dsh-voice-mode 0.2.3 → 0.5.0

package/lib/client.js

@@ -42,6 +42,23 @@ module.exports = __toCommonJS(client_exports);
 var React = __toESM(require("react"), 1);
 var import_react2 = require("react");
 
+// src/resample.ts
+function resampleLinear(src, srcRate, dstRate) {
+  if (srcRate === dstRate) return src;
+  if (src.length === 0) return src;
+  const ratio = srcRate / dstRate;
+  const outLen = Math.max(1, Math.floor(src.length / ratio));
+  const out = new Float32Array(outLen);
+  for (let i = 0; i < outLen; i++) {
+    const pos = i * ratio;
+    const i0 = Math.floor(pos);
+    const i1 = Math.min(i0 + 1, src.length - 1);
+    const frac = pos - i0;
+    out[i] = src[i0] + (src[i1] - src[i0]) * frac;
+  }
+  return out;
+}
+
 // src/wakeword.ts
 function normalizeWake(text) {
   return String(text ?? "").replace(/[\s\u3000]+/g, "").toLowerCase().replace(/[，。！？!?；;、,.]/g, "");
@@ -56,21 +73,20 @@ function matchWakeWord(partial, wakeWord) {
 }
 
 // src/asr.ts
+var workletBlobUrl = null;
 var SAMPLE_RATE = 16e3;
 var SPEECH_RMS = 0.015;
 var LEVEL_CEILING = 0.25;
 var MAX_SEGMENT_MS = 3e4;
+var MIN_SPEECH_MS = 250;
 var PRE_PAD_MS = 250;
-var PARTIAL_INTERVAL_MS = 900;
+var PARTIAL_INTERVAL_MS = 100;
 var PARTIAL_MIN_S = 0.4;
 var PARTIAL_MAX_S = 30;
 var BUFFER_SIZE = 1024;
-var INTERRUPT_LEVELS = {
-  0: { rms: 0.1, ms: 500 },
-  1: { rms: 0.06, ms: 400 },
-  2: { rms: 0.035, ms: 300 }
-};
 function createAsrEngine(config, sessionId) {
+  const wakeWord = (config.wakeWord ?? "").trim().toLowerCase().replace(/[\s\u3000]+/g, "");
+  const wakeEnabled = wakeWord !== "" && config.mode !== "hold";
   let state = "idle";
   const stateListeners = /* @__PURE__ */ new Set();
   const errorListeners = /* @__PURE__ */ new Set();
@@ -84,29 +100,45 @@ function createAsrEngine(config, sessionId) {
   };
   const transcriptListeners = /* @__PURE__ */ new Set();
   const partialListeners = /* @__PURE__ */ new Set();
-  const speechStartListeners = /* @__PURE__ */ new Set();
   const levelListeners = /* @__PURE__ */ new Set();
+  const telemetryListeners = /* @__PURE__ */ new Set();
+  let utteranceEndAt = null;
+  const emitTelemetry = (stage) => {
+    const ev = { stage, at: Date.now() };
+    for (const fn of telemetryListeners) {
+      try {
+        fn(ev);
+      } catch {
+      }
+    }
+  };
   let audioCtx = null;
   let stream = null;
   let processor = null;
+  let workletNode = null;
   let active = false;
+  let stopRequested = false;
+  let startSeq = 0;
   let inFlush = false;
   let ctxRate = SAMPLE_RATE;
   let speechActive = false;
   let segment = [];
   let segmentMs = 0;
+  let speechMs = 0;
   let silenceMs = 0;
   let prePad = [];
   let holdActive = false;
-  const wakeWord = (config.wakeWord ?? "").trim().toLowerCase().replace(/[\s\u3000]+/g, "");
-  const intLevel = INTERRUPT_LEVELS[config.interruptLevel] ?? INTERRUPT_LEVELS[0];
-  let interruptCandidateMs = 0;
-  let bargeInDampingUntil = 0;
-  let sincePartialMs = 0;
+  const echo = config.echo;
+  let lastPollAt = 0;
   let partialInFlight = false;
   let segmentEpoch = 0;
   let forcePending = false;
-  const asrUrl = (final) => `${location.origin}${config.basePath.replace(/\/+$/, "")}/asr?sessionId=${encodeURIComponent(sessionId)}&final=${final ? 1 : 0}`;
+  let uploadedSamples = 0;
+  let detectChunks = [];
+  let detectSent = 0;
+  let detectInFlight = false;
+  let detectGeneration = 0;
+  const asrUrl = (final, offset, epoch) => `${location.origin}${config.basePath.replace(/\/+$/, "")}/asr?sessionId=${encodeURIComponent(sessionId)}&final=${final ? 1 : 0}` + (offset !== void 0 ? `&offset=${offset}` : "") + (epoch !== void 0 ? `&epoch=${epoch}` : "");
   const setState = (s) => {
     state = s;
     for (const fn of stateListeners) {
@@ -126,25 +158,36 @@ function createAsrEngine(config, sessionId) {
       }
     }
   };
-  const concatSegment = () => {
-    const n = segment.reduce((acc, c) => acc + c.length, 0);
-    const out = new Float32Array(n);
+  const sliceChunks = (chunks, from) => {
+    let total = 0;
+    for (const c of chunks) total += c.length;
+    const out = new Float32Array(Math.max(0, total - from));
+    if (out.length === 0) return out;
     let off = 0;
-    for (const c of segment) {
-      out.set(c, off);
-      off += c.length;
+    let acc = 0;
+    for (const c of chunks) {
+      if (off >= out.length) break;
+      const sub = c.subarray(Math.max(0, from - acc));
+      const n = Math.min(sub.length, out.length - off);
+      out.set(sub.subarray(0, n), off);
+      off += n;
+      acc += c.length;
     }
     return out;
   };
+  const sliceSince = (from) => sliceChunks(segment, from);
   const requestPartial = async () => {
     if (partialInFlight || segment.length === 0) return;
-    const seconds = segment.reduce((n, c) => n + c.length, 0) / SAMPLE_RATE;
+    const total = segment.reduce((n, c) => n + c.length, 0);
+    const seconds = total / SAMPLE_RATE;
     if (seconds < PARTIAL_MIN_S || seconds > PARTIAL_MAX_S) return;
-    const samples = concatSegment();
+    const from = uploadedSamples;
+    if (total - from <= 0) return;
+    const samples = sliceSince(from);
     const epoch = segmentEpoch;
     partialInFlight = true;
     try {
-      let res = await fetch(asrUrl(false), {
+      let res = await fetch(asrUrl(false, from, epoch), {
         method: "POST",
         headers: { "content-type": "application/octet-stream" },
         body: samples.buffer
@@ -154,7 +197,7 @@ function createAsrEngine(config, sessionId) {
         const retry = await new Promise((resolve) => {
           setTimeout(async () => {
             try {
-              const r2 = await fetch(asrUrl(false), {
+              const r2 = await fetch(asrUrl(false, from, epoch), {
                 method: "POST",
                 headers: { "content-type": "application/octet-stream" },
                 body: samples.buffer
@@ -167,42 +210,113 @@ function createAsrEngine(config, sessionId) {
         });
         res = retry;
       }
+      if (res.status === 403 && config.onSessionExpired) {
+        const recovered = await config.onSessionExpired();
+        if (recovered && epoch === segmentEpoch) {
+          uploadedSamples = 0;
+          try {
+            res = await fetch(asrUrl(false, 0, epoch), {
+              method: "POST",
+              headers: { "content-type": "application/octet-stream" },
+              body: sliceSince(0).buffer
+            });
+          } catch {
+          }
+        }
+      }
       if (epoch !== segmentEpoch) return;
       if (!res.ok) return;
       const out = await res.json();
       if (epoch !== segmentEpoch) return;
-      if (state === "loading-model") setState("speech");
-      if (state === "wake" && wakeWord) {
+      if (state === "wake" && wakeEnabled) {
+        uploadedSamples = Math.max(uploadedSamples, from + samples.length);
         if (matchWakeWord(out.text ?? "", wakeWord)) {
           segmentEpoch++;
           segment = [];
           segmentMs = 0;
+          speechMs = 0;
           silenceMs = 0;
           prePad = [];
-          sincePartialMs = 0;
+          uploadedSamples = 0;
+          utteranceEndAt = null;
           await resetHostStream();
           if (active) setState("listening");
         }
         return;
       }
+      if (out.isSpeech !== void 0) config.onIsSpeech?.(out.isSpeech);
+      if (state === "loading-model") setState("speech");
+      uploadedSamples = Math.max(uploadedSamples, from + samples.length);
       emit(partialListeners, out.text ?? "");
+      if (out.endpoint && active && speechActive && !holdActive) finalizeSegment();
     } catch {
     } finally {
       partialInFlight = false;
     }
   };
+  const requestDetect = async () => {
+    if (detectInFlight) return;
+    let total = detectChunks.reduce((n, c) => n + c.length, 0);
+    const MAX_DETECT_PENDING = 30 * SAMPLE_RATE;
+    if (total - detectSent > MAX_DETECT_PENDING) {
+      detectSent = Math.max(0, total - MAX_DETECT_PENDING);
+    }
+    while (detectChunks.length > 0 && detectSent >= detectChunks[0].length) {
+      detectSent -= detectChunks[0].length;
+      detectChunks.shift();
+    }
+    total = detectChunks.reduce((n, c) => n + c.length, 0);
+    if (total - detectSent <= 0) return;
+    const samples = sliceChunks(detectChunks, detectSent);
+    const epoch = segmentEpoch;
+    const gen = detectGeneration;
+    detectInFlight = true;
+    try {
+      const res = await fetch(asrUrl(false, detectSent, epoch) + "&vadOnly=1", {
+        method: "POST",
+        headers: { "content-type": "application/octet-stream" },
+        body: samples.buffer,
+        signal: AbortSignal.timeout(5e3)
+        // 防服务端挂起长期锁死 detectInFlight（Important#2）
+      });
+      if (epoch !== segmentEpoch || gen !== detectGeneration) return;
+      if (!res.ok) return;
+      const out = await res.json();
+      if (epoch !== segmentEpoch || gen !== detectGeneration) return;
+      if (out.isSpeech !== void 0) config.onIsSpeech?.(out.isSpeech);
+      detectSent += samples.length;
+      while (detectChunks.length > 0 && detectSent >= detectChunks[0].length) {
+        detectSent -= detectChunks[0].length;
+        detectChunks.shift();
+      }
+    } catch {
+    } finally {
+      detectInFlight = false;
+    }
+  };
   const resetHostStream = async () => {
     try {
-      await fetch(`${asrUrl(false)}&reset=1`, { method: "POST" });
+      await fetch(`${asrUrl(false)}&reset=1`, { method: "POST", signal: AbortSignal.timeout(5e3) });
     } catch {
     }
   };
-  const finalizeSegment = () => {
+  const finalizeSegment = (force = false) => {
     if (segment.length === 0) return;
-    const samples = concatSegment();
-    const epoch = ++segmentEpoch;
+    if (config.isPlaying?.() && !forcePending && !force) return;
+    if (utteranceEndAt === null) {
+      utteranceEndAt = Date.now();
+      emitTelemetry("utterance-end");
+    }
+    emitTelemetry("endpoint-fired");
+    const from = uploadedSamples;
+    const samples = sliceSince(from);
+    const recoverySegment = segment;
+    const epochSnapshot = segmentEpoch;
+    segmentEpoch++;
     const meta = { force: forcePending };
     forcePending = false;
+    speechMs = 0;
+    uploadedSamples = 0;
     segment = [];
     speechActive = false;
     silenceMs = 0;
@@ -210,22 +324,41 @@ function createAsrEngine(config, sessionId) {
     prePad = [];
     setState("transcribing");
     void (async () => {
-      try {
-        let res = await fetch(asrUrl(true), {
-          method: "POST",
-          headers: { "content-type": "application/octet-stream" },
-          body: samples.buffer
-        });
+      emitTelemetry("submitted");
+      const MAX_FINAL_ATTEMPTS = 3;
+      const restoreState = () => setState(active ? speechActive || holdActive ? "speech" : wakeEnabled ? "wake" : "listening" : "idle");
+      for (let attempt = 0; attempt < MAX_FINAL_ATTEMPTS; attempt++) {
+        if (attempt > 0) {
+          if (segmentEpoch !== epochSnapshot + 1) return;
+          await new Promise((r) => setTimeout(r, 500 * attempt));
+        }
+        const useFull = attempt > 0;
+        const off = useFull ? 0 : from;
+        const body = useFull ? sliceChunks(recoverySegment, 0).buffer : samples.buffer;
+        let res;
+        try {
+          res = await fetch(asrUrl(true, off, epochSnapshot), {
+            signal: AbortSignal.timeout(1e4),
+            method: "POST",
+            headers: { "content-type": "application/octet-stream" },
+            body
+          });
+        } catch {
+          restoreState();
+          if (attempt === MAX_FINAL_ATTEMPTS - 1) console.warn("[dsh-voice-mode] finalize fetch \u5F02\u5E38\uFF08\u91CD\u8BD5\u8017\u5C3D\uFF09");
+          continue;
+        }
         if (res.status === 202) {
           setState("loading-model");
           res = await new Promise((resolve) => {
             setTimeout(async () => {
               try {
                 resolve(
-                  await fetch(asrUrl(true), {
+                  await fetch(asrUrl(true, off, epochSnapshot), {
+                    signal: AbortSignal.timeout(1e4),
                     method: "POST",
                     headers: { "content-type": "application/octet-stream" },
-                    body: samples.buffer
+                    body
                   })
                 );
               } catch {
@@ -234,21 +367,53 @@ function createAsrEngine(config, sessionId) {
             }, 5e3);
           });
         }
-        if (epoch !== segmentEpoch) return;
-        setState(active ? speechActive ? "speech" : "listening" : "idle");
-        if (!res.ok) return;
-        const out = await res.json();
-        if (epoch !== segmentEpoch) return;
+        if (res.status === 202) {
+          if (attempt === MAX_FINAL_ATTEMPTS - 1) console.warn("[dsh-voice-mode] finalize \u6A21\u578B\u52A0\u8F7D\u8D85\u65F6\uFF08\u91CD\u8BD5\u8017\u5C3D\uFF09");
+          continue;
+        }
+        if (res.status === 403 && config.onSessionExpired) {
+          const recovered = await config.onSessionExpired();
+          if (recovered && segmentEpoch === epochSnapshot + 1) {
+            try {
+              res = await fetch(asrUrl(true, 0, epochSnapshot), {
+                signal: AbortSignal.timeout(1e4),
+                method: "POST",
+                headers: { "content-type": "application/octet-stream" },
+                body: sliceChunks(recoverySegment, 0).buffer
+              });
+            } catch {
+            }
+          }
+        }
+        restoreState();
+        if (!res.ok) {
+          if (attempt === MAX_FINAL_ATTEMPTS - 1) console.warn("[dsh-voice-mode] finalize 5xx\uFF08\u91CD\u8BD5\u8017\u5C3D\uFF09");
+          continue;
+        }
+        let out;
+        try {
+          out = await res.json();
+        } catch {
+          if (attempt === MAX_FINAL_ATTEMPTS - 1) console.warn("[dsh-voice-mode] finalize \u54CD\u5E94\u975E JSON\uFF08\u91CD\u8BD5\u8017\u5C3D\uFF09");
+          continue;
+        }
+        if (segmentEpoch !== epochSnapshot + 1) return;
         if (out.text) emit(transcriptListeners, out.text, meta);
-      } catch {
-        setState(active ? speechActive ? "speech" : "listening" : "idle");
-        emitError("recognitionFail");
+        return;
       }
+      emitError("recognitionFail");
     })();
   };
+  let latestResidualRms = 0;
+  let echoFloorRms = 0;
+  let echoPeak = 0;
   const handleAudio = (raw) => {
     if (!active || inFlush) return;
-    const data = ctxRate !== SAMPLE_RATE ? resampleTo16k(raw, ctxRate) : raw;
+    let data = ctxRate !== SAMPLE_RATE ? resampleLinear(raw, ctxRate, SAMPLE_RATE) : raw;
+    if (echo) {
+      const ref = echo.windowAt(performance.now(), data.length);
+      data = echo.process(data, ref);
+    }
     let sum = 0;
     for (let i = 0; i < data.length; i++) sum += data[i] * data[i];
     const rms = Math.sqrt(sum / data.length);
@@ -259,23 +424,25 @@ function createAsrEngine(config, sessionId) {
       } catch {
       }
     }
-    if (Date.now() < bargeInDampingUntil) {
-      interruptCandidateMs = 0;
-    } else if (rms > intLevel.rms) {
-      interruptCandidateMs += durationMs;
-      if (interruptCandidateMs >= intLevel.ms) {
-        interruptCandidateMs = 0;
-        bargeInDampingUntil = Date.now() + 800;
-        for (const fn of speechStartListeners) {
-          try {
-            fn();
-          } catch {
-          }
-        }
-      }
+    const playingNow = config.isPlaying?.() ?? false;
+    const gateRatio = Math.pow(10, (config.echoGateDb ?? 6) / 20);
+    const peakDecay = Math.pow(0.9, durationMs / 64);
+    const floorAlpha = 1 - Math.pow(0.98, durationMs / 64);
+    if (playingNow) {
+      latestResidualRms = rms;
+      echoPeak = Math.max(echoPeak * peakDecay, rms);
+    }
+    const doubleTalk = playingNow && echoFloorRms > 0 && rms > echoFloorRms * gateRatio;
+    if (playingNow) {
+      if (echoFloorRms === 0) echoFloorRms = rms;
+      else if (!doubleTalk) echoFloorRms = echoFloorRms * (1 - floorAlpha) + rms * floorAlpha;
     } else {
-      interruptCandidateMs = 0;
+      echoPeak = 0;
     }
+    if (echo) {
+      echo.setFrozen(doubleTalk);
+    }
+    if (playingNow && !holdActive && config.mode !== "hold") detectChunks.push(data);
     if (holdActive) {
       if (!speechActive) {
         speechActive = true;
@@ -285,6 +452,7 @@ function createAsrEngine(config, sessionId) {
       silenceMs = 0;
       segment.push(data);
       if (segmentMs > MAX_SEGMENT_MS) finalizeSegment();
+    } else if (config.mode === "hold") {
     } else if (state === "wake") {
       if (rms > SPEECH_RMS) {
