Skip to content

Latest commit

 

History

History
37 lines (29 loc) · 1.9 KB

File metadata and controls

37 lines (29 loc) · 1.9 KB

Frontend Rules — Implementation Progress

Date: 2026-02-26 Based on: reports/scout-current-frontend-rules.md, reports/scout-paper-inventory-for-rules.md

Progress

Starting state: 15 rules across 4 phases After 4 batches: 42 rules (27 new rules added)

Batch Commit Rules Tests Description
1 a83cf09 8 12 Formant context: F2 locus by place, nasal FNZ, dark /l/, /r/ F3, F1 release
2 5272e0e 7 10 S-cluster aspiration, vowel reduction, nasalization, consonant duration K-factors, fricative floors
3 8c5f650 6 10 F0 microprosody (onset perturbation, continuation rise), palatal F2, vowel shortening, burst templates
4 1b37000 6 11 Nasal assimilation, stop unreleasing, word-medial shortening, F0 question baseline, pitch reset, VCV coartic

Test suite: 257 passed, 1 pre-existing failure

Remaining (13 rules — most need upstream features)

Feasible with effort (~5):

  • t_flapping: model acoustically by adjusting T_CL/T_REL params in V_T_V context
  • td_cluster_deletion: suppress T/D in coda clusters via splice
  • progressive_phrase_final_lengthening: upgrade existing rule to apply progressively
  • polysyllabic_shortening: approximate by counting vowels in word via look_back/ahead
  • stable_transition_duration: insert transition segments (complex structural)

Need upstream features (~8):

  • accent_priority_by_pos: needs POS tags on word tokens
  • f0_downstep: needs tone tracking (H/L sequence state)
  • f0_emphasis_boost: needs focus/emphasis marking
  • word_initial_unstressed_polysyllabic: needs reliable syllable count
  • phrasal_accent_lengthening: needs accent assignment
  • dac_coarticulation_scaling: needs DAC values on consonant inventory
  • t_glottalization: needs glottal stop phoneme in inventory
  • tr_dr_affrication: needs modified affricate phonemes