Research / Exploration
This is where connected ideas live before they become papers. Each thread is tied to existing work, has ingest tags, and names the next skeleton needed to turn it into a real manuscript or test.
Can speech be modeled as a persistent acoustic world rather than a transient input collapsed into text?
This is the bridge between FAC, codec-token speech models, acoustic verification, and live N'Ko ASR.
Next skeleton
Create a paper skeleton with codec tokens, FAC heads, acoustic verifier supervision, and governed transcript acceptance.
Is intelligence better framed as anticipation governed by evidence rather than prediction alone?
It unifies AGP, trajectory metrics, acoustic verification, and paper absorption into one evidential-governance loop.
Next skeleton
Define formal primitives: evidence, anticipated explanations, governance operators, survival over time, and memory formation.
Could music notation and N'Ko-like sound coding produce a richer representation of pitch, rhythm, tone, and learned sound?
It connects the user's music-note question to FAC and computational choreography without pretending it is already an ASR result.
Next skeleton
Separate notation, acoustic representation, learning task, and cultural writing system before proposing a new music substrate.
Can diffusion-style text generation help revise candidates, papers, or correction proposals after the governed core is proven?
It may be useful for proposal generation, but it should not distract from proof, governance, and live ASR evaluation.
Next skeleton
Hold until the direct proof surfaces are stable, then test it as a proposer, not a truth source.