         segmentMs += durationMs;
@@ -293,38 +461,55 @@ function createAsrEngine(config, sessionId) {
           segment = [];
           segmentMs = 0;
           silenceMs = 0;
-          prePad = [];
+          uploadedSamples = 0;
           void resetHostStream();
         }
-      } else {
-        prePad.push(data);
-        let total = 0;
-        let cut = 0;
-        for (let i = prePad.length - 1; i >= 0; i--) {
-          total += prePad[i].length / SAMPLE_RATE * 1e3;
-          if (total > PRE_PAD_MS) {
-            cut = i + 1;
-            break;
-          }
-        }
-        if (cut > 0) prePad = prePad.slice(cut);
       }
     } else if (rms > SPEECH_RMS) {
+      if (config.isPlaying?.()) {
+        if (speechActive) finalizeSegment(true);
+        return;
+      }
       if (!speechActive) {
         speechActive = true;
+        detectChunks = [];
+        detectSent = 0;
+        detectGeneration++;
+        utteranceEndAt = null;
         setState("speech");
         for (const p of prePad) segment.push(p);
         prePad = [];
       }
+      speechMs += durationMs;
       segmentMs += durationMs;
       silenceMs = 0;
       segment.push(data);
       if (segmentMs > MAX_SEGMENT_MS) finalizeSegment();
     } else if (speechActive) {
+      if (utteranceEndAt === null) {
+        utteranceEndAt = Date.now();
+        emitTelemetry("utterance-end");
+      }
       segmentMs += durationMs;
       silenceMs += durationMs;
       segment.push(data);
-      if (silenceMs > config.silenceMs) finalizeSegment();
+      if (silenceMs > config.silenceMs) {
+        if (speechMs >= MIN_SPEECH_MS) {
+          finalizeSegment();
+        } else {
+          segmentEpoch++;
+          segment = [];
+          speechActive = false;
+          speechMs = 0;
+          silenceMs = 0;
+          segmentMs = 0;
+          prePad = [];
+          utteranceEndAt = null;
+          uploadedSamples = 0;
+          void resetHostStream();
+          setState("listening");
+        }
+      }
     } else {
       prePad.push(data);
       let total = 0;
@@ -338,34 +523,41 @@ function createAsrEngine(config, sessionId) {
       }
       if (cut > 0) prePad = prePad.slice(cut);
     }
-    sincePartialMs += durationMs;
-    if ((speechActive || holdActive || state === "wake") && sincePartialMs >= PARTIAL_INTERVAL_MS) {
-      sincePartialMs = 0;
-      void requestPartial();
+    const nowMs = Date.now();
+    if (nowMs - lastPollAt >= PARTIAL_INTERVAL_MS) {
+      if (playingNow && !speechActive && !holdActive) {
+        lastPollAt = nowMs;
+        void requestDetect();
+      } else if (speechActive || holdActive || state === "wake") {
+        lastPollAt = nowMs;
+        void requestPartial();
+      }
     }
   };
-  function resampleTo16k(src, srcRate) {
-    const ratio = srcRate / SAMPLE_RATE;
-    const outLen = Math.max(1, Math.floor(src.length / ratio));
-    const out = new Float32Array(outLen);
-    for (let i = 0; i < outLen; i++) {
-      const pos = i * ratio;
-      const i0 = Math.floor(pos);
-      const i1 = Math.min(i0 + 1, src.length - 1);
-      const frac = pos - i0;
-      out[i] = src[i0] + (src[i1] - src[i0]) * frac;
-    }
-    return out;
-  }
   const startRecorder = async () => {
+    const mySeq = startSeq;
     stream = await navigator.mediaDevices.getUserMedia({
       audio: {
         channelCount: 1,
         echoCancellation: true,
-        noiseSuppression: true,
-        autoGainControl: true
+        // L0（研究核实）：AGC/NS 是时变非线性，放在 AEC 前破坏线性回声路径假设——
+        // WebRTC GainController2 与 Speex 均把增益放 AEC 之后。禁用后原生 AEC 与
+        // 自研 AEC 的输入才线性，回声消得净；AGC 放大近端语音由「残差地板门控」
+        // 的归一化语义替代，不需要采集端增益。
+        noiseSuppression: false,
+        autoGainControl: false
       }
     });
+    if (stopRequested || mySeq !== startSeq) {
+      stream.getTracks().forEach((t3) => t3.stop());
+      stream = null;
+      return;
+    }
+    const aecOn = stream.getAudioTracks()[0]?.getSettings().echoCancellation === true;
+    if (!aecOn) {
+      console.warn("[dsh-voice-mode] \u6D4F\u89C8\u5668\u539F\u751F echoCancellation \u672A\u751F\u6548\uFF08\u5916\u653E\u53EF\u80FD\u81EA\u6253\u65AD\uFF09\uFF0C\u5EFA\u8BAE\u7528\u8033\u673A\u6216\u300C\u624B\u52A8\u6253\u65AD\u300D");
+    }
+    config.onAecState?.(aecOn);
     const AC = window.AudioContext ?? window.webkitAudioContext;
     audioCtx = new AC({ sampleRate: SAMPLE_RATE });
     try {
@@ -374,6 +566,29 @@ function createAsrEngine(config, sessionId) {
     }
     ctxRate = audioCtx.sampleRate;
     const source = audioCtx.createMediaStreamSource(stream);
+    if (audioCtx.audioWorklet) {
+      try {
+        if (!workletBlobUrl) {
+          workletBlobUrl = URL.createObjectURL(new Blob(['"use strict";\n(() => {\n  // src/resample.ts\n  function resampleLinear(src, srcRate, dstRate) {\n    if (srcRate === dstRate) return src;\n    if (src.length === 0) return src;\n    const ratio = srcRate / dstRate;\n    const outLen = Math.max(1, Math.floor(src.length / ratio));\n    const out = new Float32Array(outLen);\n    for (let i = 0; i < outLen; i++) {\n      const pos = i * ratio;\n      const i0 = Math.floor(pos);\n      const i1 = Math.min(i0 + 1, src.length - 1);\n      const frac = pos - i0;\n      out[i] = src[i0] + (src[i1] - src[i0]) * frac;\n    }\n    return out;\n  }\n\n  // src/audio-worklet.ts\n  var TARGET_RATE = 16e3;\n  var CHUNK = 1024;\n  var RATIO = sampleRate / TARGET_RATE;\n  var NEED = Math.ceil(CHUNK * RATIO);\n  var VoiceCaptureProcessor = class extends AudioWorkletProcessor {\n    acc = new Float32Array(0);\n    accLen = 0;\n    process(inputs) {\n      const ch = inputs[0]?.[0];\n      if (ch && ch.length > 0) this.push(ch);\n      this.drain();\n      return true;\n    }\n    push(ch) {\n      if (this.accLen + ch.length > this.acc.length) {\n        let cap = this.acc.length > 0 ? this.acc.length : NEED * 2;\n        while (cap < this.accLen + ch.length) cap *= 2;\n        const next = new Float32Array(cap);\n        next.set(this.acc.subarray(0, this.accLen));\n        this.acc = next;\n      }\n      this.acc.set(ch, this.accLen);\n      this.accLen += ch.length;\n    }\n    drain() {\n      while (this.accLen >= NEED) {\n        const src = sampleRate === TARGET_RATE ? this.acc.subarray(0, CHUNK) : resampleLinear(this.acc.subarray(0, NEED), sampleRate, TARGET_RATE);\n        const chunk = new Float32Array(CHUNK);\n        chunk.set(src.length >= CHUNK ? src.subarray(0, CHUNK) : src);\n        this.port.postMessage(chunk, [chunk.buffer]);\n        this.acc.copyWithin(0, NEED, this.accLen);\n        this.accLen -= NEED;\n      }\n    }\n  };\n  registerProcessor("voice-capture", VoiceCaptureProcessor);\n})();\n'], { type: "text/javascript" }));
+        }
+        await audioCtx.audioWorklet.addModule(workletBlobUrl);
+        workletNode = new AudioWorkletNode(audioCtx, "voice-capture", { numberOfInputs: 1, numberOfOutputs: 1, outputChannelCount: [1] });
+        workletNode.port.onmessage = (e) => {
+          handleAudio(e.data);
+        };
+        source.connect(workletNode);
+        workletNode.connect(audioCtx.destination);
+        ctxRate = SAMPLE_RATE;
+        active = true;
+        return;
+      } catch {
+        try {
+          workletNode?.disconnect();
+        } catch {
+        }
+        workletNode = null;
+      }
+    }
     processor = audioCtx.createScriptProcessor(BUFFER_SIZE, 1, 1);
     processor.onaudioprocess = (e) => {
       handleAudio(new Float32Array(e.inputBuffer.getChannelData(0)));
@@ -383,8 +598,6 @@ function createAsrEngine(config, sessionId) {
     active = true;
   };
   const stopRecorder = async () => {
-    if (!active) return;
-    active = false;
     inFlush = true;
     segmentEpoch++;
     forcePending = false;
@@ -393,13 +606,23 @@ function createAsrEngine(config, sessionId) {
     speechActive = false;
     silenceMs = 0;
     segmentMs = 0;
+    speechMs = 0;
+    uploadedSamples = 0;
     prePad = [];
-    interruptCandidateMs = 0;
+    detectChunks = [];
+    detectSent = 0;
+    detectGeneration++;
+    utteranceEndAt = null;
     try {
       processor?.disconnect();
     } catch {
     }
     processor = null;
+    try {
+      workletNode?.disconnect();
+    } catch {
+    }
+    workletNode = null;
     try {
       stream?.getTracks().forEach((t3) => t3.stop());
     } catch {
@@ -417,13 +640,28 @@ function createAsrEngine(config, sessionId) {
     get state() {
       return state;
     },
+    get holding() {
+      return holdActive;
+    },
+    /** A2.5 回声门控：当前残差是否明显高于回声地板（marginDb 默认 6dB）——判用户人声而非回声。 */
+    aboveEchoFloor(marginDb = 6) {
+      if (echoFloorRms === 0) return false;
+      return echoPeak > echoFloorRms * Math.pow(10, marginDb / 20);
+    },
+    echoLevels() {
+      return { floorRms: echoFloorRms, residualRms: latestResidualRms, peakRms: echoPeak };
+    },
     async start() {
       if (active) return;
+      stopRequested = false;
+      startSeq++;
       segmentEpoch++;
-      sincePartialMs = 0;
-      interruptCandidateMs = 0;
+      lastPollAt = 0;
       holdActive = false;
-      setState(wakeWord ? "wake" : "listening");
+      detectChunks = [];
+      detectSent = 0;
+      detectGeneration++;
+      setState(wakeEnabled ? "wake" : "listening");
       try {
         await startRecorder();
       } catch (error) {
@@ -432,7 +670,10 @@ function createAsrEngine(config, sessionId) {
       }
     },
     async stop() {
-      if (!active) {
+      stopRequested = true;
+      const wasActive = active;
+      active = false;
+      if (!wasActive) {
         setState("idle");
         return;
       }
@@ -443,7 +684,7 @@ function createAsrEngine(config, sessionId) {
       const speechS = segment.reduce((n, c) => n + c.length, 0) / SAMPLE_RATE;
       if (speechActive && speechS >= 0.25) {
         forcePending = true;
-        sincePartialMs = 0;
+        lastPollAt = 0;
         finalizeSegment();
       }
     },
@@ -452,15 +693,39 @@ function createAsrEngine(config, sessionId) {
       holdActive = true;
       forcePending = true;
       segmentEpoch++;
+      utteranceEndAt = null;
       segment = [];
       segmentMs = 0;
+      speechMs = 0;
       silenceMs = 0;
       prePad = [];
+      uploadedSamples = 0;
+      detectChunks = [];
+      detectSent = 0;
+      detectGeneration++;
       speechActive = true;
-      sincePartialMs = 0;
-      bargeInDampingUntil = Date.now() + 800;
+      lastPollAt = 0;
       setState("speech");
     },
+    discardSegment() {
+      segmentEpoch++;
+      segment = [];
+      segmentMs = 0;
+      speechMs = 0;
+      silenceMs = 0;
+      speechActive = false;
+      prePad = [];
+      uploadedSamples = 0;
+      detectChunks = [];
+      detectSent = 0;
+      detectGeneration++;
+      utteranceEndAt = null;
+      forcePending = false;
+      lastPollAt = 0;
+      return resetHostStream().then(() => {
+        if (active) setState(wakeEnabled ? "wake" : "listening");
+      });
+    },
     endHeld(cancel = false) {
       if (!active || !holdActive) return;
       holdActive = false;
@@ -471,12 +736,18 @@ function createAsrEngine(config, sessionId) {
         silenceMs = 0;
         prePad = [];
         speechActive = false;
-        setState(wakeWord ? "wake" : "listening");
+        forcePending = false;
+        setState(wakeEnabled ? "wake" : "listening");
         return;
       }
-      forcePending = true;
-      sincePartialMs = 0;
-      finalizeSegment();
+      if (segment.length > 0) {
+        forcePending = true;
+        lastPollAt = 0;
+        finalizeSegment();
+      } else {
+        forcePending = false;
+        setState(wakeEnabled ? "wake" : "listening");
+      }
     },
     onSegment(fn) {
       transcriptListeners.add(fn);
@@ -496,12 +767,6 @@ function createAsrEngine(config, sessionId) {
         partialListeners.delete(fn);
       };
     },
-    onSpeechStart(fn) {
-      speechStartListeners.add(fn);
-      return () => {
-        speechStartListeners.delete(fn);
-      };
-    },
     onState(fn) {
       stateListeners.add(fn);
       fn(state);
@@ -514,10 +779,154 @@ function createAsrEngine(config, sessionId) {
       return () => {
         levelListeners.delete(fn);
       };
+    },
+    onTelemetry(fn) {
+      telemetryListeners.add(fn);
+      return () => {
+        telemetryListeners.delete(fn);
+      };
     }
   };
 }
 
+// src/aec.ts
+var DEFAULT_FILTER_LENGTH = 256;
+var DEFAULT_DELAY = 64;
+var DEFAULT_STEP = 0.1;
+var DEFAULT_EPSILON = 1e-6;
+var MIN_REF_NORM = 1e-6;
+var NlmsAec = class {
+  w;
+  xBuf;
+  filterLength;
+  delay;
+  mu;
+  eps;
+  /** 参考历史环形游标。 */
+  cursor = 0;
+  /** 已缓冲参考样本数（预热期）。 */
+  filled = 0;
+  /** A2.5 双讲冻结：用户说话时暂停权重更新，防滤波器被用户语音带偏。 */
+  frozen = false;
+  constructor(options = {}) {
+    this.filterLength = options.filterLength ?? DEFAULT_FILTER_LENGTH;
+    this.delay = options.delay ?? DEFAULT_DELAY;
+    this.mu = options.step ?? DEFAULT_STEP;
+    this.eps = options.epsilon ?? DEFAULT_EPSILON;
+    this.w = new Float32Array(this.filterLength);
+    this.xBuf = new Float32Array(this.delay + this.filterLength);
+  }
+  /** A2.5 双讲冻结：true 时暂停权重更新（回声相减照常，仅停止自适应）。 */
+  setFrozen(frozen) {
+    this.frozen = frozen;
+  }
+  /**
+   * 送入下一块麦克风/参考；返回去回声后的麦克风样本（与输入等长）。
+   * 参考可比麦克风块短（如静音填充）——不足部分补零。
+   */
+  process(mic, ref) {
+    const n = mic.length;
+    const out = new Float32Array(n);
+    if (n === 0) return out;
+    const xBuf = this.xBuf;
+    const bufLen = xBuf.length;
+    let cursor = this.cursor;
+    for (let i = 0; i < n; i++) {
+      xBuf[cursor] = i < ref.length ? ref[i] : 0;
+      cursor = (cursor + 1) % bufLen;
+      this.filled = Math.min(this.filled + 1, bufLen);
+      const d = mic[i];
+      if (this.filled >= this.delay + this.filterLength) {
+        let y = 0;
+        let norm = 0;
+        let idx = (cursor - this.delay + bufLen) % bufLen;
+        for (let t3 = 0; t3 < this.filterLength; t3++) {
+          const x = xBuf[idx];
+          y += this.w[t3] * x;
+          norm += x * x;
+          idx = (idx - 1 + bufLen) % bufLen;
+        }
+        const e = d - y;
+        out[i] = Number.isFinite(e) ? e : 0;
+        if (norm > MIN_REF_NORM && !this.frozen) {
+          const denom = norm + this.eps;
+          const gain = this.mu * e / denom;
+          idx = (cursor - this.delay + bufLen) % bufLen;
+          for (let t3 = 0; t3 < this.filterLength; t3++) {
+            this.w[t3] += gain * xBuf[idx];
+            idx = (idx - 1 + bufLen) % bufLen;
+          }
+        }
+      } else {
+        out[i] = d;
+      }
+    }
+    this.cursor = cursor;
+    return out;
+  }
+};
+function estimateBulkDelay(mic, ref, opts = {}) {
+  const sr = opts.sampleRate ?? 16e3;
+  const ds = Math.max(1, Math.floor(opts.downsample ?? 4));
+  const minLag = Math.max(0, opts.minLag ?? 0);
+  const maxLag = opts.maxLag ?? Math.floor(300 * sr / 1e3);
+  const n = Math.min(mic.length, ref.length);
+  if (n < ds * 64) return { lag: 0, peak: 0 };
+  const N = Math.floor(n / ds);
+  const maxLagD = Math.floor(maxLag / ds);
+  const minLagD = Math.floor(minLag / ds);
+  if (maxLagD >= N) return { lag: 0, peak: 0 };
+  const m = new Float32Array(N);
+  const r = new Float32Array(N);
+  let mSum = 0;
+  let rSum = 0;
+  for (let i = 0; i < N; i++) {
+    let mv = 0;
+    let rv = 0;
+    for (let k = 0; k < ds; k++) {
+      mv += mic[i * ds + k];
+      rv += ref[i * ds + k];
+    }
+    mv /= ds;
+    rv /= ds;
+    m[i] = mv;
+    r[i] = rv;
+    mSum += mv;
+    rSum += rv;
+  }
+  const mMu = mSum / N;
+  const rMu = rSum / N;
+  let mE = 0;
+  let rE = 0;
+  for (let i = 0; i < N; i++) {
+    m[i] -= mMu;
+    r[i] -= rMu;
+    mE += m[i] * m[i];
+    rE += r[i] * r[i];
+  }
+  if (mE < 1e-9 || rE < 1e-9) return { lag: 0, peak: 0 };
+  const prefM = new Float32Array(N + 1);
+  const prefR = new Float32Array(N + 1);
+  for (let i = 0; i < N; i++) {
+    prefM[i + 1] = prefM[i] + m[i] * m[i];
+    prefR[i + 1] = prefR[i] + r[i] * r[i];
+  }
+  let bestLag = 0;
+  let bestPeak = -Infinity;
+  for (let d = minLagD; d <= maxLagD; d++) {
+    let corr = 0;
+    for (let i = d; i < N; i++) corr += m[i] * r[i - d];
+    const em = prefM[N] - prefM[d];
+    const er = prefR[N - d];
+    const peak = em > 1e-12 && er > 1e-12 ? corr / Math.sqrt(em * er) : 0;
+    if (peak > bestPeak) {
+      bestPeak = peak;
+      bestLag = d;
+    }
+  }
+  return { lag: bestLag * ds, peak: bestPeak };
+}
+
 // src/strings.ts
 var zh = {
   stateVoiceMode: "\u8BED\u97F3\u6A21\u5F0F",
@@ -530,6 +939,7 @@ var zh = {
   hold: "\u6309\u4F4F",
   recognizing: "\u8BC6\u522B\u4E2D\u2026",
