v2.3.4
New
Subtitle regex filter (#8). Strip noise from subtitle lines before mining. Drop speaker tags like (Tanaka), sound cues like [door slams], music markers ♪♬, or Speaker: prefixes. Four one-click presets in Word Filtering settings; stack them with |. Pattern, replacement, and toggle persist across runs. Replacement uses Python backreferences (\1, \2), not asbplayer's $1 syntax.
ExpressionReading and SentenceReading fields (#7). Plain hiragana readings of the expression and sentence.
Fixes
Frequency column sorts numerically in Word Curator (#6). Unranked rows (-) cluster at the bottom.
Other
"Optional Card Fields" section renamed to "Auxiliary Data Fields". Helper text now names the files (pitch_accent.csv, frequency.csv) you need in ~/.anki_miner/ for those fields to populate.
Full Changelog: v2.3.3...v2.3.4