fix: 음성통화 STT 환각 완화 + 툴콜 턴 TTS 누락 수정 + 볼륨 슬라이더

- voice_engine.py: condition_on_previous_text=False로 whisper가 애매한
  구간에서 그럴듯한 문장을 지어내는 환각 완화
- app.js/voice-call.js: 툴 사용 턴(웹검색 등)에서 finalAnswer가 partialContent를
  통째로 덮어쓰면서 음성통화의 spokenUpTo 오프셋이 무효화돼 텍스트는 나오는데
  음성이 안 나오는 버그 수정
- index.html/voice-call.js: 통화 중 GainNode 기반 인앱 볼륨 슬라이더 추가
  (일부 모바일 브라우저에서 마이크 사용 중엔 하드웨어 볼륨 버튼이 통화
  오디오에 반영 안 되는 문제 우회)

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

src/tools/voice_engine.py

@@ -119,7 +119,9 @@ class Engine:
             f.write(audio_bytes)
             path = f.name
         try:
-            segments, info = self.stt_model.transcribe(path, language=language or None, beam_size=5, vad_filter=True)
+            segments, info = self.stt_model.transcribe(
+                path, language=language or None, beam_size=5, vad_filter=True, condition_on_previous_text=False,
+            )
             text = ''.join(seg.text for seg in segments).strip()
             return text, (info.language if info else language)
         finally:
@@ -140,7 +142,9 @@ class Engine:
     # ── Streaming STT (partial decode of an accumulating PCM buffer) ──────
     def transcribe_pcm(self, pcm: bytes, language: str):
         audio = pcm16_bytes_to_float(pcm)
-        segments, info = self.stt_model.transcribe(audio, language=language or None, beam_size=1, vad_filter=True)
+        segments, info = self.stt_model.transcribe(
+            audio, language=language or None, beam_size=1, vad_filter=True, condition_on_previous_text=False,
+        )
         text = ''.join(seg.text for seg in segments).strip()
         return text, (info.language if info else language)
 

web-ui/app.js

@@ -1561,6 +1561,7 @@ async function sendChat(queuedMessage = null) {
             const token = event.text || event.content || event.delta || '';
             if (token) {
               partialContent += token;
+              if (window._voiceCallOnToken) window._voiceCallOnToken(partialContent);
             }
             break;
           }
@@ -1813,6 +1814,11 @@ async function sendChat(queuedMessage = null) {
             if (s.finalAnswer) {
               addProcessEntry('final', s.finalAnswer);
               partialContent = s.finalAnswer; // track for stop
+              // partialContent just got replaced wholesale (not appended to), so
+              // any voice-call spokenUpTo offset computed against the old string
+              // is now meaningless — reset before feeding the new text through.
+              if (window._voiceCallResetSpoken) window._voiceCallResetSpoken();
+              if (window._voiceCallOnToken) window._voiceCallOnToken(partialContent);
             }
             break;
           }
@@ -1903,6 +1909,8 @@ async function sendChat(queuedMessage = null) {
       }
     }
 
+    if (window._voiceCallOnTurnDone) window._voiceCallOnTurnDone(finalReply || partialContent);
+
     if (finalReply) {
       // Append file cards for any attachments sent via the 'files' SSE event
       if (turnFileLinks.length > 0) {

web-ui/index.html

@@ -387,10 +387,18 @@
 	      </div>
 	      <!-- 첨부 파일 카드 영역 (전송 전 표시) -->
       <div id="staged-attachments" style="display:none;flex-wrap:wrap;gap:8px;padding:8px 12px 0;border-top:1px solid var(--line)"></div>
+      <div id="voice-call-bar" style="display:none;align-items:center;gap:8px;padding:8px 12px;border-top:1px solid var(--line);font-size:13px;color:var(--text-dim,#888)">
+        <span id="voice-call-status">통화 연결 중…</span>
+        <span id="voice-call-caption" style="flex:1;opacity:0.85;font-style:italic"></span>
+        <span aria-hidden="true">🔈</span>
+        <input type="range" id="voice-call-volume" min="0" max="1.5" step="0.05" value="1" oninput="voiceCallSetVolume(this.value)" title="통화 음량" style="width:70px" />
+        <button class="send-btn" onclick="voiceCallHangup()" title="통화 종료" style="font-size:14px;width:32px;height:32px;background:#c0392b" aria-label="통화 종료">✕</button>
+      </div>
       <div class="chat-input-row">
         <input type="file" id="file-upload" accept="image/*,.pdf,.txt,.md,.csv,.xlsx,.xls,.docx,.pptx,.mid,.midi" style="display:none" onchange="handleFileUpload(this)" />
         <button class="send-btn" onclick="document.getElementById('file-upload').click()" title="파일 첨부 (이미지·PDF·Excel·txt 등)" style="font-size:18px;width:44px;height:44px" aria-label="파일 첨부">📎</button>
         <button class="send-btn" id="mic-btn" onclick="toggleMicRecording()" title="음성 입력 (클릭: 시작/중지)" style="font-size:18px;width:44px;height:44px" aria-label="음성 입력">🎤</button>
+        <button class="send-btn" id="voice-call-btn" onclick="toggleVoiceCall()" title="실시간 음성 대화 (연속 대화 모드)" style="font-size:18px;width:44px;height:44px" aria-label="실시간 음성 대화">📞</button>
         <textarea
           id="chat-input"
           placeholder="메시지를 입력하세요... (Enter로 전송, Shift+Enter로 줄바꿈)"
@@ -1905,6 +1913,15 @@
       <div id="cv-sessions-list" style="display:flex;flex-direction:column;gap:10px">
         <div style="color:var(--muted);font-size:13px;text-align:center;padding:20px">로딩 중...</div>
       </div>
+
+      <div style="display:flex;align-items:center;justify-content:space-between;margin:20px 0 6px">
+        <div class="right-section-title">fail2ban 차단 IP</div>
+        <button class="btn btn-sm" style="background:var(--panel-2);border:1px solid var(--line);color:var(--muted)" onclick="loadBannedIpsTab()">새로고침</button>
+      </div>
+      <div id="cv-banned-ips-stat" style="font-size:11px;color:var(--muted);margin-bottom:8px"></div>
+      <div id="cv-banned-ips-list" style="display:flex;flex-direction:column;gap:6px">
+        <div style="color:var(--muted);font-size:13px;text-align:center;padding:20px">로딩 중...</div>
+      </div>
     </div>
 