   holdToTalk: "\u6309\u4F4F\u8BF4\u8BDD",
+  releaseToSend: "\u677E\u5F00\u53D1\u9001",
   voiceDetected: "\u8BED\u97F3\u4E2D",
   entering: "\u8FDB\u5165\u4E2D\u2026",
   voiceBtn: "\u8BED\u97F3",
@@ -540,15 +950,18 @@ var zh = {
   titleEnter: "\u8FDB\u5165\u8BED\u97F3\u5BF9\u8BDD\u6A21\u5F0F\uFF08Ctrl+Shift+V\uFF09",
   loadingModel: "\u6B63\u5728\u52A0\u8F7D\u6A21\u578B\u2026",
   listening: "\u8046\u542C\u4E2D\u2026",
-  wakeWord: "\u5524\u9192\u8BCD",
+  thinking: "\u601D\u8003\u4E2D\u2026",
   barHold: "\u8BED\u97F3\u6A21\u5F0F \xB7 \u6309\u4F4F\u8BF4\u8BDD\uFF08\u77ED\u6309\u9000\u51FA\uFF09",
   barListening: "\u8BED\u97F3\u6A21\u5F0F \xB7 \u8046\u542C\u4E2D\u2026",
+  wakeWord: "\u5524\u9192\u8BCD",
+  sayWake: "\u8BF4\u300C{wake}\u300D\u5F00\u59CB",
   reading: "\u6717\u8BFB\u4E2D\u2026",
   recognitionFail: "\u8BC6\u522B\u5931\u8D25\uFF0C\u8BF7\u91CD\u8BD5",
+  sessionExpired: "\u8BED\u97F3\u4F1A\u8BDD\u5DF2\u65AD\u5F00\uFF0C\u6B63\u5728\u91CD\u8FDE\u2026",
+  sessionExpiredFail: "\u8BED\u97F3\u4F1A\u8BDD\u91CD\u8FDE\u5931\u8D25\uFF0C\u8BF7\u91CD\u65B0\u5F00\u542F\u8BED\u97F3\u6A21\u5F0F",
   modelDownloadFail: "\u8BED\u97F3\u6A21\u578B\u4E0B\u8F7D\u5931\u8D25\uFF08{file}\uFF09\uFF1A\u8BF7\u68C0\u67E5\u7F51\u7EDC\u540E\u91CD\u65B0\u8FDB\u5165\u8BED\u97F3\u6A21\u5F0F\u91CD\u8BD5",
   startFail: "\u8BED\u97F3\u6A21\u5F0F\u542F\u52A8\u5931\u8D25\uFF1A{err}",
   holdDots: "\u6309\u4F4F\u8BF4\u8BDD\u2026",
-  sayWake: "\u8BF4\u300C{wake}\u300D\u5F00\u59CB",
   exit: "\u9000\u51FA",
   skip: "\u8DF3\u8FC7",
   configUnavailableNote: "\uFF08\u8BBE\u7F6E\u6587\u6863\u672A\u5C31\u7EEA\uFF0C\u9762\u677F\u5C31\u7EEA\u540E\u4F1A\u81EA\u52A8\u51FA\u73B0\uFF09\u3002",
@@ -558,29 +971,95 @@ var zh = {
   previewPlayFail: "\u8BD5\u542C\u5931\u8D25\uFF1A\u65E0\u6CD5\u64AD\u653E\u8BE5\u97F3\u8272",
   previewAutoplay: "\u6D4F\u89C8\u5668\u62E6\u622A\u4E86\u81EA\u52A8\u64AD\u653E\uFF0C\u8BF7\u518D\u70B9\u4E00\u6B21\u8BD5\u542C",
   previewCheck: "\u8BD5\u542C\u5931\u8D25\uFF1A\u8BF7\u68C0\u67E5\u7F51\u7EDC\u6216\u97F3\u8272\u540D\uFF08ShortName\uFF09\u662F\u5426\u6B63\u786E",
+  previewRateLimited: "\u8BD5\u542C\u592A\u9891\u7E41\u4E86\uFF0C\u7A0D\u5019\u51E0\u79D2\u518D\u70B9\uFF08\u6BCF\u5206\u949F\u9650 20 \u6B21\uFF09",
+  previewTimeout: "\u5408\u6210\u8D85\u65F6\uFF1A\u6A21\u578B\u4ECD\u5728\u52A0\u8F7D\uFF0C\u8BF7\u7A0D\u5019\u51E0\u79D2\u518D\u8BD5",
+  previewSynthesisFail: "\u5408\u6210\u5931\u8D25",
   previewBtnTitle: "\u8BD5\u542C\u5F53\u524D\u97F3\u8272\uFF08\u5F53\u524D\u8BED\u901F\uFF09",
   synthesizing: "\u5408\u6210\u4E2D\u2026",
   preview: "\u8BD5\u542C",
   custom: "\u81EA\u5B9A\u4E49",
   // settings rows
-  descVoice: "Edge TTS \u97F3\u8272\uFF08\u4E0B\u62C9\u5E38\u7528\uFF0C\u5176\u4F59\u9009\u300C\u81EA\u5B9A\u4E49\u300D\u624B\u52A8\u586B ShortName\uFF09",
+  voicePrev: "\u4E0A\u4E00\u97F3\u8272",
+  voiceNext: "\u4E0B\u4E00\u97F3\u8272",
+  modeBtnToggle: "\u6301\u7EED\u8046\u542C",
+  modeBtnHold: "\u6309\u4F4F\u8BF4",
+  modeBtnTitle: "\u70B9\u51FB\u5207\u6362\u4EA4\u4E92\u6A21\u5F0F\uFF08\u4FDD\u5B58\u5230\u8BBE\u7F6E\uFF09",
+  descVoice: "Edge \u4E91\u7AEF\u97F3\u8272\uFF08\u8FDB\u5165\u65F6\u81EA\u52A8\u52A0\u8F7D\u5FAE\u8F6F\u5168\u90E8\u97F3\u8272\uFF0C\u4E0B\u62C9\u6216 \u25C0\u25B6 \u9009\uFF1B\u7F51\u7EDC\u5F02\u5E38\u56DE\u9000\u5E38\u7528 14 \u4E2A\uFF0C\u4E5F\u53EF\u300C\u81EA\u5B9A\u4E49\u300D\u586B ShortName\uFF09",
+  descVoiceLocal: "\u672C\u5730\u97F3\u8272\uFF08vits \u4E94\u4E2A\u8BF4\u8BDD\u4EBA\u5168\u90E8\u5217\u51FA\uFF0C\u4E0B\u62C9\u9009\u6216 \u25C0\u25B6 \u5207\u6362\uFF1B\u65E0\u9700\u81EA\u5B9A\u4E49\uFF09",
+  ttsEngine: "\u6717\u8BFB\u5F15\u64CE",
+  descTtsEngine: "\u672C\u5730 VITS\uFF08\u7EAF\u4E2D\u6587\uFF09/ \u672C\u5730 Kokoro\uFF08\u4E2D\u82F1\u5747\u53EF\uFF09/ Edge \u4E91\u7AEF\uFF08\u97F3\u8D28\u6700\u81EA\u7136\uFF0C\u6587\u672C\u4E0A\u5FAE\u8F6F\uFF09",
+  engineVits: "\u672C\u5730 VITS",
+  engineKokoro: "\u672C\u5730\u4E2D\u82F1",
+  engineEdge: "Edge \u4E91\u7AEF",
+  descVoiceKokoro: "Kokoro \u4E2D\u82F1\u97F3\u8272\uFF08103 \u4E2A\u5168\u90E8\u5217\u51FA\uFF0C\u4E0B\u62C9\u9009\u6216 \u25C0\u25B6 \u5207\u6362\uFF1B48-51 \u4E2D\u6587\u540D\uFF0C\u5176\u4F59\u6309\u7F16\u53F7+\u5B9E\u6D4B\u6027\u522B\u6807\u6CE8\uFF0C\u4E2D\u82F1\u6DF7\u8BFB\u5747\u53EF\uFF09",
   descRate: "\u6717\u8BFB\u8BED\u901F\u500D\u7387\uFF080.5 \u6162\u901F \uFF5E 2.0 \u5FEB\u901F\uFF0C1.0 \u6B63\u5E38\uFF09",
-  descInterrupt: "\u53D1\u58F0\u6253\u65AD\u7075\u654F\u5EA6\uFF080 \u9AD8\u95E8\u69DB / 1 \u4E2D / 2 \u4F4E\uFF09",
+  descInterrupt: "\u53D1\u58F0\u6253\u65AD\u7075\u654F\u5EA6\uFF080 \u9AD8\u95E8\u69DB / 1 \u4E2D / 2 \u4F4E\uFF1B\u53D1\u58F0\u786E\u8BA4\u7EA6 0.3/0.2/0.1 \u79D2\uFF09",
+  descBargeIn: "\u6253\u65AD\u65B9\u5F0F\uFF08auto \u81EA\u52A8\u6253\u65AD\uFF1A\u5F00\u53E3\u5373\u6253\u65AD\uFF0C\u8033\u673A/\u5B89\u9759\u73AF\u5883\u63A8\u8350\uFF1Bmanual \u624B\u52A8\u6253\u65AD\uFF1A\u5916\u653E\u63A8\u8350\u2014\u2014\u56DE\u58F0\u4E0D\u4F1A\u8BEF\u89E6\u53D1\u81EA\u6253\u65AD\uFF0C\u6309\u4F4F\u9EA6\u514B\u98CE/Ctrl \u663E\u5F0F\u6253\u65AD\uFF09",
+  bargeInAuto: "\u81EA\u52A8",
+  bargeInManual: "\u624B\u52A8",
+  descEchoGate: "\u56DE\u58F0\u95E8\u63A7\u9608\u503C\uFF08dB\uFF0C\u9ED8\u8BA4 6\uFF09\uFF1A\u81EA\u52A8\u6253\u65AD\u8981\u6C42\u6B8B\u5DEE\u9AD8\u4E8E\u56DE\u58F0\u5730\u677F\u6B64\u503C\uFF1B\u5916\u653E\u4ECD\u8BEF\u6253\u65AD\u8C03\u5927\uFF088~10\uFF09\uFF0C\u592A\u96BE\u6253\u65AD\u8C03\u5C0F\uFF083~4\uFF09",
+  descShortcut: "\u8FDB\u5165/\u9000\u51FA\u8BED\u97F3\u6A21\u5F0F\u7684\u5FEB\u6377\u952E\uFF08\u5F62\u5982 Ctrl+Shift+V\uFF1B\u7559\u7A7A\u7981\u7528\u5FEB\u6377\u952E\uFF0C\u53EA\u7528\u9EA6\u514B\u98CE\u6309\u94AE\uFF1B\u907F\u514D\u6D4F\u89C8\u5668\u4FDD\u7559\u7EC4\u5408\u5982 Ctrl+W/N/T\uFF09",
+  vadDetected: "VAD \u68C0\u6D4B\u5230\u8BED\u97F3",
+  aecOff: "\u539F\u751F\u56DE\u58F0\u6D88\u9664\u672A\u751F\u6548",
+  aecOffHint: "\u6D4F\u89C8\u5668\u539F\u751F\u56DE\u58F0\u6D88\u9664\u672A\u751F\u6548\uFF08\u5916\u653E\u53EF\u80FD\u81EA\u6253\u65AD\uFF09\uFF0C\u5EFA\u8BAE\u7528\u8033\u673A\u6216\u5207\u6362\u300C\u624B\u52A8\u6253\u65AD\u300D",
+  interruptConfirm: "\u6253\u65AD\u786E\u8BA4",
   sev0: "0 \u9AD8\u95E8\u69DB",
   sev1: "1 \u4E2D",
   sev2: "2 \u4F4E",
-  descSilence: "\u8BF4\u5B8C\u6574\u4E00\u53E5\u7684\u9759\u97F3\u505C\u987F\u6BEB\u79D2\u6570\uFF08\u9ED8\u8BA4 2000 = 2 \u79D2\uFF09",
+  descSilence: "\u8BF4\u5B8C\u6574\u4E00\u53E5\u7684\u9759\u97F3\u505C\u987F\u6BEB\u79D2\u6570\uFF08\u9ED8\u8BA4 700 \u6BEB\u79D2\uFF1B\u81F3\u5C11 250ms \u8BED\u97F3\u624D\u5224\u53E5\uFF0C\u9632\u77ED\u4FC3\u566A\u58F0\u8BEF\u89E6\u53D1\uFF09",
   descIdle: "\u65E0\u6D3B\u52A8\u81EA\u52A8\u9000\u51FA\u8BED\u97F3\u6A21\u5F0F\u7684\u5206\u949F\u6570\uFF08\u9ED8\u8BA4 10\uFF09",
   descModelHost: "ASR \u6A21\u578B\u4E0B\u8F7D\u6E90\uFF08\u5B98\u65B9\u6E90 / \u56FD\u5185\u955C\u50CF\uFF0C\u6216\u9009\u300C\u81EA\u5B9A\u4E49\u300D\u586B\u4EFB\u610F\u955C\u50CF\uFF09",
   descAutoSend: "\u8BC6\u522B\u5B9A\u7A3F\u540E\u81EA\u52A8\u53D1\u9001\uFF08\u5173=\u53EA\u8FDB\u8349\u7A3F\uFF1B\u6309\u4F4F Ctrl / hold \u677E\u624B\u4ECD\u53D1\u9001\uFF09",
-  descSpokenFormat: "\u8BED\u97F3\u4F1A\u8BDD\u6CE8\u5165\u53E3\u8BED\u5316\u63D0\u793A\u8BCD\uFF08\u56DE\u590D\u53E3\u8BED\u5316\u3001\u4E0D\u7528 Markdown \u6392\u7248\u7B26\u53F7\uFF0C\u6717\u8BFB\u66F4\u987A\uFF1B\u9ED8\u8BA4\u5173\uFF0C\u6539\u52A8\u5373\u65F6\u751F\u6548\uFF09",
+  descAutoResume: "\u5207\u6362\u56DE\u4E0A\u6B21\u8BED\u97F3\u4F1A\u8BDD\u65F6\u81EA\u52A8\u6062\u590D\u8BED\u97F3\u6A21\u5F0F\uFF08\u9ED8\u8BA4\u5173\uFF0C\u9700\u9EA6\u514B\u98CE\u6743\u9650\u5DF2\u6388\u4E88\uFF1B\u7701\u53BB\u6BCF\u6B21\u5207\u6362\u4F1A\u8BDD\u540E\u91CD\u65B0\u70B9\u9EA6\u514B\u98CE\uFF09",
+  descSpokenFormat: "\u8BED\u97F3\u4F1A\u8BDD\u6CE8\u5165\u53E3\u8BED\u5316\u63D0\u793A\u8BCD\uFF08\u56DE\u590D\u53E3\u8BED\u5316\u3001\u77ED\u53E5\u3001\u4E0D\u7528 Markdown \u6392\u7248\u7B26\u53F7\uFF0C\u6717\u8BFB\u66F4\u987A\u66F4\u5FEB\uFF1B\u9ED8\u8BA4\u5F00\uFF0C\u6539\u52A8\u5373\u65F6\u751F\u6548\uFF09",
+  descSenseVoice: "\u5B9A\u7A3F\u7528 SenseVoice \u91CD\u8BD1\uFF08\u5E26\u6807\u70B9 + \u6570\u5B57\u5F52\u4E00\u5316\u3001\u8BC6\u522B\u66F4\u51C6\uFF1B\u9ED8\u8BA4\u5F00\u3002\u5173\u95ED\u53EF\u7701 228MB \u6A21\u578B\uFF0C\u53EA\u8D70\u6D41\u5F0F\u8BC6\u522B\uFF09",
+  descToolBeep: '\u5DE5\u5177\u8C03\u7528\u63D0\u793A\u97F3\uFF08\u9ED8\u8BA4\u5173\uFF09\uFF1AAI \u601D\u8003/\u8C03\u7528\u5DE5\u5177\u65F6"\u6EF4"\u4E00\u58F0\uFF1B\u5ACC\u5435\u5C31\u4FDD\u6301\u5173\u95ED',
   descMode: "\u4EA4\u4E92\u6A21\u5F0F\uFF08toggle \u6301\u7EED\u8046\u542C+\u9759\u97F3\u65AD\u53E5 / hold \u6309\u4F4F\u8BF4\u8BDD\uFF09",
   modeToggle: "\u6301\u7EED\u8046\u542C",
   modeHold: "\u6309\u4F4F\u8BF4\u8BDD",
   descWakeWord: "\u5524\u9192\u8BCD\uFF08\u9ED8\u8BA4\u5173\uFF1B\u5982\u300C\u4F60\u597D\u5C0FD\u300D\uFF0C\u8BF4\u51FA\u540E\u5F00\u59CB\u8BC6\u522B\uFF09",
   wakePlaceholder: "\u5982\uFF1A\u4F60\u597D\u5C0FD",
-  settingsCardDesc: "\u97F3\u8272 / \u8BED\u901F / \u6253\u65AD\u7075\u654F\u5EA6 / \u9759\u97F3\u505C\u987F / \u7A7A\u95F2\u8D85\u65F6 / \u6A21\u578B\u955C\u50CF / \u81EA\u52A8\u53D1\u9001 / \u4EA4\u4E92\u6A21\u5F0F / \u5524\u9192\u8BCD / \u53E3\u8BED\u5316\u63D0\u793A\u8BCD",
-  configUnavailable: "\u914D\u7F6E\u6682\u4E0D\u53EF\u7528"
+  settingsCardDesc: "\u6717\u8BFB\u5F15\u64CE / \u97F3\u8272 / \u8BED\u901F / \u6253\u65AD\u7075\u654F\u5EA6 / \u6253\u65AD\u65B9\u5F0F / \u9759\u97F3\u505C\u987F / \u7A7A\u95F2\u8D85\u65F6 / \u6A21\u578B\u955C\u50CF / \u81EA\u52A8\u53D1\u9001 / \u4EA4\u4E92\u6A21\u5F0F / \u5524\u9192\u8BCD / \u53E3\u8BED\u5316\u63D0\u793A\u8BCD",
+  settingsEffectiveNote: "\u6717\u8BFB\u5F15\u64CE / \u97F3\u8272 / \u8BED\u901F / \u53E3\u8BED\u5316\u63D0\u793A\u8BCD / \u91CD\u8BD1 \u5373\u65F6\u751F\u6548\uFF1B\u5176\u4F59\uFF08\u6253\u65AD\u7075\u654F\u5EA6 / \u6253\u65AD\u65B9\u5F0F / \u56DE\u58F0\u95E8\u63A7 / \u5FEB\u6377\u952E / \u9759\u97F3 / \u7A7A\u95F2 / \u955C\u50CF / \u81EA\u52A8\u53D1\u9001 / \u81EA\u52A8\u6062\u590D / \u4EA4\u4E92\u6A21\u5F0F / \u5524\u9192\u8BCD / \u5DE5\u5177\u63D0\u793A\u97F3\uFF09\u4E0B\u6B21\u8FDB\u5165\u8BED\u97F3\u6A21\u5F0F\u65F6\u751F\u6548\u3002",
+  configUnavailable: "\u914D\u7F6E\u6682\u4E0D\u53EF\u7528",
+  // telemetry（P1-5 开发模式延迟埋点状态条：各段耗时标签）
+  telUtteranceEnd: "\u8BF4\u5B8C",
+  telEndpoint: "\u7AEF\u70B9",
+  telSubmitted: "\u5B9A\u7A3F",
+  telFirstToken: "\u9996Token",
+  telFirstSentence: "\u9996\u53E5",
+  telFirstChunk: "\u9996chunk",
+  telFirstPlayed: "\u9996\u97F3",
+  // 模型管理（设置面板实时状态/重试）
+  modelsTitle: "\u8BED\u97F3\u6A21\u578B",
+  modelsDisabled: "\u5DF2\u5173\u95ED\uFF08\u8BBE\u7F6E\u4E2D\u5F00\u542F\uFF09",
+  modelStreamingAsr: "\u6D41\u5F0F\u8BC6\u522B",
+  modelVad: "\u7AEF\u70B9 VAD",
+  modelSense: "\u5B9A\u7A3F\u91CD\u8BD1",
+  modelsReady: "\u5C31\u7EEA",
+  modelsDownloading: "{file} {percent}%",
+  modelsFail: "\u4E0B\u8F7D\u5931\u8D25\uFF08{sec} \u79D2\u540E\u81EA\u52A8\u91CD\u8BD5\uFF09",
+  modelsMissing: "\u672A\u4E0B\u8F7D",
+  modelsRetry: "\u91CD\u8BD5\u4E0B\u8F7D",
+  modelsRetrying: "\u91CD\u8BD5\u4E2D\u2026",
+  modelsRetryHint: "\u955C\u50CF\u5207\u6362\u6216\u4E0B\u8F7D\u5931\u8D25\u540E\u70B9\u51FB\u7ACB\u5373\u91CD\u8BD5",
+  modelsHint: "\u4E0B\u8F7D/\u8FDB\u5EA6\u5B9E\u65F6\u8DDF\u8FDB\uFF1B\u5931\u8D25\u81EA\u52A8\u9000\u907F 60s \u91CD\u8BD5\u3002\u955C\u50CF\u6E90\u5207\u6362\u540E\u70B9\u300C\u91CD\u8BD5\u4E0B\u8F7D\u300D\u7ACB\u5373\u751F\u6548\uFF1B\u4E5F\u53EF\u7528 npm run prefetch \u9884\u4E0B\u8F7D\u3002",
+  // 朗读引擎状态（设置面板内联展示）
+  engineLoading: "\u52A0\u8F7D\u4E2D\u2026",
+  engineReady: "\u5C31\u7EEA",
+  engineError: "\u52A0\u8F7D\u5931\u8D25",
+  engineIdle: "\u53EF\u8BD5\u542C",
+  ttsModelsMissing: "\u672C\u5730\u6A21\u578B\u672A\u5C31\u7EEA",
+  ttsRedownload: "\u91CD\u65B0\u4E0B\u8F7D",
+  ttsRedownloadHint: "\u6E05\u7406\u8BE5\u5F15\u64CE\u6A21\u578B\u7F13\u5B58\u5E76\u91CD\u65B0\u4E0B\u8F7D",
+  ttsCleaning: "\u6E05\u7406\u4E2D\u2026",
+  // 设置分组标题
+  secRead: "\u6717\u8BFB\u4E0E\u97F3\u8272",
+  secInterrupt: "\u6253\u65AD\u4E0E\u9759\u97F3",
+  secInteraction: "\u4EA4\u4E92\u4E0E\u8F93\u5165",
+  secRecognition: "\u8BC6\u522B\u4E0E\u53E3\u8BED",
+  secModel: "\u6A21\u578B\u4E0E\u955C\u50CF",
+  telTotal: "\u5408\u8BA1"
 };
 var en = {
   stateVoiceMode: "Voice Mode",
@@ -593,6 +1072,7 @@ var en = {
   hold: "Hold",
   recognizing: "Recognizing\u2026",
   holdToTalk: "Hold to talk",
+  releaseToSend: "Release to send",
   voiceDetected: "Voice active",
   entering: "Entering\u2026",
   voiceBtn: "Voice",
@@ -603,15 +1083,18 @@ var en = {
   titleEnter: "Enter voice mode (Ctrl+Shift+V)",
   loadingModel: "Loading model\u2026",
   listening: "Listening\u2026",
-  wakeWord: "Wake word",
+  thinking: "Thinking\u2026",
   barHold: "Voice mode \xB7 hold to talk (tap to exit)",
   barListening: "Voice mode \xB7 listening\u2026",
+  wakeWord: "Wake word",
+  sayWake: 'Say "{wake}" to start',
   reading: "Reading\u2026",
   recognitionFail: "Recognition failed, try again",
+  sessionExpired: "Voice session expired, reconnecting\u2026",
+  sessionExpiredFail: "Voice session reconnect failed; please re-enter voice mode",
   modelDownloadFail: "Model download failed ({file}): check network and re-enter voice mode",
   startFail: "Voice mode failed to start: {err}",
   holdDots: "Hold to talk\u2026",
-  sayWake: 'Say "{wake}" to start',
   exit: "Exit",
   skip: "Skip",
   configUnavailableNote: " (settings document not ready; the panel will appear when it is).",
@@ -620,28 +1103,90 @@ var en = {
   previewPlayFail: "Preview failed: cannot play this voice",
   previewAutoplay: "Autoplay blocked \u2014 click preview again",
   previewCheck: "Preview failed: check network or ShortName",
+  previewRateLimited: "Preview too frequent \u2014 wait a few seconds (20/min limit)",
+  previewTimeout: "Synthesis timed out: model still loading, retry in a few seconds",
+  previewSynthesisFail: "Synthesis failed",
   previewBtnTitle: "Preview voice (current rate)",
   synthesizing: "Synthesizing\u2026",
   preview: "Preview",
   custom: "Custom",
-  descVoice: "Edge TTS voice (presets, or a custom ShortName)",
+  voicePrev: "Previous voice",
+  voiceNext: "Next voice",
+  modeBtnToggle: "Continuous",
+  modeBtnHold: "Hold to talk",
+  modeBtnTitle: "Click to switch interaction mode (saved to settings)",
+  descVoice: "Edge cloud voices (auto-loads all Microsoft voices on entry; \u25C0\u25B6 or dropdown; falls back to 14 common on network failure; custom ShortName allowed)",
+  descVoiceLocal: "Local voice (vits, all 5 speakers listed; dropdown or \u25C0\u25B6; no custom needed)",
+  ttsEngine: "Read-aloud engine",
+  descTtsEngine: "Local VITS (Chinese only) / Local Kokoro (Chinese + English) / Edge cloud (most natural, text sent to Microsoft)",
+  engineVits: "Local VITS",
+  engineKokoro: "Local zh-en",
+  engineEdge: "Edge cloud",
+  descVoiceKokoro: "Kokoro zh-en voices (103; \u25C0\u25B6 to cycle; 48-51 named Chinese, others numbered with measured gender; mixed zh-en supported)",
   descRate: "Speech rate (0.5 slow \u2013 2.0 fast, 1.0 normal)",
-  descInterrupt: "Interrupt sensitivity (0 high barrier / 2 low)",
+  descInterrupt: "Interrupt sensitivity (0 high barrier / 1 medium / 2 low; ~0.3/0.2/0.1 s speech confirmation)",
+  descBargeIn: "Barge-in mode (auto: interrupt by speaking \u2014 headphones/quiet; manual: for loudspeaker, no echo-triggered self-interrupt \u2014 hold mic/Ctrl to interrupt)",
+  bargeInAuto: "Auto",
+  bargeInManual: "Manual",
+  descEchoGate: "Echo gate threshold (dB, default 6): auto barge-in requires the residual to exceed the echo floor by this value; raise (8-10) if speaker echo still interrupts, lower (3-4) if hard to interrupt",
+  descShortcut: "Shortcut to enter/exit voice mode (e.g. Ctrl+Shift+V; empty disables it, mic button only; avoid browser-reserved combos like Ctrl+W/N/T)",
+  vadDetected: "VAD speech",
+  aecOff: "Native AEC off",
+  aecOffHint: "Native echo cancellation is not active (speaker echo may self-interrupt); use headphones or Manual barge-in",
+  interruptConfirm: "interrupt confirm",
   sev0: "0 high",
   sev1: "1 medium",
   sev2: "2 low",
-  descSilence: "Silence pause before a sentence is committed (default 2000 ms)",
+  descSilence: "Silence pause before a sentence is committed (default 700 ms; at least 250 ms of speech required, guards against noise triggers)",
   descIdle: "Auto-exit voice mode after idle minutes (default 10)",
   descModelHost: "ASR model download source (official source / mirror, or any custom URL)",
   descAutoSend: "Auto-send after finalized recognition (off = draft only; Ctrl / hold still sends)",
