Correct release-candidate architecture description

README.md

@@ -47,22 +47,21 @@ The parameter count describes the inference-only text-to-waveform model. It does
 not include optimizer state, discriminators, or training-only posterior paths,
 and it does not hide a separate vocoder.
 
-## Why It Is Not Stock VITS
+## Architecture
 
-Inflect-Micro-v2 uses a compact aligned VITS-derived backbone with several
-small-model-specific changes:
+Inflect-Micro-v2 uses a compact VITS-derived, single-stage text-to-waveform
+backbone configured specifically for the sub-10M regime:
 
-- a phoneme frontend with explicit word and punctuation boundary classes
+- a normalized English phoneme frontend with stress and preserved punctuation
 - a deterministic duration predictor for compact, stable deployment
 - monotonic alignment between phonemes and acoustic frames
-- a hierarchical token-level prosody planner with multiple receptive fields
-- bounded energy and log-F0 intercept/slope prediction plus voiced occupancy
-- prosody conditioning applied to both the latent prior and waveform decoder
-- smoothed continuous controls without smoothing consonant boundaries
+- a residual-coupling latent flow and posterior encoder during training
+- an integrated HiFi-GAN-style waveform decoder; no external vocoder is needed
 - multi-resolution STFT supervision during waveform training
 
-The prosody planner is intended to preserve phrase-level movement without the
-frame-level jitter that repeatedly damaged earlier Inflect experiments.
+The selected checkpoint does not use the separate experimental HierJoint or
+HierNAR prosody-planner branches evaluated during development. Those models are
+not part of this release candidate.
 
 ### Core Configuration
 
@@ -168,4 +167,3 @@ Before this repository becomes public:
 - [ ] publish representative, average, OOD, long, and failure samples
 - [ ] verify installation and inference from a clean machine
 - [ ] finalize model and dataset licensing
-