RITMOM - Repetition is the mother of memory.

It's for learning languages. Creates audio tracks from:

• Your Google Translate favorites list. Feeds word and phrase pairs from.
• Plain text file with foreign word list. Offline dictionaries are used for translation (WordNet, GoldenDict, Lingoes).

The idea is to use a sequence

• foreign female slow
• native female normal
• example usage phrase female
• foreign male slow
• native female normal
• example usage phrase male

for better memorization.

Text to speech out of a box. It uses 3rd party TTS engines via Windows COM. You have to use some proprietary TTS engine to generate a good quality narration.

1. Install TTS engines
2. You will need ffmpeg executable in your PATH to convert your tracks to MP3.
3. Save phrases to phrases/ dir, both csv or xls will fit:
4. pip install -r requirements.txt
5. Edit config.json file.

foreign and native parameters must contain substring of description, which can be listed with this command: python main.py -l

Download desired text corpuses to be able to generate context usage phrases. Corpuses are listed in a phraseExamples configuration option:

import nltk; nltk.download()

6. Generate audio python main.py

For furigana support you'll need MeCab executable accessible in your PATH.

You'll need to install the requirements pip install -r requirements-jp.txt

Supported dictionary formats:

• DSL (Used by GoldenDict)
• LDX (Used by ABBYY Lingvo and Lingoes)
• WordNet (through NLTK)

• Fix TidyUpText filter
• Download and cache required resources (dictionaries, pre-trained models, etc)
• Use TextChunk instead of plain text
• Extract examples from offline dictionaries
• Add subtitle (.srt) corpus reader and examples
• Do try fuzzy matching if no translation found
• Text: write approximate audio timestamp
• Arbitrary Text Source (book, subtitles, news)
• Generate text output too
• Make a sequence configurable from a config
• Fix wordnet for languages differ from English
• Search for collocations
• Download missing NLTK packages and corpuses automatically if missing
• Fix index before running tasks
• Add synonyms and antonyms
• Use more threads for TTS
• Pickle indices for corpus for quicker start
• Add wordnet: definition (thesaurus, examples, excerpts)
• Update requirements.txt