-  descSpokenFormat: "Inject spoken-format prompt into voice replies (colloquial, no Markdown; default off, live)",
+  descAutoResume: "Auto-resume voice mode when switching back to the last voice session (default off, requires granted mic permission)",
+  descSpokenFormat: "Inject spoken-format prompt into voice replies (colloquial, short sentences, no Markdown; default on, live)",
+  descSenseVoice: "Re-transcribe the finalized utterance with SenseVoice (punctuation + ITN, more accurate; default on \u2014 turn off to skip the 228 MB model and keep streaming only)",
+  descToolBeep: "Tool-call beep (default off): beep when the agent is thinking/calling tools; keep off if it annoys you",
   descMode: "Interaction mode (toggle: continuous listen + auto-send / hold: press to talk)",
   modeToggle: "Continue listen",
   modeHold: "Hold to talk",
   descWakeWord: "Wake word (default off; e.g. Hey D)",
   wakePlaceholder: "e.g. Hey D",
-  settingsCardDesc: "Voice / rate / interrupt / silence / idle / model host / auto-send / mode / wake word / spoken format",
-  configUnavailable: "Configuration unavailable"
+  settingsCardDesc: "Engine / voice / rate / interrupt / barge-in / silence / idle / model host / auto-send / mode / wake word / spoken format",
+  settingsEffectiveNote: "Engine / voice / rate / spoken format / re-transcribe apply immediately; the rest (interrupt / barge-in / echo gate / shortcut / silence / idle / mirror / auto-send / auto-resume / mode / wake word / tool beep) apply next time you enter voice mode.",
+  configUnavailable: "Configuration unavailable",
+  telUtteranceEnd: "end",
+  telEndpoint: "endpoint",
+  telSubmitted: "submit",
+  telFirstToken: "1st token",
+  telFirstSentence: "1st sentence",
+  telFirstChunk: "1st chunk",
+  telFirstPlayed: "1st audio",
+  modelsTitle: "Voice models",
+  modelsDisabled: "off (enable in settings)",
+  modelStreamingAsr: "Streaming ASR",
+  modelVad: "Endpoint VAD",
+  modelSense: "Finalize",
+  modelsReady: "Ready",
+  modelsDownloading: "{file} {percent}%",
+  modelsFail: "Download failed (auto-retry in {sec}s)",
+  modelsMissing: "not downloaded",
+  modelsRetry: "Retry",
+  modelsRetrying: "Retrying\u2026",
+  modelsRetryHint: "Click to retry now after switching mirror or a failure",
+  modelsHint: "Live download state; failures auto-backoff 60s. After switching the mirror, click Retry to take effect immediately; npm run prefetch pre-downloads.",
+  engineLoading: "loading\u2026",
+  engineReady: "ready",
+  engineError: "failed",
+  engineIdle: "ready to preview",
+  ttsModelsMissing: "local models missing",
+  ttsRedownload: "Re-download",
+  ttsRedownloadHint: "Clear this engine model cache and re-download",
+  ttsCleaning: "Clearing\u2026",
+  secRead: "Reading & voice",
+  secInterrupt: "Interrupt & silence",
+  secInteraction: "Interaction",
+  secRecognition: "Recognition & speech",
+  secModel: "Model & mirror",
+  telTotal: "total"
 };
 var guess = () => /^zh\b/i.test(
   typeof document !== "undefined" && document.documentElement.lang || (typeof navigator !== "undefined" ? navigator.language : "") || ""
@@ -666,6 +1211,25 @@ var cardStyle = {
   borderRadius: 12,
   overflow: "hidden"
 };
+var FIELD_LABELS = {
+  ttsEngine: "\u6717\u8BFB\u5F15\u64CE",
+  voice: "\u97F3\u8272",
+  rate: "\u8BED\u901F",
+  interruptLevel: "\u6253\u65AD\u7075\u654F\u5EA6",
+  bargeInMode: "\u6253\u65AD\u65B9\u5F0F",
+  echoGateDb: "\u56DE\u58F0\u95E8\u63A7",
+  mode: "\u4EA4\u4E92\u6A21\u5F0F",
+  shortcut: "\u5FEB\u6377\u952E",
+  wakeWord: "\u5524\u9192\u8BCD",
+  toolBeep: "\u5DE5\u5177\u63D0\u793A\u97F3",
+  autoSend: "\u81EA\u52A8\u53D1\u9001",
+  autoResume: "\u81EA\u52A8\u6062\u590D",
+  senseVoice: "\u5B9A\u7A3F\u91CD\u8BD1",
+  spokenFormat: "\u53E3\u8BED\u5316\u63D0\u793A\u8BCD",
+  silenceMs: "\u9759\u97F3\u505C\u987F",
+  idleTimeoutMinutes: "\u7A7A\u95F2\u8D85\u65F6",
+  modelHost: "\u6A21\u578B\u955C\u50CF"
+};
 var setHeader = {
   appearance: "none",
   width: "100%",
@@ -742,6 +1306,149 @@ var VOICE_OPTIONS = [
   { v: "en-US-AriaNeural", label: "Aria \xB7 \u5973 \xB7 English" },
   { v: "en-US-GuyNeural", label: "Guy \xB7 \u7537 \xB7 English" }
 ];
+var VOICE_OPTIONS_LOCAL = [
+  { v: "suyingxue", label: "\u7D20\u6620\u96EA \xB7 \u5973" },
+  { v: "gunian", label: "\u987E\u5FF5 \xB7 \u7537" },
+  { v: "fushiyu", label: "\u5085\u65AF\u9047 \xB7 \u5973" },
+  { v: "bingjiao", label: "\u51B0\u5A07 \xB7 \u7537" },
+  { v: "bazong", label: "\u9738\u603B \xB7 \u7537" }
+];
+var KOKORO_F0 = [
+  224,
+  189,
+  154,
+  261,
+  226,
+  222,
+  220,
+  229,
+  198,
+  186,
+  212,
+  293,
+  233,
+  161,
+  247,
+  207,
+  218,
+  216,
+  220,
+  238,
+  242,
+  229,
+  198,
+  286,
+  211,
+  190,
+  264,
+  261,
+  226,
+  147,
+  216,
+  240,
+  233,
+  188,
+  222,
+  247,
+  253,
+  270,
+  276,
+  276,
+  279,
+  320,
+  247,
+  296,
+  276,
+  235,
+  139,
+  240,
+  282,
+  282,
+  238,
+  226,
+  273,
+  216,
+  286,
+  270,
+  198,
+  179,
+  117,
+  130,
+  114,
+  128,
+  108,
+  106,
+  122,
+  136,
+  190,
+  112,
+  108,
+  128,
+  131,
+  111,
+  110,
+  132,
+  138,
+  189,
+  137,
+  148,
+  151,
+  127,
+  135,
+  111,
+  138,
+  114,
+  125,
+  158,
+  128,
+  156,
+  132,
+  162,
+  131,
+  136,
+  142,
+  124,
+  129,
+  136,
+  126,
+  135,
+  161,
+  150,
+  124,
+  104,
+  124
+];
+var KOKORO_NAMED = {
+  48: { v: "zf_xiaobei", label: "\u5C0F\u5317 \xB7 \u4E2D\u6587\u5973" },
+  49: { v: "zf_xiaoni", label: "\u5C0F\u59AE \xB7 \u4E2D\u6587\u5973" },
+  50: { v: "zf_xiaoxiao", label: "\u5C0F\u5C0F \xB7 \u4E2D\u6587\u5973" },
+  51: { v: "zf_xiaoyi", label: "\u5C0F\u827A \xB7 \u4E2D\u6587\u5973" }
+};
+var KOKORO_LABEL_OVERRIDES = {
+  62: "62 \xB7 \u6DF1\u6C89 \xB7 \u5E38\u7528\u7537\u58F0",
+  68: "68 \xB7 \u6D51\u539A \xB7 \u5E38\u7528\u7537\u58F0",
+  75: "75 \xB7 \u6E05\u4EAE \xB7 \u5E38\u7528\u7537\u58F0",
+  76: "76 \xB7 \u78C1\u6027 \xB7 \u5E38\u7528\u7537\u58F0"
+};
+var KOKORO_PINNED = [62, 68, 75, 76];
+function kokoroOption(sid) {
+  const custom = KOKORO_LABEL_OVERRIDES[sid];
+  if (custom) return { v: String(sid), label: custom };
+  const named = KOKORO_NAMED[sid];
+  if (named) return { v: named.v, label: named.label };
+  const hz = KOKORO_F0[sid] ?? null;
+  if (hz === null) return { v: String(sid), label: `${sid} \xB7 \u97F3\u8272` };
+  return { v: String(sid), label: `${sid} \xB7 ${hz < 180 ? "\u7537\u58F0" : "\u5973\u58F0"} \xB7 ${hz}Hz` };
+}
+var VOICE_OPTIONS_KOKORO = [
+  ...KOKORO_PINNED.map((sid) => kokoroOption(sid)),
+  ...KOKORO_F0.map((_, sid) => kokoroOption(sid)).filter((o) => !KOKORO_PINNED.includes(Number(o.v)))
+];
+var ENGINE_DEFAULT_VOICE = {
+  vits: "fushiyu",
+  kokoro: "zf_xiaoni",
+  edge: "zh-CN-XiaoxiaoNeural"
+};
 var HOST_OPTIONS = [
   { v: "https://huggingface.co", label: "\u5B98\u65B9\u6E90 huggingface.co" },
   { v: "https://hf-mirror.com", label: "\u56FD\u5185\u955C\u50CF hf-mirror.com" }
@@ -869,6 +1576,100 @@ function SelectField({
     footer?.(inOptions ? cur : custom)
   ] });
 }
+var stepBtn = {
+  boxSizing: "border-box",
+  width: 36,
+  flex: "0 0 auto",
+  cursor: "pointer",
+  border: `1px solid ${t2.border}`,
+  borderRadius: 8,
+  background: "var(--dsw-alias-bg-layer-2)",
+  color: t2.label,
+  fontSize: 16,
+  lineHeight: "28px",
+  textAlign: "center",
+  padding: 0,
+  fontFamily: "inherit"
+};
+function VoiceSelect({
+  score,
+  field,
+  value,
+  options,
+  placeholder,
+  footer,
+  showCustom = true
+}) {
+  const cur = String(value ?? "");
+  const inOptions = options.some((o) => o.v === cur);
+  const idx = options.findIndex((o) => o.v === cur);
+  const [custom, setCustom] = (0, import_react.useState)(inOptions ? "" : cur);
+  (0, import_react.useEffect)(() => {
+    if (!options.some((o) => o.v === cur)) setCustom(cur);
+  }, [cur, options]);
+  const move = (delta) => {
+    if (options.length === 0) return;
+    if (inOptions) {
+      const n = options.length;
+      const next = options[((idx + delta) % n + n) % n];
+      void score.set(field, next.v);
+    } else {
+      void score.set(field, options[0].v);
+    }
+  };
+  const selectStyle = {
+    ...inputStyle,
+    appearance: "none",
+    cursor: "pointer",
+    backgroundImage: `url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' width='12' height='12' viewBox='0 0 12 12' fill='none'%3E%3Cpath d='M3 4.5L6 7.5L9 4.5' stroke='%2381858C' stroke-width='1.5' stroke-linecap='round' stroke-linejoin='round'/%3E%3C/svg%3E")`,
+    backgroundPosition: "right 12px center",
+    backgroundRepeat: "no-repeat",
+    backgroundSize: "12px 12px",
+    paddingRight: 32
+  };
+  const label = inOptions ? options[idx].label : custom || placeholder || "";
+  const selectValue = inOptions ? cur : showCustom ? "__custom__" : options[0]?.v ?? "";
+  return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: { display: "flex", flexDirection: "column", gap: 6, width: 320, maxWidth: "100%", alignItems: "stretch" }, children: [
+    /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: { display: "flex", gap: 6, alignItems: "center" }, children: [
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("button", { type: "button", "aria-label": t("voicePrev"), onClick: () => move(-1), style: stepBtn, children: "\u2039" }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsxs)(
+        "select",
+        {
+          style: selectStyle,
+          value: selectValue,
+          "aria-label": label,
+          onChange: (e) => {
+            const v = e.target.value;
+            if (v === "__custom__") setCustom(inOptions ? "" : custom);
+            else void score.set(field, v);
+          },
+          children: [
+            options.map((o) => /* @__PURE__ */ (0, import_jsx_runtime.jsx)("option", { value: o.v, children: o.label }, o.v)),
+            showCustom && /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("option", { value: "__custom__", children: [
+              t("custom"),
+              "\u2026"
+            ] })
+          ]
+        }
+      ),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("button", { type: "button", "aria-label": t("voiceNext"), onClick: () => move(1), style: stepBtn, children: "\u203A" })
+    ] }),
+    showCustom && !inOptions && /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+      "input",
+      {
+        style: inputStyle,
+        value: custom,
+        placeholder,
+        onChange: (e) => setCustom(e.target.value),
+        onBlur: () => void score.set(field, custom),
+        onKeyDown: (e) => {
+          if (e.key === "Enter") void score.set(field, custom);
+        }
+      }
+    ),
+    footer?.(inOptions ? cur : showCustom ? custom : cur)
+  ] });
+}
 function VoicePreviewButton({ voice, rate }) {
   const [busy, setBusy] = (0, import_react.useState)(false);
   const [note, setNote] = (0, import_react.useState)(null);
@@ -895,13 +1696,26 @@ function VoicePreviewButton({ voice, rate }) {
           method: "POST",
           headers: { "content-type": "application/json" },
           body: JSON.stringify({ voice: v, rate }),
-          signal: AbortSignal.timeout(15e3)
+          signal: AbortSignal.timeout(9e4)
         });
         if (res.status === 403) {
           setNote(t("previewDisabled"));
           return;
         }
-        if (!res.ok) throw new Error(`preview http ${res.status}`);
+        if (res.status === 429) {
+          setNote(t("previewRateLimited"));
+          return;
+        }
+        if (!res.ok) {
+          let detail = "";
+          try {
+            const parsed = await res.json();
+            if (parsed && typeof parsed.error === "string") detail = parsed.error;
+          } catch {
+          }
+          setNote(detail ? `${t("previewSynthesisFail")}\uFF1A${detail}` : t("previewCheck"));
+          return;
+        }
         const blob = await res.blob();
         const url = URL.createObjectURL(blob);
         audio.src = url;
@@ -918,8 +1732,8 @@ function VoicePreviewButton({ voice, rate }) {
             e instanceof DOMException && e.name === "NotAllowedError" ? t("previewAutoplay") : t("previewPlayFail")
           );
         }
-      } catch {
-        setNote(t("previewCheck"));
+      } catch (e) {
+        setNote(e instanceof DOMException && e.name === "TimeoutError" ? t("previewTimeout") : t("previewCheck"));
       } finally {
         setBusy(false);
       }
@@ -949,41 +1763,264 @@ function VoicePreviewButton({ voice, rate }) {
   ] });
 }
 function Row({ name, desc, children }) {
+  const label = FIELD_LABELS[name] ?? name;
   return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: setRow, children: [
     /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: setLabelBox, children: [
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: setLabel, children: name }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: setLabel, children: label }),
       /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: setHint, children: desc })
     ] }),
     /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { flexShrink: 0, maxWidth: 300 }, children })
   ] });
 }
+function Section({ title, children }) {
+  return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { marginTop: 2 }, children: [
+    /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { display: "flex", alignItems: "center", gap: 8, padding: "10px 0 2px" }, children: [
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 11, fontWeight: 700, letterSpacing: ".06em", textTransform: "uppercase", color: "var(--dsw-alias-label-secondary)" }, children: title }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { flex: 1, height: 1, background: t2.border } })
+    ] }),
+    children
+  ] });
+}
 function SegGroup({
   score,
   field,
   value,
-  options
+  options,
+  onSelect
 }) {
-  return /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { role: "group", style: setSeg, children: options.map((o) => /* @__PURE__ */ (0, import_jsx_runtime.jsx)("button", { style: setSegBtn(value === o.v), "aria-pressed": value === o.v, onClick: () => void score.set(field, o.v), children: o.label }, String(o.v))) });
+  return /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { role: "group", style: setSeg, children: options.map((o) => /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+    "button",
+    {
+      style: setSegBtn(value === o.v),
+      "aria-pressed": value === o.v,
+      onClick: () => {
+        void score.set(field, o.v);
+        onSelect?.(o.v);
+      },
+      children: o.label
+    },
+    String(o.v)
+  )) });
 }
-function VoiceSettingsCard({ scope }) {
-  const [snap, setSnap] = (0, import_react.useState)(() => scope.getSnapshot());
-  const [collapsed, setCollapsed] = (0, import_react.useState)(true);
-  (0, import_react.useEffect)(
-    () => scope.subscribe(() => {
-      setSnap({ ...scope.getSnapshot() });
-    }),
-    [scope]
-  );
-  const value = snap?.value ?? {};
-  const unavailable = snap?.status === "unavailable" || snap?.status === "error";
-  if (unavailable) {
-    return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { "data-dshvm-settings": "card", style: { color: t2.term, fontSize: 12, padding: "14px 16px", ...cardStyle }, children: [
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { color: "var(--dsw-alias-state-error-primary)" }, children: t("configUnavailable") }),
-      t("configUnavailableNote")
-    ] });
+var fmtMB = (b) => b >= 1048576 ? `${(b / 1048576).toFixed(0)}MB` : b > 0 ? `${Math.round(b / 1024)}KB` : "\u2013";
+function EngineStatusInline() {
+  const [st, setSt] = (0, import_react.useState)(null);
+  const [cleaning, setCleaning] = (0, import_react.useState)(false);
+  (0, import_react.useEffect)(() => {
+    let alive = true;
+    const poll = async () => {
+      try {
+        const res = await fetch(location.origin + BASE_PATH + "/models/status");
+        if (res.ok && alive) setSt(await res.json());
+      } catch {
+      }
+    };
+    void poll();
+    const timer = setInterval(() => void poll(), 3e3);
+    return () => {
+      alive = false;
+      clearInterval(timer);
+    };
+  }, []);
+  const clean = (engine) => {
+    setCleaning(true);
+    void fetch(location.origin + BASE_PATH + "/models/clean", {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify({ engine })
+    }).catch(() => void 0).finally(() => setTimeout(() => setCleaning(false), 1500));
+  };
+  const tts = st?.tts;
+  if (!tts) return /* @__PURE__ */ (0, import_jsx_runtime.jsx)(import_jsx_runtime.Fragment, {});
+  const engineName = tts.engine === "vits" ? t("engineVits") : tts.engine === "kokoro" ? t("engineKokoro") : t("engineEdge");
+  const isLocal = !!tts.local;
+  let statusText;
