What
A new export mode: per-character "sides" (the actor's scenes + cue lines) where Malayalam dialogue gets a phonetic Latin column alongside the original Malayalam, so a non-Malayalam-speaking crew member can read the lines aloud with correct pronunciation.
Three output formats:
- IPA:
/ɲaːn poːkunnu/ — for foreign DOPs, language coaches, dub artists.
- Mozhi-Latin:
njaan pokunnu — readable by anyone who's seen Mozhi-typed Malayalam.
- Simplified phonetic:
nyaan po-kun-nu (with stress + syllable hyphenation) — for actors who want a non-IPA reading guide.
Why this matters
Real production scenarios for Malayalam films:
- Foreign DOP needs to follow the rhythm of dialogue without reading Malayalam.
- Dub artist for the Tamil/Telugu version needs to lip-sync with phonetic accuracy.
- Festival-circuit script reader (foreign producer) wants to say the lines aloud.
- Sound designer / boom operator anticipating phonetic stress for mic placement.
No screenwriting tool ships this. We have the unique infrastructure (Mozhi engine + mlphon's phoneme transcription) to do it cleanly.
Dependency
mlphon (https://gitlab.com/smc/mlphon) for the phoneme stream. mlphon is an HFST-based phonological analyzer — same FST runtime as mlmorph, so once we have the SFST/HFST runtime in Rust, mlphon comes along for the ride.
Technical sketch
- New ExportModal section: "Actor sides" — pick one or more characters.
- For each selected character: extract their Character cues + adjacent Dialogue + N lines of context.
- Run mlphon on every Malayalam dialogue line, render the phoneme stream as IPA / Mozhi-Latin / simplified.
- Layout: two-column PDF, Malayalam left, phonetic right.
Out of scope
Subtitle export (a different problem — needs translation, not transliteration). See #11 in the v1.0+ planning conversation; that's not on the roadmap.
What
A new export mode: per-character "sides" (the actor's scenes + cue lines) where Malayalam dialogue gets a phonetic Latin column alongside the original Malayalam, so a non-Malayalam-speaking crew member can read the lines aloud with correct pronunciation.
Three output formats:
/ɲaːn poːkunnu/— for foreign DOPs, language coaches, dub artists.njaan pokunnu— readable by anyone who's seen Mozhi-typed Malayalam.nyaan po-kun-nu(with stress + syllable hyphenation) — for actors who want a non-IPA reading guide.Why this matters
Real production scenarios for Malayalam films:
No screenwriting tool ships this. We have the unique infrastructure (Mozhi engine + mlphon's phoneme transcription) to do it cleanly.
Dependency
mlphon (https://gitlab.com/smc/mlphon) for the phoneme stream. mlphon is an HFST-based phonological analyzer — same FST runtime as mlmorph, so once we have the SFST/HFST runtime in Rust, mlphon comes along for the ride.
Technical sketch
Out of scope
Subtitle export (a different problem — needs translation, not transliteration). See #11 in the v1.0+ planning conversation; that's not on the roadmap.