Regenerate audio even if cached
  -q, --quiet             Minimal output (for scripts)
      --json              Machine-readable JSON output

init

Create a new deck project with sample files.

Usage: anki-voiced init [OPTIONS]

Options:
  -l, --lang LANG    Language for sample content [default: english]
  -t, --template     Template to use [default: double-card]

voices

List available voices for text-to-speech.

Usage: anki-voiced voices [OPTIONS]

Options:
  -l, --lang LANG    Filter by language
      --json         Machine-readable JSON output

join

Combine multiple .apkg files into one master deck.

Usage: anki-voiced join [OPTIONS] DECK_FILES...

Options:
  -o, --output PATH    Output .apkg file (required)
  -n, --name TEXT      Master deck name (required)

Templates

basic - Simple Front/Back

• Fields: front, back
• Cards: 1 card (Front + Audio -> Back)
• Use case: Simple vocabulary, phrases

front,back
Hello,Bonjour
Goodbye,Au revoir

double-card - Comprehension + Production

• Fields: sentence, translation, pronunciation, tags
• Cards: 2 cards per entry
  • Comprehension: Audio + Sentence -> Translation
  • Production: Translation + Hint -> Sentence + Audio
• Use case: Language learning with active recall

sentence,translation,pronunciation,tags
The meeting starts at 10 AM.,La réunion commence à 10h.,The meeting starts at 10 AM.,business

cloze - Fill-in-the-blank

• Fields: text (with {{c1::cloze}}), extra
• Cards: Auto-generated from cloze markers
• Use case: Grammar patterns, vocabulary in context

text,extra
I {{c1::like}} apples.,verb: to enjoy
She {{c1::runs}} every day.,verb: to run

Languages & Voices

Language	Code	Default Voice	Available Voices
English	en	af_heart	af_heart, af_bella, af_nicole, af_sarah, af_sky, am_adam, am_michael
Japanese	ja	jm_kumo	jf_alpha, jf_gongitsune, jf_nezumi, jf_tebukuro, jm_kumo
French	fr	ff_siwis	ff_siwis
Portuguese	pt	pf_camila	pf_camila

Configuration

Project Config (deck.toml)

name = "Japanese IT Vocabulary"
language = "japanese"
voice = "jm_kumo"
template = "double-card"
output = "japanese-it-complete.apkg"

[[tiers]]
name = "Tier 1 - Foundations"
data = "tier1.csv"

[[tiers]]
name = "Tier 2 - Development"
data = "tier2.csv"

Environment Variables

ANKI_VOICED_LANG=japanese      # Default language
ANKI_VOICED_VOICE=female       # Default voice
ANKI_VOICED_TEMPLATE=basic     # Default template
NO_COLOR=1                     # Disable colors

Configuration Precedence

1. Command-line flags (highest)
2. Environment variables
3. Project config (deck.toml in current directory)
4. User config (~/.config/anki-voiced/config.toml)
5. Built-in defaults (lowest)

Caching

Audio files are cached in ~/.cache/anki-voiced/ for faster regeneration. To clear the cache:

rm -rf ~/.cache/anki-voiced

Development

# Clone the repo
git clone https://github.com/kakkoidev/anki-voiced
cd anki-voiced

# Install with uv (recommended)
uv sync --all-extras

# Run tests
uv run pytest

# Run locally
uv run anki-voiced --help

# Or with pip
pip install -e ".[dev]"
pytest
python -m anki_voiced --help

License

MIT

Credits

• Kokoro TTS - High-quality text-to-speech
• genanki - Anki deck generation
• lameenc - MP3 encoding
• Typer - CLI framework