+  let statusColor;
+  if (tts.loading) {
+    statusText = t("engineLoading");
+    statusColor = t2.term;
+  } else if (tts.ready) {
+    statusText = t("engineReady");
+    statusColor = "var(--dsw-alias-state-success-primary)";
+  } else if (tts.error) {
+    statusText = t("engineError");
+    statusColor = "var(--dsw-alias-state-error-primary)";
+  } else if (isLocal && !tts.local.ready) {
+    statusText = t("ttsModelsMissing");
+    statusColor = "var(--dsw-alias-state-error-primary)";
+  } else {
+    statusText = t("engineIdle");
+    statusColor = t2.term;
   }
-  return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { "data-dshvm-settings": "card", style: cardStyle, children: [
-    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("style", { children: focusVisibleCss }),
+  return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { display: "flex", alignItems: "center", flexWrap: "wrap", gap: 8, padding: "2px 0 10px" }, children: [
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 12, color: t2.term }, children: engineName }),
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 12, fontWeight: tts.ready || tts.error ? 600 : 400, color: statusColor }, children: statusText }),
+    tts.loading && st?.progress?.file && /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: { fontSize: 12, color: t2.term }, children: [
+      st.progress.file,
+      " ",
+      st.progress.percent,
+      "%"
+    ] }),
+    isLocal && /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+      "button",
+      {
+        type: "button",
+        onClick: () => clean(tts.engine),
+        disabled: cleaning || tts.loading,
+        style: {
+          font: "inherit",
+          fontSize: 12,
+          cursor: tts.loading ? "default" : "pointer",
+          color: t2.label,
+          background: "var(--dsw-alias-bg-layer-2)",
+          border: `1px solid ${t2.border}`,
+          borderRadius: 8,
+          padding: "3px 10px",
+          flexShrink: 0
+        },
+        title: t("ttsRedownloadHint"),
+        children: cleaning ? t("ttsCleaning") : t("ttsRedownload")
+      }
+    ),
+    tts.error && /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 11, color: "var(--dsw-alias-state-error-primary)", flexBasis: "100%" }, children: tts.error })
+  ] });
+}
+function ModelStatusView() {
+  const [st, setSt] = (0, import_react.useState)(null);
+  const [retrying, setRetrying] = (0, import_react.useState)(null);
+  (0, import_react.useEffect)(() => {
+    let alive = true;
+    const poll = async () => {
+      try {
+        const res = await fetch(`${location.origin}${BASE_PATH}/models/status`);
+        if (res.ok && alive) setSt(await res.json());
+      } catch {
+      }
+    };
+    void poll();
+    const timer = setInterval(() => void poll(), 3e3);
+    return () => {
+      alive = false;
+      clearInterval(timer);
+    };
+  }, []);
+  const retry = (kind) => {
+    setRetrying(kind);
+    void fetch(`${location.origin}${BASE_PATH}/models/retry`, {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify({ kind })
+    }).catch(() => void 0).finally(() => {
+      setTimeout(() => setRetrying(null), 2e3);
+    });
+  };
+  const mkRow = (label, info, key, progressFor) => /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { display: "flex", alignItems: "center", gap: 10, padding: "6px 0" }, children: [
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { width: 92, flexShrink: 0, fontSize: 12, color: t2.label }, children: label }),
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { flex: 1, minWidth: 0 }, children: info.disabledText ? /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 12, color: t2.term }, children: info.disabledText }) : info.ready ? /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 12, color: "var(--dsw-alias-state-success-primary)", fontWeight: 600 }, children: t("modelsReady") }) : progressFor && progressFor.file ? /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: { fontSize: 12, color: t2.term }, children: [
+      t("modelsDownloading").replace("{file}", progressFor.file).replace("{percent}", String(progressFor.percent)),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { display: "block", height: 4, borderRadius: 99, background: t2.border, marginTop: 4, overflow: "hidden" }, children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { display: "block", height: "100%", width: `${progressFor.percent}%`, background: "var(--dsw-alias-brand-primary)", transition: "width .3s" } }) })
+    ] }) : info.failLatchMs !== void 0 && info.failLatchMs > 0 ? /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 12, color: "var(--dsw-alias-state-error-primary)" }, children: t("modelsFail").replace("{sec}", String(Math.ceil(info.failLatchMs / 1e3))) }) : /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: { fontSize: 12, color: t2.term }, children: [
+      fmtMB(info.size),
+      t("modelsMissing")
+    ] }) }),
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+      "button",
+      {
+        type: "button",
+        disabled: retrying === key || info.ready || !!info.disabledText,
+        onClick: () => retry(key),
+        style: {
+          font: "inherit",
+          fontSize: 12,
+          cursor: info.ready ? "default" : "pointer",
+          color: info.ready ? t2.term : t2.label,
+          background: "var(--dsw-alias-bg-layer-2)",
+          border: `1px solid ${t2.border}`,
+          borderRadius: 8,
+          padding: "3px 10px",
+          opacity: info.ready || info.disabledText ? 0.5 : 1,
+          flexShrink: 0
+        },
+        title: t("modelsRetryHint"),
+        children: retrying === key ? t("modelsRetrying") : t("modelsRetry")
+      }
+    )
+  ] });
+  const anyDownloading = !!st?.progress;
+  return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { marginTop: 4 }, children: [
+    /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { display: "flex", alignItems: "center", gap: 8, padding: "8px 0" }, children: [
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { fontSize: 13, fontWeight: 600, color: t2.label }, children: t("modelsTitle") }),
+      anyDownloading && st?.progress && /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: { fontSize: 12, color: t2.term }, children: [
+        st.progress.file,
+        " ",
+        st.progress.percent,
+        "%"
+      ] })
+    ] }),
+    mkRow(
+      t("modelStreamingAsr"),
+      { ready: !!st?.asr.ready, size: st?.asr.files.reduce((a, f) => a + f.size, 0) ?? 0, failLatchMs: st?.asr.failLatchMs ?? 0 },
+      "asr",
+      anyDownloading ? st.progress : null
+    ),
+    mkRow(t("modelVad"), { ready: !!st?.vad.ready, size: st?.vad.size ?? 0, failLatchMs: st?.vad.failLatchMs ?? 0 }, "vad", anyDownloading ? st.progress : null),
+    mkRow(
+      t("modelSense"),
+      {
+        ready: !!st?.sense.ready,
+        size: st?.sense.size ?? 0,
+        failLatchMs: st?.sense.enabled ? st?.sense.failLatchMs ?? 0 : 0,
+        disabledText: st?.sense.enabled ? void 0 : t("modelsDisabled")
+      },
+      "sense",
+      anyDownloading ? st.progress : null
+    ),
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("div", { style: { fontSize: 12, color: t2.term, lineHeight: "18px", padding: "4px 0 8px" }, children: t("modelsHint") })
+  ] });
+}
+function VoiceSettingsCard({ scope }) {
+  const [snap, setSnap] = (0, import_react.useState)(() => scope.getSnapshot());
+  const [collapsed, setCollapsed] = (0, import_react.useState)(true);
+  (0, import_react.useEffect)(
+    () => scope.subscribe(() => {
+      setSnap({ ...scope.getSnapshot() });
+    }),
+    [scope]
+  );
+  const value = snap?.value ?? {};
+  const unavailable = snap?.status === "unavailable" || snap?.status === "error";
+  const engine = value.ttsEngine === "edge" ? "edge" : value.ttsEngine === "kokoro" ? "kokoro" : "vits";
+  const [edgeVoices, setEdgeVoices] = (0, import_react.useState)(null);
+  (0, import_react.useEffect)(() => {
+    if (engine !== "edge") return;
+    let alive = true;
+    void fetch(location.origin + BASE_PATH + "/voices").then((res) => res.ok ? res.json() : null).then((data) => {
+      if (!alive || !data?.voices) return;
+      const localeName = (lk) => {
+        const m = /^([a-z]{2,3})(?:-[A-Z]{2})?/.exec(lk);
+        return m ? m[1].toUpperCase() : lk;
+      };
+      const genderName = (g) => g === "Female" ? "\u5973" : g === "Male" ? "\u7537" : g === "Neutral" ? "\u4E2D\u6027" : g;
+      setEdgeVoices(
+        data.voices.map((v) => ({
+          v: v.ShortName,
+          label: v.FriendlyName.replace(/ Microsoft.*/, "") + " \xB7 " + genderName(v.Gender) + " \xB7 " + localeName(v.Locale) + " (" + v.ShortName + ")"
+        })).sort((a, b) => a.label.localeCompare(b.label))
+      );
+    }).catch(() => void 0);
+    return () => {
+      alive = false;
+    };
+  }, [engine]);
+  const voiceOptions = engine === "edge" ? edgeVoices ?? VOICE_OPTIONS : engine === "kokoro" ? VOICE_OPTIONS_KOKORO : VOICE_OPTIONS_LOCAL;
+  if (unavailable) {
+    return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { "data-dshvm-settings": "card", style: { color: t2.term, fontSize: 12, padding: "14px 16px", ...cardStyle }, children: [
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { color: "var(--dsw-alias-state-error-primary)" }, children: t("configUnavailable") }),
+      t("configUnavailableNote")
+    ] });
+  }
+  return /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { "data-dshvm-settings": "card", style: cardStyle, children: [
+    /* @__PURE__ */ (0, import_jsx_runtime.jsx)("style", { children: focusVisibleCss }),
     /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("button", { type: "button", "aria-expanded": !collapsed, onClick: () => setCollapsed((c) => !c), style: { ...setHeader, background: collapsed ? "transparent" : t2.bgOpen }, children: [
       /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("span", { style: setHeadText, children: [
         /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: setName, children: t("stateVoiceMode") }),
@@ -992,59 +2029,205 @@ function VoiceSettingsCard({ scope }) {
       /* @__PURE__ */ (0, import_jsx_runtime.jsx)("span", { style: { ...setChevron, transform: collapsed ? "rotate(0deg)" : "rotate(180deg)" }, "aria-hidden": "true", children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("svg", { viewBox: "0 0 16 16", width: 14, height: 14, children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("path", { fill: "currentColor", d: "M4 6l4 4 4-4z" }) }) })
     ] }),
     !collapsed && /* @__PURE__ */ (0, import_jsx_runtime.jsx)("div", { style: setBody, children: /* @__PURE__ */ (0, import_jsx_runtime.jsxs)("div", { style: { marginTop: 4 }, children: [
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "voice", desc: t("descVoice"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
-        SelectField,
-        {
-          score: scope,
-          field: "voice",
-          value: value.voice ?? "",
-          options: VOICE_OPTIONS,
-          placeholder: "zh-CN-XiaoxiaoNeural",
-          footer: (v) => /* @__PURE__ */ (0, import_jsx_runtime.jsx)(VoicePreviewButton, { voice: v, rate: Number(value.rate ?? 1) })
-        }
-      ) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "rate", desc: t("descRate"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "rate", value: value.rate ?? 1, min: 0.5, max: 2, step: 0.1 }) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "interruptLevel", desc: t("descInterrupt"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
-        SegGroup,
-        {
-          score: scope,
-          field: "interruptLevel",
-          value: value.interruptLevel,
-          options: [
-            { v: 0, label: t("sev0") },
-            { v: 1, label: t("sev1") },
-            { v: 2, label: t("sev2") }
-          ]
-        }
-      ) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "silenceMs", desc: t("descSilence"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "silenceMs", value: value.silenceMs ?? 2e3, min: 500, max: 3e4, step: 100 }) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "idleTimeoutMinutes", desc: t("descIdle"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "idleTimeoutMinutes", value: value.idleTimeoutMinutes ?? 10, min: 1, max: 120, step: 1 }) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "modelHost", desc: t("descModelHost"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(SelectField, { score: scope, field: "modelHost", value: value.modelHost ?? "", options: HOST_OPTIONS, placeholder: "https://..." }) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "autoSend", desc: t("descAutoSend"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.autoSend), onChange: (e) => void scope.set("autoSend", e.target.checked) }) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "spokenFormat", desc: t("descSpokenFormat"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.spokenFormat), onChange: (e) => void scope.set("spokenFormat", e.target.checked) }) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "mode", desc: t("descMode"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
-        SegGroup,
-        {
-          score: scope,
-          field: "mode",
-          value: value.mode,
-          options: [
-            { v: "toggle", label: t("modeToggle") },
-            { v: "hold", label: t("modeHold") }
-          ]
-        }
-      ) }),
-      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "wakeWord", desc: t("descWakeWord"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(TextField, { score: scope, field: "wakeWord", value: value.wakeWord ?? "", placeholder: t("wakePlaceholder") }) })
+      /* @__PURE__ */ (0, import_jsx_runtime.jsxs)(Section, { title: t("secRead"), children: [
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "ttsEngine", desc: t("descTtsEngine"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+          SegGroup,
+          {
+            score: scope,
+            field: "ttsEngine",
+            value: engine,
+            options: [
+              { v: "vits", label: t("engineVits") },
+              { v: "kokoro", label: t("engineKokoro") },
+              { v: "edge", label: t("engineEdge") }
+            ],
+            onSelect: (v) => {
+              if (v !== engine) {
+                void scope.set("voice", ENGINE_DEFAULT_VOICE[String(v)] ?? "zh-CN-XiaoxiaoNeural");
+              }
+            }
+          }
+        ) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(EngineStatusInline, {}),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+          Row,
+          {
+            name: "voice",
+            desc: engine === "edge" ? t("descVoice") : engine === "kokoro" ? t("descVoiceKokoro") : t("descVoiceLocal"),
+            children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+              VoiceSelect,
+              {
+                score: scope,
+                field: "voice",
+                value: value.voice ?? "",
+                options: voiceOptions,
+                placeholder: ENGINE_DEFAULT_VOICE[engine] ?? "zh-CN-XiaoxiaoNeural",
+                showCustom: engine === "edge",
+                footer: (v) => /* @__PURE__ */ (0, import_jsx_runtime.jsx)(VoicePreviewButton, { voice: v, rate: Number(value.rate ?? 1) })
+              }
+            )
+          }
+        ),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "rate", desc: t("descRate"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "rate", value: value.rate ?? 1, min: 0.5, max: 2, step: 0.1 }) })
+      ] }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsxs)(Section, { title: t("secInterrupt"), children: [
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "interruptLevel", desc: t("descInterrupt"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+          SegGroup,
+          {
+            score: scope,
+            field: "interruptLevel",
+            value: value.interruptLevel,
+            options: [
+              { v: 0, label: t("sev0") },
+              { v: 1, label: t("sev1") },
+              { v: 2, label: t("sev2") }
+            ]
+          }
+        ) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "bargeInMode", desc: t("descBargeIn"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+          SegGroup,
+          {
+            score: scope,
+            field: "bargeInMode",
+            value: value.bargeInMode,
+            options: [
+              { v: "auto", label: t("bargeInAuto") },
+              { v: "manual", label: t("bargeInManual") }
+            ]
+          }
