Standard translation AI (2+ years old). Has been used for over 30,000 media files. Ok performance, Prone to pronoun inconsistencies more so than other models.
Our previous best translation AI. Maintains and tries to builds upon vega-preview’s performance by making fewer mistakes, adding more nuance to conversations and limiting pronoun inconsistencies.
Our current best translation AI. A new experimental model with very high performance potential. Still prone to pronoun errors but usually covers more dialogue than lulu. Only supports Japanese input audio/video. Please report any issues to support@subtitlenexus.com.