fix(frontend): improve responsive transcripts and simplify theme styles

docs/client/architecture.md

@@ -57,7 +57,8 @@ audio. It is not an unconditional byte-for-byte tunnel.
 | [client.ts](../../vocero-frontend/src/realtime/client.ts) | Realtime protocol, session lifecycle, audio and transcript events |
 | [components](../../vocero-frontend/src/components) | Chat, caller profile, settings and voice components |
 | [audio](../../vocero-frontend/src/audio) | TypeScript capture/playback worklets, PCM codec, and visualizer |
-| [legacy-theme.css](../../vocero-frontend/src/styles/legacy-theme.css) | Retained themes and responsive visual styles alongside Tailwind |
+| [legacy-theme.css](../../vocero-frontend/src/styles/legacy-theme.css) | Theme tokens, voice artwork and component appearance |
+| [layout.css](../../vocero-frontend/src/styles/layout.css) | Shared responsive layout, transcript sizing and reduced-motion behavior |
 
 ## Server routes
 
@@ -95,6 +96,12 @@ describe service-owned structure and retain loose upstream values. See
 4. Assistant PCM16 audio at the current 16 kHz protocol rate is decoded and sent to
    [audio-playback.worklet.ts](../../vocero-frontend/src/audio/audio-playback.worklet.ts), which buffers,
    resamples and fades playback edges. Input speech can clear playback for barge-in.
+   The client flushes short final chunks on response completion and remains in
+   the speaking state until the worklet reports that playback has drained.
+   Sequence numbers reject drain notifications from superseded playback.
+
+Microphone meter subscribers are separate from application-state subscribers,
+so level updates do not rerender the main React tree.
 
 Audio worklets are loaded through `AudioWorklet.addModule`; keep their processor
 registration structure intact when editing them. The speech backend's thread
@@ -105,6 +112,8 @@ and cancellation model is documented in [realtime API](../speech/realtime-api.md
 The browser triggers external reasoning on completed user transcripts only in
 `OPENAI` mode. A non-empty answer replaces the native response through the
 external-answer path. These HTTP operations supplement the realtime audio path.
+New user speech and session teardown cancel pending reasoning. Turn identity
+checks reject stale HTTP results and speech callbacks even if cancellation is late.
 
 Read [reasoning](reasoning.md), [context injection](context-injection.md), and
 [tool calling](tool-calling.md) for the conditions and limitations of each path.