+        ) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "echoGateDb", desc: t("descEchoGate"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "echoGateDb", value: value.echoGateDb ?? 6, min: 3, max: 12, step: 1 }) })
+      ] }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsxs)(Section, { title: t("secInteraction"), children: [
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "mode", desc: t("descMode"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(
+          SegGroup,
+          {
+            score: scope,
+            field: "mode",
+            value: value.mode,
+            options: [
+              { v: "toggle", label: t("modeToggle") },
+              { v: "hold", label: t("modeHold") }
+            ]
+          }
+        ) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "shortcut", desc: t("descShortcut"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(TextField, { score: scope, field: "shortcut", value: value.shortcut ?? "Ctrl+Shift+V", placeholder: "Ctrl+Shift+V" }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "wakeWord", desc: t("descWakeWord"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(TextField, { score: scope, field: "wakeWord", value: value.wakeWord ?? "", placeholder: t("wakePlaceholder") }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "toolBeep", desc: t("descToolBeep"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.toolBeep), onChange: (e) => void scope.set("toolBeep", e.target.checked) }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "autoSend", desc: t("descAutoSend"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.autoSend), onChange: (e) => void scope.set("autoSend", e.target.checked) }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "autoResume", desc: t("descAutoResume"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.autoResume), onChange: (e) => void scope.set("autoResume", e.target.checked) }) })
+      ] }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsxs)(Section, { title: t("secRecognition"), children: [
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "senseVoice", desc: t("descSenseVoice"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.senseVoice), onChange: (e) => void scope.set("senseVoice", e.target.checked) }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "spokenFormat", desc: t("descSpokenFormat"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)("input", { type: "checkbox", checked: Boolean(value.spokenFormat), onChange: (e) => void scope.set("spokenFormat", e.target.checked) }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "silenceMs", desc: t("descSilence"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "silenceMs", value: value.silenceMs ?? 700, min: 500, max: 3e4, step: 100 }) }),
+        /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "idleTimeoutMinutes", desc: t("descIdle"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(NumberField, { score: scope, field: "idleTimeoutMinutes", value: value.idleTimeoutMinutes ?? 10, min: 1, max: 120, step: 1 }) })
+      ] }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Section, { title: t("secModel"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(Row, { name: "modelHost", desc: t("descModelHost"), children: /* @__PURE__ */ (0, import_jsx_runtime.jsx)(SelectField, { score: scope, field: "modelHost", value: value.modelHost ?? "", options: HOST_OPTIONS, placeholder: "https://..." }) }) }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)("div", { style: { fontSize: 12, color: t2.term, lineHeight: "18px", padding: "4px 0 8px" }, children: t("settingsEffectiveNote") }),
+      /* @__PURE__ */ (0, import_jsx_runtime.jsx)(ModelStatusView, {})
     ] }) })
   ] });
 }
 
 // src/client.tsx
 var import_jsx_runtime2 = require("react/jsx-runtime");
-var beepCtx = null;
+var isSpeechTrueCount = 0;
+var interruptFirstAt = 0;
+var isSpeechFalseRun = 0;
+var lastReenterAt = 0;
+var INT_CONFIRM_FRAMES = { 0: 3, 1: 2, 2: 1 };
 var inject = ["slots", "sessions", "settingsScope"];
+var TELEMETRY_VIEW = [
+  { stage: "utterance-end", key: "telUtteranceEnd" },
+  { stage: "endpoint-fired", key: "telEndpoint" },
+  { stage: "submitted", key: "telSubmitted" },
+  { stage: "first-llm-token", key: "telFirstToken" },
+  { stage: "first-sentence-text", key: "telFirstSentence" },
+  { stage: "first-tts-chunk", key: "telFirstChunk" },
+  { stage: "first-audio-played", key: "telFirstPlayed" }
+];
+var BUILD_TAG = "a76aad8";
+var TELEMETRY_FLAG = "dsh-voice-mode.telemetry";
+var telemetryEnabled = typeof localStorage !== "undefined" && localStorage.getItem(TELEMETRY_FLAG) === "1";
+console.log("[dsh-voice] build=" + BUILD_TAG);
+var debugLog = (event, fields = {}) => {
+  if (!telemetryEnabled) return;
+  const out = {};
+  for (const [k, v] of Object.entries(fields)) {
+    out[k] = typeof v === "number" && !Number.isInteger(v) ? Number(v.toFixed(4)) : v;
+  }
+  console.log("[dsh-voice]", event, JSON.stringify(out));
+};
+var beepCtx = null;
+function playToolBeep() {
+  try {
+    if (!beepCtx) beepCtx = new AudioContext();
+    void beepCtx.resume?.();
+    const ctx = beepCtx;
+    const osc = ctx.createOscillator();
+    const gain = ctx.createGain();
+    gain.gain.setValueAtTime(0.08, ctx.currentTime);
+    gain.gain.exponentialRampToValueAtTime(1e-3, ctx.currentTime + 0.1);
+    osc.connect(gain);
+    gain.connect(ctx.destination);
+    osc.start();
+    osc.stop(ctx.currentTime + 0.1);
+  } catch {
+  }
+}
+var SAMPLE_RATE_16K = 16e3;
+var ECHO_DELAY_MS = 0;
+var ECHO_TAIL_MS = 400;
 var WAVE_BARS = 14;
 var BASE_PATH2 = "/voice-mode";
+function getTabId() {
+  try {
+    const KEY = "dshvm-tabId";
+    let id = sessionStorage.getItem(KEY);
+    if (!id) {
+      id = typeof crypto !== "undefined" && "randomUUID" in crypto ? crypto.randomUUID() : Math.random().toString(36).slice(2) + Date.now().toString(36);
+      sessionStorage.setItem(KEY, id);
+    }
+    return id;
+  } catch {
+    return Math.random().toString(36).slice(2) + Date.now().toString(36);
+  }
+}
+var TAB_ID = getTabId();
+function parseShortcut(s) {
+  const parts = (s || "").split("+");
+  const mods = { ctrl: false, shift: false, alt: false, meta: false };
+  let key = "";
+  for (const raw of parts) {
+    const t3 = raw.trim().toLowerCase();
+    if (t3 === "ctrl" || t3 === "control") mods.ctrl = true;
+    else if (t3 === "shift") mods.shift = true;
+    else if (t3 === "alt" || t3 === "option") mods.alt = true;
+    else if (t3 === "meta" || t3 === "cmd" || t3 === "command") mods.meta = true;
+    else if (t3.length === 1 && /^[a-z0-9]$/.test(t3)) {
+      if (key) return null;
+      key = t3;
+    } else return null;
+  }
+  if (!key) return null;
+  if (!mods.ctrl && !mods.shift && !mods.alt && !mods.meta) return null;
+  return { ...mods, key };
+}
+function getLastVoiceSession() {
+  try {
+    return localStorage.getItem("dshvm-last-voice");
+  } catch {
+    return null;
+  }
+}
+function setLastVoiceSession(id) {
+  try {
+    if (id) localStorage.setItem("dshvm-last-voice", id);
+    else localStorage.removeItem("dshvm-last-voice");
+  } catch {
+  }
+}
 function apply(ctx) {
   const bus = createVoiceBus(void 0, ctx);
   ctx.slots.inject(
@@ -1089,6 +2272,7 @@ function apply(ctx) {
       () => ctx.slots.register(
         {
           name: "settings.plugin.item",
+          id: "voice-mode",
           key: "voice-mode",
           order: 100,
           label: t("stateVoiceMode")
@@ -1098,74 +2282,173 @@ function apply(ctx) {
     );
   }
 }
-function createAudioEngine(setUi) {
-  const queue = [];
-  const audio = new Audio();
-  const playNext = () => {
-    const frame = queue.shift();
+function createAudioEngine(setUi, onPlayed, onPlaybackRef, onAllPlayed) {
+  const pending = [];
+  const fallbackAudio = new Audio();
+  let fallback = false;
+  let ctx = null;
+  let duckGain = null;
+  let nextEndAt = 0;
+  const activeSrcs = /* @__PURE__ */ new Set();
+  let decoding = false;
+  const captionQueue = [];
+  const warm = () => {
+    if (ctx) {
+      void ctx.resume?.();
+      return;
+    }
+    try {
+      const AC = window.AudioContext ?? window.webkitAudioContext;
+      ctx = new AC();
+      duckGain = ctx.createGain();
+      duckGain.gain.value = 1;
+      duckGain.connect(ctx.destination);
+      void ctx.resume?.();
+    } catch {
+      ctx = null;
+    }
+  };
+  const playFallback = () => {
+    const frame = pending.shift() ?? null;
     if (!frame) {
       setUi({ playing: false, playingCaption: null });
       return;
     }
-    const bin = atob(frame.audio);
-    const bytes = new Uint8Array(bin.length);
-    for (let i = 0; i < bin.length; i++) bytes[i] = bin.charCodeAt(i);
-    const url = URL.createObjectURL(new Blob([bytes], { type: "audio/mpeg" }));
-    audio.src = url;
-    audio.onended = () => {
+    const url = URL.createObjectURL(new Blob([frame.audio], { type: frame.mime === "audio/wav" ? "audio/wav" : "audio/mpeg" }));
+    fallbackAudio.src = url;
+    fallbackAudio.onended = () => {
       URL.revokeObjectURL(url);
-      playNext();
+      playFallback();
     };
-    audio.onerror = () => {
+    fallbackAudio.onerror = () => {
       URL.revokeObjectURL(url);
-      playNext();
+      playFallback();
+    };
+    fallbackAudio.onplaying = () => {
+      try {
+        onPlayed?.();
+      } catch {
+      }
+      try {
+        if (ctx && frame.audio.length) {
+          void ctx.decodeAudioData(frame.audio.buffer.slice(0)).then((buf) => {
+            onPlaybackRef?.(buf.getChannelData(0), buf.sampleRate, performance.now());
+          }).catch(() => {
+          });
+        }
+      } catch {
+      }
     };
     setUi({ playing: true, playingCaption: frame.text, ttsNotice: null });
-    void audio.play().catch(() => playNext());
+    void fallbackAudio.play().catch(() => playFallback());
   };
-  const toolBeep = () => {
-    try {
-      if (!beepCtx) {
-        beepCtx = new AudioContext();
-        void beepCtx.resume?.();
+  const drainPending = () => {
+    if (decoding || !ctx || !duckGain || pending.length === 0) return;
+    decoding = true;
+    void (async () => {
+      try {
+        while (pending.length > 0) {
+          const frame = pending[0];
+          const buf = await ctx.decodeAudioData(frame.audio.buffer.slice(0));
+          if (pending.length === 0 || pending[0] !== frame) return;
+          pending.shift();
+          const t0 = ctx.currentTime;
+          const at = Math.max(t0 + 0.02, nextEndAt);
+          const src = ctx.createBufferSource();
+          src.buffer = buf;
+          src.connect(duckGain);
+          activeSrcs.add(src);
+          src.onended = () => {
+            activeSrcs.delete(src);
+            captionQueue.shift();
+            if (activeSrcs.size === 0 && pending.length === 0) {
+              setUi({ playing: false, playingCaption: null });
+              onAllPlayed?.();
+            } else if (captionQueue.length > 0) {
+              setUi({ playingCaption: captionQueue[0] });
+            }
+          };
+          src.start(at);
+          nextEndAt = at + buf.duration;
+          try {
+            const outLat = ctx.outputLatency ?? 0;
+            const wallMs = performance.now() + (at + outLat - ctx.currentTime) * 1e3;
+            onPlaybackRef?.(buf.getChannelData(0), buf.sampleRate, wallMs);
+          } catch {
+          }
+          try {
+            onPlayed?.();
+          } catch {
+          }
+          captionQueue.push(frame.text);
+          setUi({ playing: true, playingCaption: captionQueue[0], ttsNotice: null });
+        }
+      } catch {
+        for (const src of activeSrcs) {
+          try {
+            src.stop();
+          } catch {
+          }
+        }
+        activeSrcs.clear();
+        captionQueue.length = 0;
+        fallback = true;
+        playFallback();
+      } finally {
+        decoding = false;
       }
-      const osc = beepCtx.createOscillator();
-      const gain = beepCtx.createGain();
-      osc.frequency.value = 880;
-      gain.gain.setValueAtTime(0.08, beepCtx.currentTime);
-      gain.gain.exponentialRampToValueAtTime(1e-3, beepCtx.currentTime + 0.1);
-      osc.connect(gain);
-      gain.connect(beepCtx.destination);
-      osc.start();
-      osc.stop(beepCtx.currentTime + 0.1);
-    } catch {
-    }
+    })();
   };
   return {
     push(frame) {
-      queue.push(frame);
-      if (audio.paused) playNext();
+      if (fallback || !ctx) {
+        pending.push(frame);
+        if (fallbackAudio.paused) playFallback();
+        return;
+      }
+      pending.push(frame);
+      drainPending();
     },
     skip() {
-      queue.length = 0;
-      audio.pause();
-      audio.onended = null;
-      audio.onerror = null;
+      pending.length = 0;
+      nextEndAt = 0;
+      fallbackAudio.pause();
+      fallbackAudio.onended = null;
+      fallbackAudio.onerror = null;
+      for (const src of activeSrcs) {
+        try {
+          src.stop();
+        } catch {
+        }
+      }
+      activeSrcs.clear();
+      captionQueue.length = 0;
       setUi({ playing: false, playingCaption: null });
     },
-    toolBeep
+    warm,
+    unduck() {
+      if (!ctx || !duckGain) return;
+      const now = ctx.currentTime;
+      duckGain.gain.cancelScheduledValues(now);
+      duckGain.gain.setTargetAtTime(1, now, 0.035);
+    }
   };
 }
 function createVoiceBus(basePath = BASE_PATH2, ctx) {
   let activeSessionId = null;
   const DEFAULT_BOOT = {
     basePath: BASE_PATH2,
-    silenceMs: 2e3,
+    silenceMs: 700,
     interruptLevel: 0,
     idleTimeoutMinutes: 10,
     autoSend: true,
+    autoResume: false,
     mode: "toggle",
-    wakeWord: ""
+    bargeInMode: "auto",
+    echoGateDb: 6,
+    shortcut: "Ctrl+Shift+V",
+    wakeWord: "",
+    toolBeep: false
   };
   const ui = {
     state: "idle",
@@ -1178,16 +2461,142 @@ function createVoiceBus(basePath = BASE_PATH2, ctx) {
     ttsNotice: null,
     boot: DEFAULT_BOOT,
     mode: "toggle",
+    telemetry: null,
+    turn: "idle",
     wakeWord: ""
   };
   const listeners = /* @__PURE__ */ new Set();
   const audioListeners = /* @__PURE__ */ new Set();
-  const toolListeners = /* @__PURE__ */ new Set();
   let source = null;
-  const engine = createAudioEngine((patch) => {
-    Object.assign(ui, patch);
+  let playingEndAt = 0;
+  const telemetryStages = {};
+  const stampTelemetry = (stage, at) => {
+    if (!telemetryEnabled) return;
+    if (stage === "utterance-end") {
+      for (const k of Object.keys(telemetryStages)) delete telemetryStages[k];
+    }
+    if (telemetryStages[stage] === void 0) {
+      telemetryStages[stage] = at ?? Date.now();
+      ui.telemetry = { ...telemetryStages };
+      notify();
+    }
+  };
+  const resetTelemetry = () => {
+    if (!telemetryEnabled) return;
+    for (const k of Object.keys(telemetryStages)) delete telemetryStages[k];
+    ui.telemetry = null;
+    ui.interruptConfirmMs = void 0;
     notify();
-  });
+  };
+  const refChunks = [];
+  let refTotal = 0;
+  let refStartWall = 0;
+  let refActive = false;
+  const pushRef = (pcmSrc, srcRate, startWallMs) => {
+    const pcm = resampleLinear(pcmSrc, srcRate, SAMPLE_RATE_16K);
+    if (!refActive) {
+      refActive = true;
+      refStartWall = startWallMs;
+      refChunks.length = 0;
+      refTotal = 0;
+    }
+    const tailWall = refStartWall + refTotal / SAMPLE_RATE_16K * 1e3;
+    const gapMs = startWallMs - tailWall;
+    if (gapMs > 250) {
+      refChunks.length = 0;
+      refTotal = 0;
+      refStartWall = startWallMs;
+    } else if (gapMs > 1) {
+      const padN = Math.floor(gapMs / 1e3 * SAMPLE_RATE_16K);
+      refChunks.push(new Float32Array(padN));
+      refTotal += padN;
+    }
+    refChunks.push(pcm);
+    refTotal += pcm.length;
+    const maxTotal = SAMPLE_RATE_16K * 60;
+    while (refTotal - (refChunks[0]?.length ?? 0) > maxTotal) {
+      refTotal -= refChunks.shift().length;
+    }
+  };
+  const refWindowAt = (tWallMs, n) => {
+    const out = new Float32Array(n);
+    if (!refActive || refTotal === 0) return out;
+    const idx = Math.floor((tWallMs - ECHO_DELAY_MS - refStartWall) / 1e3 * SAMPLE_RATE_16K);
+    if (idx < 0 || idx >= refTotal) return out;
+    let acc = 0;
+    let outOff = 0;
+    for (const c of refChunks) {
+      if (outOff >= n) break;
+      if (idx >= acc + c.length) {
+        acc += c.length;
+        continue;
+      }
+      const start = Math.max(0, idx - acc);
+      const cnt = Math.min(c.length - start, n - outOff);
+      out.set(c.subarray(start, start + cnt), outOff);
+      outOff += cnt;
+      acc += c.length;
+    }
+    return out;
+  };
+  const aec = new NlmsAec({ filterLength: 1024, delay: 0 });
