feat(dashboard): per-stage STT/LLM/TTS timing + rename 두뇌 -> LLM

- voice_turn now records three separate timed steps (STT, LLM, TTS) instead of
  one combined "STT+두뇌+TTS" step, so each stage's latency shows in the turn.
- Rename 두뇌 -> LLM everywhere user-facing: component chip, sub header, demo
  banner, model label, brain-failure log, startup print, pipeline step name.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

wsai/__main__.py

@@ -142,7 +142,7 @@ def _run_voice_server(host: str, port: int) -> None:
     monitor.log("info", f"음성 서버 준비 완료 (STT device={sdev}). 디스코드 봇 연결 대기.", cat="READY")
 
     shown = host if host not in ("0.0.0.0", "") else _lan_ip()
-    print(f"\n  음성 서버 준비 완료 (STT device: {sdev}, 두뇌: {brain_name})")
+    print(f"\n  음성 서버 준비 완료 (STT device: {sdev}, LLM: {brain_name})")
     print(f"  대시보드/상태:   http://{shown}:{port}")
     print(f"  봇 연결 엔드포인트: http://127.0.0.1:{port}/api/voice-turn\n")
     try:

wsai/dashboard.py

@@ -650,6 +650,13 @@ class Dashboard:
             self.monitor.log("info", f"LLM 모델 변경: {model} (다음 답변부터 적용)", cat="MODEL")
         return {**self.models_settings()["llm"], "changed": changed}
 
+    @staticmethod
+    def _step(turn, name: str, ms: float) -> None:
+        """Record one finished timing step (STT / LLM / TTS) on a turn."""
+        st = turn.step(name)
+        st.ok, st.ms = True, float(ms)
+        turn._steps.append(st)
+
     def voice_turn(self, audio_bytes: bytes, speaker: str = "",
                    guild: str = "", channel: str = "") -> dict:
         """One Discord voice turn: decode the uploaded utterance, recognise it
@@ -680,6 +687,7 @@ class Dashboard:
                 check=True, capture_output=True,
             )
             heard = (self._submit(self.stt.transcribe(wav)) or "").strip()
+            self._step(turn, "STT", (time.monotonic() - t0) * 1000)  # 인식(+디코드)
             turn.heard(heard or "(빈 결과)")
             if not heard:
                 # Nothing recognised (silence/noise): mark it as [잡음] and skip
@@ -688,16 +696,16 @@ class Dashboard:
                 turn.replied("[잡음]")
                 turn.finish()
                 return {"heard": heard, "reply": "[잡음]", "wav": b""}
+            t_llm = time.monotonic()
             reply_text = self._think(heard)
+            self._step(turn, "LLM" if self.brain else "echo", (time.monotonic() - t_llm) * 1000)
             turn.thought(_thought_summary(reply_text))
             turn.replied(reply_text)
+            t_tts = time.monotonic()
             out_path = self._submit(self.tts.synth(_speech_text(reply_text)))
             with open(out_path, "rb") as f:
                 reply_wav = f.read()
-            ms = int((time.monotonic() - t0) * 1000)
+            self._step(turn, "TTS", (time.monotonic() - t_tts) * 1000)  # 합성
-            step = turn.step("STT+두뇌+TTS" if self.brain else "STT+TTS(GPU)")
-            step.ok, step.ms = True, float(ms)
-            turn._steps.append(step)
             turn.finish()
             try:
                 os.remove(out_path)
@@ -733,7 +741,7 @@ class Dashboard:
                 self.monitor.add_claude_usage(u.get("input", 0), u.get("output", 0))
         except Exception as exc:  # noqa: BLE001
             log.exception("brain failed")
-            self.monitor.log("error", f"두뇌 응답 실패: {exc}", cat="BRAIN")
+            self.monitor.log("error", f"LLM 응답 실패: {exc}", cat="BRAIN")
             blob = f"{getattr(exc, 'status_code', '')} {exc}".lower()
             if "529" in blob or "overload" in blob:
                 # Transient server overload survived the SDK retries.
@@ -1008,7 +1016,7 @@ PAGE = r"""<!DOCTYPE html>
 <header>
   <div>
     <h1>watch_sceen_ai · 실시간 상태</h1>
-    <div class="sub">STT → 두뇌 → TTS 음성 루프를 단계별로 관찰</div>
+    <div class="sub">STT → LLM → TTS 음성 루프를 단계별로 관찰</div>
   </div>
   <div class="pill"><span id="dot" class="dot off"></span><span id="listen">연결 대기</span></div>
   <button id="promptBtn" class="btn hbtn">📝 프롬프트</button>
@@ -1081,7 +1089,7 @@ PAGE = r"""<!DOCTYPE html>
         <span id="mSTTstat" class="sttstat"></span>
       </div>
       <div class="sttrow" style="margin-top:8px">
-        <label style="font-size:12.5px;color:var(--muted)">LLM(두뇌)
+        <label style="font-size:12.5px;color:var(--muted)">LLM
           <select id="mLLM" class="ttssel"></select></label>
         <button id="mLLMapply" class="btn">적용</button>
         <span id="mLLMstat" class="sttstat"></span>
@@ -1194,14 +1202,14 @@ function renderStatus(s){
   const bar = $('demobar');
   if(mockParts.length){
     bar.style.display='block';
-    bar.innerHTML = '⚠ <b>데모 모드</b> — 실제 음성/STT/두뇌/TTS가 아직 연결되지 않아, 아래 대화는 '
+    bar.innerHTML = '⚠ <b>데모 모드</b> — 실제 음성/STT/LLM/TTS가 아직 연결되지 않아, 아래 대화는 '
       + '실제로 들은 내용이 아니라 <b>목(mock) 예시 스크립트</b>입니다. '
       + '실제 엔진(faster-whisper·Claude·MeloTTS)을 붙이면 이 자리에 진짜 발화·지연·오류가 표시됩니다.';
   } else {
     bar.style.display='none';
   }
   const el = $('comp'); el.innerHTML = '';
-  const names = {source:'눈(소스)', vision:'시각', stt:'귀(STT)', brain:'두뇌', tts:'입(TTS)', text:'텍스트'};
+  const names = {source:'눈(소스)', vision:'시각', stt:'귀(STT)', brain:'LLM', tts:'입(TTS)', text:'텍스트'};
   for(const k of Object.keys(names)){
     if(!(k in comps)) continue;
     const v = comps[k];

wsai/pipeline.py

@@ -87,7 +87,7 @@ class Pipeline:
         try:
             async with turn.step("화면 맥락"):
                 screen = await self.context.latest()
-            async with turn.step("두뇌(생각)"):
+            async with turn.step("LLM(생각)"):
                 reply = await self.brain.respond(utt.text, screen, self._history)
             turn.replied(reply.text)
             self._remember(utt.text, reply.text)