Hi AlexData-Hawkhill,
Yeah, the GPU enabled version is much faster. Glad you got this update.
In terms of the translation, it actually doesn't work exactly like that. The model itself basically only does transcription to the same language. The model just has a "bonus" feature that allows direct translation from audio to English, no intermediary text to translate.
Having a full translation from any language to any other language is outside the scope for this app. If there's interest, I could publish an independent app that does full translation of subtitles and text from any language to any other one, which would complement this app.