+  let refDelaySamples = 0;
+  let estMic = [];
+  let estRef = [];
+  const EST_CAP = SAMPLE_RATE_16K;
+  let lastEstimateAt = 0;
+  let echoBypass = false;
+  const echoSource = {
+    process: (mic, ref) => {
+      if (echoBypass) return mic;
+      const now = performance.now();
+      if (ui.playing) {
+        for (let i = 0; i < mic.length; i++) estMic.push(mic[i]);
+        for (let i = 0; i < ref.length; i++) estRef.push(ref[i]);
+        if (estMic.length > EST_CAP) {
+          const drop = estMic.length - EST_CAP;
+          estMic.splice(0, drop);
+          estRef.splice(0, drop);
+        }
+        if (now - lastEstimateAt > 2e3 && estMic.length > SAMPLE_RATE_16K * 0.5) {
+          lastEstimateAt = now;
+          const est = estimateBulkDelay(
+            Float32Array.from(estMic),
+            Float32Array.from(estRef),
+            { sampleRate: SAMPLE_RATE_16K, maxLag: Math.floor(0.25 * SAMPLE_RATE_16K) }
+          );
+          if (est.peak > 0.5) {
+            refDelaySamples = refDelaySamples === 0 ? est.lag : Math.round(refDelaySamples * 0.8 + est.lag * 0.2);
+          }
+          estMic.length = 0;
+          estRef.length = 0;
+        }
+      }
+      let refForAec = ref;
+      if (refDelaySamples > 0 && refActive && refTotal > refDelaySamples) {
+        const shiftMs = refDelaySamples / SAMPLE_RATE_16K * 1e3;
+        refForAec = refWindowAt(now - shiftMs, ref.length);
+      }
+      return aec.process(mic, refForAec);
+    },
+    windowAt: refWindowAt,
+    // A2.5 双讲冻结：用户说话时暂停 NLMS 自适应。
+    setFrozen: (frozen) => aec.setFrozen(frozen)
+  };
+  const engine = createAudioEngine(
+    (patch) => {
+      Object.assign(ui, patch);
+      notify();
+    },
+    () => stampTelemetry("first-audio-played"),
+    (pcm, sampleRate, wallMs) => pushRef(pcm, sampleRate, wallMs),
+    // Fix：自然播完（无 TTS 在播）即清参考池——AEC 不再拿旧回合参考适配新语音。
+    () => {
+      refActive = false;
+      refChunks.length = 0;
+      refTotal = 0;
+    }
+  );
   const notify = () => {
     for (const fn of listeners) {
       try {
@@ -1198,13 +2607,23 @@ function createVoiceBus(basePath = BASE_PATH2, ctx) {
   };
   const connect = () => {
     if (source) return;
-    source = new EventSource(`${location.origin}${basePath}/stream`);
+    source = new EventSource(`${location.origin}${basePath}/stream?tabId=${encodeURIComponent(TAB_ID)}`);
+    source.addEventListener("open", () => {
+      rejectSeqUpTo.clear();
+      lastFinalSeq.clear();
+    });
     source.addEventListener("mode", (e) => {
       try {
-        const active = JSON.parse(e.data).active ?? null;
-        if (active !== activeSessionId) {
-          activeSessionId = active;
-          if (active !== null || ui.playing) engine.skip();
+        const data = JSON.parse(e.data);
+        const active = data.active ?? null;
+        const ownerTabId = data.ownerTabId ?? null;
+        const preempted = active !== activeSessionId || ownerTabId !== null && activeSessionId !== null && ownerTabId !== TAB_ID;
+        if (activeSessionId !== null && preempted) {
+          const prev = activeSessionId;
+          activeSessionId = null;
+          if (ui.turn !== "idle") ui.turn = "idle";
+          doSkipAudio(prev);
+          resetTelemetry();
           notify();
         }
       } catch {
@@ -1223,18 +2642,23 @@ function createVoiceBus(basePath = BASE_PATH2, ctx) {
       } catch {
       }
     });
-    source.addEventListener("tool", (e) => {
+    source.addEventListener("turn", (e) => {
       try {
         const ev = JSON.parse(e.data);
-        for (const fn of toolListeners) {
-          try {
-            fn(ev);
-          } catch {
-          }
+        if (ev.sessionId === activeSessionId && ev.state) {
+          ui.turn = ev.state;
+          notify();
         }
       } catch {
       }
     });
+    source.addEventListener("latency", (e) => {
+      try {
+        const ev = JSON.parse(e.data);
+        if (ev.sessionId === activeSessionId && ev.stage) stampTelemetry(ev.stage);
+      } catch {
+      }
+    });
     source.addEventListener("asr-progress", (e) => {
       try {
         const p = JSON.parse(e.data);
@@ -1268,12 +2692,85 @@ function createVoiceBus(basePath = BASE_PATH2, ctx) {
       } catch {
       }
     });
+    source.addEventListener("tool", (e) => {
+      try {
+        const p = JSON.parse(e.data);
+        if (p.sessionId === activeSessionId && ui.boot.toolBeep === true) playToolBeep();
+      } catch {
+      }
+    });
   };
   connect();
+  const rejectSeqUpTo = /* @__PURE__ */ new Map();
+  const lastFinalSeq = /* @__PURE__ */ new Map();
+  let curSentenceId = null;
+  let curChunks = [];
+  let curBytes = 0;
+  let curChunkCount = 0;
   audioListeners.add((frame) => {
-    if (frame.sessionId === activeSessionId) engine.push(frame);
+    if (frame.sessionId !== activeSessionId) return;
+    const rejectLine = rejectSeqUpTo.get(frame.sessionId);
+    if (rejectLine !== void 0 && frame.sentenceId <= rejectLine) return;
+    stampTelemetry("first-tts-chunk");
+    if (frame.sentenceId !== curSentenceId) {
+      curSentenceId = frame.sentenceId;
+      curChunks = [];
+      curBytes = 0;
+      curChunkCount = 0;
+    }
+    if (frame.final) {
+      if (frame.chunkId !== curChunkCount) {
+        curSentenceId = null;
+        curChunks = [];
+        curBytes = 0;
+        curChunkCount = 0;
+        return;
+      }
+      const buf = new Uint8Array(curBytes);
+      let off = 0;
+      for (const c of curChunks) {
+        buf.set(c, off);
+        off += c.length;
+      }
+      curSentenceId = null;
+      curChunks = [];
+      curBytes = 0;
+      curChunkCount = 0;
+      if (buf.length === 0) return;
+      const isMp3 = buf[0] === 255;
+      const isWav = buf.length >= 4 && buf[0] === 82 && buf[1] === 73 && buf[2] === 70 && buf[3] === 70;
+      if (!isMp3 && !isWav) return;
+      engine.push({
+        sessionId: frame.sessionId,
+        seq: frame.sentenceId,
+        text: frame.text ?? "",
+        audio: buf,
+        mime: frame.mime
+      });
+      lastFinalSeq.set(frame.sessionId, frame.sentenceId);
+      return;
+    }
+    const bin = atob(frame.audio);
+    const bytes = new Uint8Array(bin.length);
+    for (let i = 0; i < bin.length; i++) bytes[i] = bin.charCodeAt(i);
+    curChunks.push(bytes);
+    curBytes += bytes.length;
+    curChunkCount += 1;
   });
-  toolListeners.add(() => engine.toolBeep());
+  const doSkipAudio = (sidArg) => {
+    const sid = sidArg ?? activeSessionId;
+    if (sid) {
+      rejectSeqUpTo.set(sid, Math.max(lastFinalSeq.get(sid) ?? -1, curSentenceId ?? -1));
+    }
+    curSentenceId = null;
+    curChunks = [];
+    curBytes = 0;
+    refActive = false;
+    refChunks.length = 0;
+    refTotal = 0;
+    engine.skip();
+    playingEndAt = 0;
+  };
   return {
     get activeSessionId() {
       return activeSessionId;
@@ -1287,34 +2784,47 @@ function createVoiceBus(basePath = BASE_PATH2, ctx) {
       };
     },
     setUi(patch) {
+      if (patch.playing === false && ui.playing === true) playingEndAt = Date.now();
       Object.assign(ui, patch);
       notify();
     },
+    /** isPlaying 尾音截止墙钟：playing 或尾音宽限期内均视为「AI 正在朗读」。 */
+    playingTailUntil() {
+      return playingEndAt + ECHO_TAIL_MS + refDelaySamples / SAMPLE_RATE_16K * 1e3;
+    },
     async enter(sessionId) {
       try {
         const res = await fetch(`${location.origin}${basePath}/toggle`, {
           method: "POST",
           headers: { "content-type": "application/json" },
-          body: JSON.stringify({ sessionId, on: true })
+          body: JSON.stringify({ sessionId, on: true, tabId: TAB_ID })
         });
         const out = await res.json();
-        activeSessionId = out.active ?? null;
+        activeSessionId = out.active === sessionId ? sessionId : null;
         notify();
         if (!res.ok) return { ok: false, error: out.error ?? t("enterFail") };
-        return { ok: out.active === sessionId, error: out.active === sessionId ? void 0 : t("enterFail") };
+        if (out.active === sessionId) setLastVoiceSession(sessionId);
+        return {
+          ok: out.active === sessionId,
+          preempted: out.active !== null && out.active !== sessionId,
+          error: out.active === sessionId ? void 0 : t("enterFail")
+        };
       } catch {
         return { ok: false, error: t("enterFail") };
       }
     },
     async exit(sessionId) {
+      resetTelemetry();
+      ui.turn = "idle";
+      doSkipAudio();
       try {
         const res = await fetch(`${location.origin}${basePath}/toggle`, {
           method: "POST",
           headers: { "content-type": "application/json" },
-          body: JSON.stringify({ sessionId, on: false })
+          body: JSON.stringify({ sessionId, on: false, tabId: TAB_ID })
         });
-        const out = await res.json();
-        activeSessionId = out.active ?? null;
+        await res.json();
+        activeSessionId = null;
         notify();
       } catch {
       }
@@ -1325,20 +2835,31 @@ function createVoiceBus(basePath = BASE_PATH2, ctx) {
         audioListeners.delete(fn);
       };
     },
-    onToolEvent(fn) {
-      toolListeners.add(fn);
-      return () => {
-        toolListeners.delete(fn);
-      };
-    },
     skipAudio() {
-      engine.skip();
+      doSkipAudio();
+    },
+    echoForAsr() {
+      return echoSource;
+    },
+    setEchoBypass(on) {
+      echoBypass = on;
+    },
+    echoDelayMs() {
+      return refDelaySamples / SAMPLE_RATE_16K * 1e3;
+    },
+    unduckAudio() {
+      engine.unduck();
     },
     cancelTurn(sessionId) {
       try {
         ctx?.sessions?.binding?.(sessionId)?.session.cancel?.();
       } catch {
       }
+    },
+    stampTelemetry,
+    resetTelemetry,
+    warmAudio() {
+      engine.warm();
     }
   };
 }
@@ -1372,8 +2893,12 @@ function MicButton({
   const submitTimerRef = (0, import_react2.useRef)(null);
   const idleTimerRef = (0, import_react2.useRef)(null);
   const runningRef = (0, import_react2.useRef)(false);
+  const mountedRef = (0, import_react2.useRef)(true);
   const holdCtrlRef = (0, import_react2.useRef)(false);
-  const bootNow = () => bus.ui.boot ?? { basePath: "/voice-mode", silenceMs: 2e3, interruptLevel: 0, idleTimeoutMinutes: 10, autoSend: true, mode: "toggle", wakeWord: "" };
+  const manualHoldRef = (0, import_react2.useRef)(false);
+  const breakRef = (0, import_react2.useRef)(null);
+  const pausedForHiddenRef = (0, import_react2.useRef)(false);
+  const bootNow = () => bus.ui.boot ?? { basePath: "/voice-mode", silenceMs: 700, interruptLevel: 0, idleTimeoutMinutes: 10, autoSend: true, autoResume: false, mode: "toggle", bargeInMode: "auto", echoGateDb: 6, shortcut: "Ctrl+Shift+V", wakeWord: "", toolBeep: false };
   useVoiceCss();
   const [, bumpUi] = (0, import_react2.useState)(0);
   (0, import_react2.useEffect)(
@@ -1398,8 +2923,13 @@ function MicButton({
         interruptLevel: c.interruptLevel ?? cur.interruptLevel,
         idleTimeoutMinutes: c.idleTimeoutMinutes ?? cur.idleTimeoutMinutes,
         autoSend: c.autoSend ?? cur.autoSend,
+        autoResume: c.autoResume === true,
         mode: c.mode === "hold" ? "hold" : "toggle",
-        wakeWord: c.wakeWord ?? cur.wakeWord
+        bargeInMode: c.bargeInMode === "manual" ? "manual" : "auto",
+        echoGateDb: typeof c.echoGateDb === "number" ? Math.min(12, Math.max(3, c.echoGateDb)) : cur.echoGateDb,
+        shortcut: typeof c.shortcut === "string" ? c.shortcut : cur.shortcut,
+        wakeWord: typeof c.wakeWord === "string" ? c.wakeWord : cur.wakeWord,
+        toolBeep: c.toolBeep === true
       };
       bus.setUi({ boot: next, mode: next.mode, wakeWord: next.wakeWord });
       return next;
@@ -1428,14 +2958,24 @@ function MicButton({
       if (bus.activeSessionId !== sid) {
         setLocalMode("off");
         clearIdle();
-        void engineRef.current?.stop();
+        cancelPendingSubmit();
+        clearBreakTimer();
+        setHolding(false);
+        isSpeechTrueCount = 0;
+        breakRef.current = null;
+        manualHoldRef.current = false;
+        holdCtrlRef.current = false;
+        const engine = engineRef.current;
         engineRef.current = null;
+        if (engine) void engine.stop();
+        bus.resetTelemetry();
+        bus.setUi({ state: "idle", partial: "", levels: [], error: null, model: null, ttsNotice: null, isSpeech: void 0 });
       }
     });
   }, [bus]);
   const cancelPendingSubmit = () => {
     if (submitTimerRef.current) {
-      clearTimeout(submitTimerRef.current);
+      clearInterval(submitTimerRef.current);
       submitTimerRef.current = null;
     }
   };
@@ -1444,33 +2984,175 @@ function MicButton({
     setLocalMode("off");
     clearIdle();
     cancelPendingSubmit();
+    isSpeechTrueCount = 0;
+    breakRef.current = null;
+    manualHoldRef.current = false;
+    clearBreakTimer();
+    setHolding(false);
     const engine = engineRef.current;
     engineRef.current = null;
-    if (engine) void engine.stop();
-    bus.setUi({ state: "idle", partial: "", levels: [], error: null, model: null, ttsNotice: null });
+    if (engine) await engine.stop();
+    bus.resetTelemetry();
+    bus.setUi({ state: "idle", partial: "", levels: [], error: null, model: null, ttsNotice: null, isSpeech: void 0 });
     const sid = sidRef.current;
-    if (sid) void bus.exit(sid);
+    if (sid) await bus.exit(sid);
   };
   const enterMode = async () => {
     const sid = sidRef.current;
     if (!sid || localRef.current !== "off") return;
+    isSpeechTrueCount = 0;
     setLocalMode("pending");
     try {
       const entered = await bus.enter(sid);
+      if (!mountedRef.current) {
+        if (entered.ok) void bus.exit(sid);
+        return;
+      }
       if (!entered.ok) {
         setLocalMode("off");
-        bus.setUi({
-          error: entered.error === "voice mode disabled" ? t("disabled") : entered.error ?? t("enterFail")
-        });
+        if (!entered.preempted) {
+          bus.setUi({
+            error: entered.error === "voice mode disabled" ? t("disabled") : entered.error ?? t("enterFail")
+          });
+        }
         return;
       }
       const cfg = await fetchConfig();
       const basePath = cfg.basePath;
       const silenceMs = cfg.silenceMs;
       const interruptLevel = cfg.interruptLevel;
-      const engine = createAsrEngine({ silenceMs, interruptLevel, basePath, wakeWord: cfg.wakeWord }, sid);
-      bus.setUi({ mode: cfg.mode, wakeWord: cfg.wakeWord });
+      const confirmFrames = INT_CONFIRM_FRAMES[interruptLevel] ?? 2;
+      const bargeInMode = cfg.bargeInMode;
+      debugLog("enter", {
+        build: BUILD_TAG,
+        mode: cfg.mode,
+        bargeInMode,
+        echoGateDb: cfg.echoGateDb,
+        interruptLevel,
+        silenceMs,
+        sessionId: sid
+      });
+      const hardBreak = async () => {
+        bus.skipAudio();
+        bus.unduckAudio();
+        if (runningRef.current && sidRef.current) {
+          bus.cancelTurn(sidRef.current);
+        }
+        const cancelP = fetch(`${location.origin}${BASE_PATH2}/cancel`, {
+          method: "POST",
+          headers: { "content-type": "application/json" },
+          // hold 按压中保留 host ASR 段（松手定稿续传前半句，防吃句）。
+          body: JSON.stringify({ sessionId: sidRef.current, keepAsr: engineRef.current?.holding === true }),
+          signal: AbortSignal.timeout(3e3)
+        }).catch(() => {
+        });
+        if (engineRef.current && !engineRef.current.holding) await engineRef.current.discardSegment();
+        await cancelP;
+        bus.setUi({ partial: "\u2026" });
+      };
+      breakRef.current = hardBreak;
+      const engine = createAsrEngine(
+        {
+          silenceMs,
+          basePath,
+          mode: cfg.mode,
+          wakeWord: cfg.wakeWord,
+          echoGateDb: cfg.echoGateDb,
+          echo: bus.echoForAsr(),
+          // 回声尾音宽限：playing 或尾音窗口内均视为朗读中，防句播完瞬间的残响漏入 ASR。
+          isPlaying: () => bus.ui.playing || Date.now() < bus.playingTailUntil(),
+          // 打断根治阶段二：服务端 Silero VAD 帧级检测下行 → 驱动打断（替代 RMS 能量快
+          // 路径）。连续 confirmFrames 次 true（墙钟节拍 100ms/拍，三档确认约 0.3/0.2/0.1s）
+          // 判真实人声前沿；仅 AI 朗读中（bus.ui.playing）触发 hardBreak，
+          // 防 TTS 回声被 VAD 误判为语音而自打断。
+          onIsSpeech: (speech) => {
+            if (bargeInMode === "manual") return;
+            if (!bus.ui.playing) {
+              isSpeechTrueCount = 0;
+              interruptFirstAt = 0;
+              bus.setUi({ isSpeech: speech, echoDelayMs: bus.echoDelayMs(), echoLevels: engineRef.current?.echoLevels() });
+              return;
+            }
