Scripter

Create interactive scripts from recordings for language learning

Create interactive HTML5 pages with a script from a recorded conversation. Powered by Deepgram. Requires API key from Deepgram. Submission for Deepgram+DEV hackathon, 2022.

Yout might want to read the submission post, where you will also find a demo.

Get the API key

• Deepgram (required): Create an account in deepgram.com and get an API key.

Store it in a file named deepgramApiKey in the root folder or pass it directly in the CLI using the --deepgram-api-key argument.

Features

• Tune the voice recognition process with the Deepgram query parameters for transcriptions pre-recorded audio with -P|--param KEY=VALUE arguments.
• Downloads and results are cached to reduce redundant traffic, but you can force it using the -F|--force flag or directly removing the _cache folder.
• All log information outputs through stderr and the search output through stdout (or a file, with the -o|--output-file FILE argument). This makes it easy to redirect and pipe different information safely.
• Use the -H|--host HOST option to set the LibreTranslate host to call for translations. By default, https://libretranslate.de is called, but a local server is recommended.
• Your language of use must be supported by both Deepgram and LibreTranslate. Set the source and target languages respectively with -f|--from LANG and -t|--to LANG.

Quick start

• Install all dependencies:

  pip install -r requirements.txt

• Generate an interactive script from a conversation in Spanish, using English translations.

  python main.py --from es --to en example_audio.ogg -o index.html

Set up a local LibreTranslate server (optional)

# Install the application
pip install libretranslate

# Launch it
# IMPORTANT: use --load-only to avoid loading
# the innecesary models of all supported languages
libretranslate --load-only es,en --disable-web-ui

Now you can use Scripter with the --host http://localhost:5000 argument.

Contributing

• See the CONTRIBUTING file to make a PR.
• ⭐ Star this repository
• Raise an issue .

Contributors

Empty section, for now ;)

Ideas for the future

• Allow usage of output templates.
• Enable options to change which APIs to use for the transcription and translation services.
• Add a demo.

License

This code is licensed under the MIT license, a copy of which is available in the repository.