• hear word/phrase in context (audio or preferably audio+visual). no use of additional subtitles in any language (incl not looking at JP subtitles)
• replay section until it's possible to catch the target word/phrase, then look it up - first time to see its written form. best to do JP definition rather than JP<->EN translation.
• replay section, now knowing the exact / written word/phrase - able to recognise the single new word(etc) in a long or unclear string of text
• record audio for the section as an anki card (front contains audio only with no text, reverse shows the new vocab with ふりがな and JP definition, + EN translation for just new content and only if required.)
• make use of screenshots of moment/scene from which vocab is derived

note

• importance of recognising(or being shown) word through audio(/+picture) rather than reading it in advance or simultaneously　∴ forces closer concentration and creates more ingrained picture/illustration in the mind. + seeing it in its context from the start is also helpful