 <div style="display:flex;justify-content:flex-end;gap:8px;padding:12px 16px;border-top:1px solid var(--line)">
@@ -1914,6 +1931,7 @@
   </div>
 </div>
 
+<script src="voice-call.js"></script>
 <script src="app.js"></script>
 
 <!-- Code folder browser dropdown (fixed, rendered outside any overflow:hidden container) -->

web-ui/voice-call.js

@@ -19,7 +19,7 @@
   let ws = null;
   let captureCtx = null, playCtx = null;
   let micStream = null;
-  let captureNode = null, playerNode = null;
+  let captureNode = null, playerNode = null, gainNode = null;
   let state = 'idle'; // idle | listening | user_speaking | processing | assistant_speaking
   let speechChunks = 0;
   let silenceMs = 0;
@@ -62,6 +62,13 @@
     else startVoiceCall();
   };
 
+  window.voiceCallSetVolume = function (v) {
+    const vol = parseFloat(v);
+    if (!Number.isFinite(vol)) return;
+    localStorage.setItem('voiceCallVolume', String(vol));
+    if (gainNode) gainNode.gain.value = vol;
+  };
+
   window.voiceCallHangup = function () {
     active = false;
     state = 'idle';
@@ -70,7 +77,7 @@
     try { micStream && micStream.getTracks().forEach(t => t.stop()); } catch {}
     try { captureCtx && captureCtx.close(); } catch {}
     try { playCtx && playCtx.close(); } catch {}
-    captureCtx = playCtx = micStream = captureNode = playerNode = null;
+    captureCtx = playCtx = micStream = captureNode = playerNode = gainNode = null;
     try { wakeLock && wakeLock.release(); } catch {}
     wakeLock = null;
     ttsQueue = []; ttsBusy = false; currentTtsId = null;
@@ -110,7 +117,16 @@
     playCtx = new (window.AudioContext || window.webkitAudioContext)({ sampleRate: 24000 });
     await playCtx.audioWorklet.addModule('voice-call-worklets.js');
     playerNode = new AudioWorkletNode(playCtx, 'pcm-player-processor');
-    playerNode.connect(playCtx.destination);
+    // A GainNode we control from the in-app slider — on some mobile browsers
+    // the hardware volume rocker maps to the call/mic audio session while a
+    // getUserMedia stream is open, not to this Web Audio output, so the phone's
+    // own volume slider doesn't reliably affect playback here.
+    gainNode = playCtx.createGain();
+    const savedVolume = parseFloat(localStorage.getItem('voiceCallVolume') || '1');
+    gainNode.gain.value = savedVolume;
+    const volumeSlider = document.getElementById('voice-call-volume');
+    if (volumeSlider) volumeSlider.value = String(savedVolume);
+    playerNode.connect(gainNode).connect(playCtx.destination);
 
     const proto = location.protocol === 'https:' ? 'wss' : 'ws';
     const token = getAuthToken();
@@ -298,6 +314,13 @@
     return { sentences, consumedLength: start };
   }
 
+  // Hooked from app.js when partialContent is replaced wholesale rather than
+  // appended to (e.g. a multi-step tool turn's finalAnswer) — spokenUpTo was an
+  // offset into the old string and must be dropped before the new text arrives.
+  window._voiceCallResetSpoken = function () {
+    spokenUpTo = 0;
+  };
+
   // Hooked from app.js's SSE token handler — only acts while a call is active.
   window._voiceCallOnToken = function (fullPartialContent) {
     if (!active) return;