Assets

SubSync assets overview

Assets are downloadable packages used by SubSync. Currently, there are three types of assets:

• speech recognition model,
• dictionary,
• self updater.

Assets available on server are listed in assets.json.

Dictionary

Used for translation subtitles when reference is in another language. Languages are encoded in file name as [lang1]-[lang2].dict. It is text file with one phrase per line with translations separated by |. Lines beginning with # are omitted (commented out).

First line contains header information in form #dictionary/[lang1]/[lang2]/[version].

Example:

#dictionary/cze/eng/1.0.0
# Example dictionary file
phrase1|translation1
phrase2|translation2a|translation2b

Speech recognition model

Used by speech recognition engine.

Consist of several files, including acoustic and language model, list of words and JSON description. Description file is named [lang].speech and contains model parameters. This is the configuration for pocketsphinx library. My knowledge about Sphinx models is very limited, I never created one by myself, just using publicly available ones. If you created such model and want to share, please let me know and I will make asset for SubSync from it.

For more information please refer to CMUSphinx docs.