N'Ko Acoustic Coding — Featural Acoustic Coding (FAC)

Research code and paper for the N'Ko tone-resolution seam: using acoustic evidence, especially F0, to restore tone marks that a toneless N'Ko ASR pipeline cannot recover from text alone. The recognizer emits toneless N'Ko, but N'Ko tone is written in the signal: speaker-relative F0 supplies evidence that a text-only prior cannot see. FAC is the acoustic likelihood side of that estimator; AGP is the governance gate that decides where a correction is safe. | N'Ko slot | Acoustic axis it encodes | Native? | |---|---|---| | Tone marks (7 marks, `U+07EB`–`U+07F1`) | pitch register + contour, with short/long variants | **native** | | Nucleus (7 vowels) | spectral centroid / formant color | **native** | | Onset (consonant manner) | attack transient type | **native** | | Nasal coda | resonant sustain | **native** | | harmonicity / spread / roughness / dynamics | timbre | designed extension | The correction that matters: **falling tone is native**. Unicode names U+07EE as `NKO COMBINING LONG DESCENDING TONE`; no designed pitch mark is needed for falling. Designed FAC extensions are reserved for higher timbral descriptors, not pitch. The measured corpus prior is now generated by code, not copied by hand: 4,139 parsed syllables across 105 entries show roughly 65.8% marked high/low, 33.3% unmarked mid, and 0.9% contour. That makes the first practical correction problem register-first and confidence-gated, not a broad claim that contour dominates all tone.