+            if (speech === true && isSpeechTrueCount === 0) {
+              const lv = engineRef.current?.echoLevels();
+              debugLog("vad-speech-start", {
+                playing: bus.ui.playing,
+                delayMs: Math.round(bus.echoDelayMs()),
+                floor: lv?.floorRms,
+                resid: lv?.residualRms,
+                peak: lv?.peakRms
+              });
+            }
+            if (speech === true) {
+              isSpeechFalseRun = 0;
+              isSpeechTrueCount++;
+              if (isSpeechTrueCount === 1) interruptFirstAt = Date.now();
+              if (isSpeechTrueCount >= confirmFrames) {
+                if (engineRef.current && !engineRef.current.aboveEchoFloor(cfg.echoGateDb ?? 6)) {
+                  const lv2 = engineRef.current?.echoLevels();
+                  debugLog("echo-gate-reject", {
+                    gateDb: cfg.echoGateDb ?? 6,
+                    floor: lv2?.floorRms,
+                    resid: lv2?.residualRms,
+                    peak: lv2?.peakRms,
+                    confirmFrames
+                  });
+                  isSpeechTrueCount = 0;
+                  interruptFirstAt = 0;
+                  return;
+                }
+                const confirmMs = interruptFirstAt > 0 ? Date.now() - interruptFirstAt : 0;
+                const lv = engineRef.current?.echoLevels();
+                debugLog("interrupt-trigger", {
+                  confirmMs,
+                  floor: lv?.floorRms,
+                  resid: lv?.residualRms,
+                  peak: lv?.peakRms,
+                  delayMs: Math.round(bus.echoDelayMs())
+                });
+                interruptFirstAt = 0;
+                isSpeechTrueCount = 0;
+                resetIdle();
+                bus.resetTelemetry();
+                bus.setUi({ interruptConfirmMs: confirmMs });
+                void hardBreak();
+              }
+            } else {
+              isSpeechFalseRun++;
+              if (isSpeechFalseRun >= 2) {
+                isSpeechFalseRun = 0;
+                isSpeechTrueCount = Math.max(0, isSpeechTrueCount - 1);
+                if (isSpeechTrueCount === 0) interruptFirstAt = 0;
+              }
+            }
+            bus.setUi({ isSpeech: speech, echoDelayMs: bus.echoDelayMs(), echoLevels: engineRef.current?.echoLevels() });
+          },
+          onSessionExpired: async () => {
+            if (localRef.current !== "on") return false;
+            if (Date.now() - lastReenterAt < 2e3) return false;
+            lastReenterAt = Date.now();
+            bus.setUi({ error: t("sessionExpired") });
+            const reentered = await bus.enter(sid);
+            if (!reentered.ok) {
+              bus.setUi({ error: t("sessionExpiredFail") });
+            } else {
+              bus.setUi({ error: null });
+            }
+            return reentered.ok;
+          },
+          // A1：原生 AEC 生效状态 → 状态条提示（外放且原生 AEC 失效时引导用耳机/手动打断）。
+          onAecState: (on2) => {
+            debugLog("aec-state", { nativeEchoCancellation: on2 });
+            bus.setEchoBypass(on2);
+            bus.setUi({ aecOff: !on2 });
+          }
+        },
+        sid
+      );
+      bus.setUi({ mode: cfg.mode });
       engineRef.current = engine;
+      engine.onTelemetry((e) => bus.stampTelemetry(e.stage, e.at));
+      bus.warmAudio();
       try {
         if (!beepCtx) beepCtx = new AudioContext();
         void beepCtx.resume?.();
@@ -1491,6 +3173,7 @@ function MicButton({
       engine.onPartial((text) => bus.setUi({ partial: text }));
       engine.onSegment((text, meta) => {
         resetIdle();
+        bus.setUi({ partial: "" });
         const actions = actionsRef.current;
         const trimmed = text.trim();
         if (!trimmed) return;
@@ -1507,6 +3190,7 @@ function MicButton({
           } catch {
           }
         }
+        if (bus.ui.playing && !meta?.force) return;
         if (bootNow().autoSend === false && !meta?.force) return;
         const doSubmit = () => {
           try {
@@ -1522,29 +3206,35 @@ function MicButton({
         };
         cancelPendingSubmit();
         doSubmit();
-        submitTimerRef.current = setTimeout(() => {
+        let retryCount = 0;
+        submitTimerRef.current = setInterval(() => {
+          retryCount++;
           const phase = phaseRef.current;
-          if (phase !== "submitting" && phase !== "adjudicating" && draftRef.current.trim()) doSubmit();
-        }, 800);
-      });
-      engine.onSpeechStart(async () => {
-        resetIdle();
-        bus.skipAudio();
-        try {
-          await fetch(`${location.origin}${BASE_PATH2}/cancel`, {
-            method: "POST",
-            headers: { "content-type": "application/json" },
-            body: JSON.stringify({ sessionId: sidRef.current })
-          });
-        } catch {
-        }
-        if (runningRef.current && sidRef.current) {
-          bus.cancelTurn(sidRef.current);
-        }
-        bus.setUi({ partial: "\u2026" });
+          if (retryCount > 3 || phase === "submitting" || phase === "adjudicating" || draftRef.current.trim() !== trimmed) {
+            cancelPendingSubmit();
+            return;
+          }
+          doSubmit();
+        }, 500);
       });
       bus.setUi({ state: "idle", partial: "", levels: [], error: null, model: null, ttsNotice: null });
+      if (!mountedRef.current) {
+        engineRef.current = null;
+        void bus.exit(sid);
+        return;
+      }
       await engine.start();
+      if (!mountedRef.current) {
+        engineRef.current = null;
+        await engine.stop();
+        void bus.exit(sid);
+        return;
+      }
+      if (bus.activeSessionId !== sid) {
+        engineRef.current = null;
+        await engine.stop();
+        return;
+      }
       setLocalMode("on");
       resetIdle();
     } catch (e) {
@@ -1569,21 +3259,40 @@ function MicButton({
   (0, import_react2.useEffect)(() => {
     sidRef.current = sessionId;
   }, [sessionId]);
+  const autoResumeTriedForRef = (0, import_react2.useRef)(null);
+  (0, import_react2.useEffect)(() => {
+    const sid = sessionId;
+    if (!sid || sid === autoResumeTriedForRef.current) return;
+    autoResumeTriedForRef.current = sid;
+    void (async () => {
+      const cfg = await fetchConfig();
+      if (!cfg.autoResume) return;
+      if (getLastVoiceSession() !== sid) return;
+      if (bus.activeSessionId !== null) return;
+      if (localRef.current !== "off") return;
+      await enterMode().catch(() => {
+        setLocalMode("off");
+      });
+    })();
+  }, [sessionId]);
   const runningSel = useSession ? useSession((s) => s === void 0 ? void 0 : s.running) : void 0;
   (0, import_react2.useEffect)(() => {
     runningRef.current = runningSel === true;
   }, [runningSel]);
   (0, import_react2.useEffect)(() => {
+    mountedRef.current = true;
     return () => {
+      mountedRef.current = false;
       clearIdle();
       cancelPendingSubmit();
+      isSpeechTrueCount = 0;
       const sid = sidRef.current;
-      if (localRef.current === "on" && sid) {
+      if ((localRef.current === "on" || localRef.current === "pending") && sid) {
         void engineRef.current?.stop();
         void fetch(`${location.origin}${BASE_PATH2}/toggle`, {
           method: "POST",
           headers: { "content-type": "application/json" },
-          body: JSON.stringify({ sessionId: sid, on: false }),
+          body: JSON.stringify({ sessionId: sid, on: false, tabId: TAB_ID }),
           keepalive: true
         }).catch(() => {
         });
@@ -1592,6 +3301,8 @@ function MicButton({
   }, []);
   (0, import_react2.useEffect)(() => {
     let ctrlTimer = null;
+    let ctrlHoldStart = 0;
+    let otherKeyDuringCtrl = false;
     const cancelCtrl = () => {
       if (ctrlTimer) {
         clearTimeout(ctrlTimer);
@@ -1599,33 +3310,68 @@ function MicButton({
       }
       if (holdCtrlRef.current) {
         holdCtrlRef.current = false;
+        setHolding(false);
+        engineRef.current?.endHeld(false);
+      }
+      if (manualHoldRef.current) {
+        manualHoldRef.current = false;
+        setHolding(false);
         engineRef.current?.endHeld(false);
       }
     };
     const onKeyDown = (e) => {
-      if (e.ctrlKey && e.shiftKey && e.key.toLowerCase() === "v" && !e.repeat) {
-        e.preventDefault();
-        cancelCtrl();
-        toggleRef.current();
+      const combo = parseShortcut(bootNow().shortcut);
+      const codeKey = e.code.replace("Key", "").replace("Digit", "").toLowerCase();
+      if (combo && !e.repeat && (e.key.toLowerCase() === combo.key || codeKey === combo.key) && e.ctrlKey === combo.ctrl && e.shiftKey === combo.shift && e.altKey === combo.alt && e.metaKey === combo.meta) {
+        const el = e.target;
+        const editable = el instanceof HTMLElement && (el.tagName === "TEXTAREA" || el.tagName === "INPUT" || el.isContentEditable);
+        if (!editable && !e.isComposing) {
+          e.preventDefault();
+          cancelCtrl();
+          toggleRef.current();
+        }
         return;
       }
       const eng = engineRef.current;
-      if (e.key !== "Control" || e.shiftKey || e.altKey || e.metaKey || e.repeat || !eng) return;
-      if (bootNow().mode === "hold") {
-        ctrlTimer = setTimeout(() => {
-          ctrlTimer = null;
-          holdCtrlRef.current = true;
+      if (e.key === "Control" && !e.shiftKey && !e.altKey && !e.metaKey && !e.repeat && eng) {
+        ctrlHoldStart = Date.now();
+        otherKeyDuringCtrl = false;
+        if (bootNow().mode === "hold") {
+          ctrlTimer = setTimeout(() => {
+            ctrlTimer = null;
+            holdCtrlRef.current = true;
+            setHolding(true);
+            eng.beginHeld();
+          }, 600);
+        } else if (bootNow().bargeInMode === "manual" && bus.ui.playing) {
+          manualHoldRef.current = true;
+          setHolding(true);
           eng.beginHeld();
-        }, 600);
-      } else {
-        eng.forceSend();
+          void breakRef.current?.();
+        }
+        return;
+      }
+      if (ctrlHoldStart > 0 && e.key !== "Control") {
+        otherKeyDuringCtrl = true;
+        if (ctrlTimer) {
+          clearTimeout(ctrlTimer);
+          ctrlTimer = null;
+        }
       }
     };
     const onKeyUp = (e) => {
-      if (e.key === "Control") cancelCtrl();
+      if (e.key !== "Control") return;
+      if (bootNow().mode !== "hold" && !manualHoldRef.current && !otherKeyDuringCtrl && ctrlHoldStart > 0 && Date.now() - ctrlHoldStart >= 250) {
+        engineRef.current?.forceSend();
+      }
+      cancelCtrl();
+      ctrlHoldStart = 0;
+      otherKeyDuringCtrl = false;
     };
     const onBlur = () => {
       cancelCtrl();
+      ctrlHoldStart = 0;
+      otherKeyDuringCtrl = false;
       if (localRef.current === "on" && bootNow().mode === "hold") engineRef.current?.endHeld(true);
     };
     window.addEventListener("keydown", onKeyDown);
@@ -1654,12 +3400,25 @@ function MicButton({
       if (localRef.current !== "on" || bootNow().mode !== "hold") return;
       engineRef.current?.endHeld(true);
       holdCtrlRef.current = false;
+      setHolding(false);
       bus.setUi({ partial: "" });
     };
     const onVisibility = () => {
-      if (document.hidden && bootNow().mode === "hold") {
-        engineRef.current?.endHeld(true);
-        holdCtrlRef.current = false;
+      if (document.hidden) {
+        if (localRef.current === "on" && engineRef.current) {
+          if (bootNow().mode === "hold") {
+            engineRef.current?.endHeld(true);
+            holdCtrlRef.current = false;
+            setHolding(false);
+          }
+          pausedForHiddenRef.current = true;
+          void engineRef.current.stop();
+        }
+      } else if (pausedForHiddenRef.current && localRef.current === "on") {
+        pausedForHiddenRef.current = false;
+        void engineRef.current?.start().catch(() => {
+          setLocalMode("off");
+        });
       }
     };
     window.addEventListener("keydown", onKeyDown);
@@ -1690,28 +3449,76 @@ function MicButton({
   const livePhase = useInput ? useInput((s) => s?.phase ?? "") : "";
   const phaseRef = (0, import_react2.useRef)("");
   phaseRef.current = livePhase;
-  const label = on ? busy ? t("recognizing") : holdMode ? t("holdToTalk") : t("voiceDetected") : local === "pending" ? t("entering") : t("voiceBtn");
+  const [holding, setHolding] = (0, import_react2.useState)(false);
+  const label = on ? busy ? t("recognizing") : holdMode ? holding ? t("releaseToSend") : t("holdToTalk") : bus.ui.state === "wake" ? t("sayWake").replace("{wake}", bus.ui.wakeWord || t("wakeWord")) : t("voiceDetected") : local === "pending" ? t("entering") : t("voiceBtn");
   const holdPtrRef = (0, import_react2.useRef)(null);
+  const toggleHoldRef = (0, import_react2.useRef)(false);
+  const suppressClickUntilRef = (0, import_react2.useRef)(0);
+  const breakTimerRef = (0, import_react2.useRef)(null);
+  const clearBreakTimer = () => {
+    if (breakTimerRef.current !== null) {
+      clearTimeout(breakTimerRef.current);
+      breakTimerRef.current = null;
+    }
+  };
   const onPointerDown = (e) => {
-    if (bootNow().mode !== "hold") return;
     holdPtrRef.current = { t: Date.now(), y: e.clientY, id: e.pointerId };
     e.currentTarget.setPointerCapture?.(e.pointerId);
-    if (localRef.current === "on") engineRef.current?.beginHeld();
+    if (bootNow().mode === "hold") {
+      if (localRef.current === "on") {
+        setHolding(true);
+        const eng = engineRef.current;
+        eng?.beginHeld();
+        if (eng && bootNow().bargeInMode === "manual" && bus.ui.playing) {
+          breakTimerRef.current = setTimeout(() => {
+            breakTimerRef.current = null;
+            if (holdPtrRef.current && bus.ui.playing) void breakRef.current?.();
+          }, 250);
+        }
+      }
+    } else if (localRef.current === "on" && bus.ui.playing) {
+      toggleHoldRef.current = true;
+      setHolding(true);
+      const eng = engineRef.current;
+      eng?.beginHeld();
+      breakTimerRef.current = setTimeout(() => {
+        breakTimerRef.current = null;
+        if (holdPtrRef.current && bus.ui.playing) void breakRef.current?.();
+      }, 250);
+    }
   };
   const onPointerMove = (e) => {
     const p = holdPtrRef.current;
     if (!p || p.id !== e.pointerId) return;
     if (p.y - e.clientY >= 40) {
       holdPtrRef.current = null;
+      toggleHoldRef.current = false;
+      clearBreakTimer();
+      setHolding(false);
       engineRef.current?.endHeld(true);
       bus.setUi({ partial: "" });
+      suppressClickUntilRef.current = Date.now() + 500;
     }
   };
   const onPointerUp = (e) => {
     const p = holdPtrRef.current;
     holdPtrRef.current = null;
+    clearBreakTimer();
+    setHolding(false);
     if (!p || p.id !== e.pointerId) return;
     const ms = Date.now() - p.t;
+    if (bootNow().mode !== "hold") {
+      if (toggleHoldRef.current) {
+        toggleHoldRef.current = false;
+        if (ms >= 250) {
+          suppressClickUntilRef.current = Date.now() + 500;
+          engineRef.current?.endHeld(false);
+        } else {
+          engineRef.current?.endHeld(true);
+        }
+      }
+      return;
+    }
     if (ms < 250) {
       if (localRef.current === "on") {
         engineRef.current?.endHeld(true);
@@ -1725,12 +3532,16 @@ function MicButton({
   };
   const onPointerCancel = () => {
     holdPtrRef.current = null;
+    toggleHoldRef.current = false;
+    clearBreakTimer();
+    setHolding(false);
     engineRef.current?.endHeld(true);
   };
   return /* @__PURE__ */ (0, import_jsx_runtime2.jsxs)(
     "button",
     {
       onClick: (e) => {
+        if (Date.now() < suppressClickUntilRef.current) return;
         if (holdMode) {
           if (e.detail !== 0) return;
         }
@@ -1740,29 +3551,34 @@ function MicButton({
       onPointerMove,
       onPointerUp,
       onPointerCancel,
+      onContextMenu: (e) => e.preventDefault(),
       "data-dshvm": "mic",
       "aria-label": on ? t("ariaActive") : t("ariaEnter"),
       "aria-pressed": on,
       title: on ? holdMode ? t("titleHold") : t("titleToggle") : t("titleEnter"),
       style: {
-        border: "none",
-        background: on ? holdMode ? "rgba(88, 166, 255, 0.16)" : "rgba(63, 185, 80, 0.16)" : local === "pending" ? "rgba(88, 166, 255, 0.14)" : "transparent",
+        border: holding ? "1px solid rgba(248, 81, 73, 0.6)" : on ? holdMode ? "1px solid rgba(88, 166, 255, 0.45)" : "1px solid rgba(63, 185, 80, 0.45)" : "1px solid rgba(139, 148, 158, 0.35)",
+        background: holding ? "rgba(248, 81, 73, 0.2)" : on ? holdMode ? "rgba(88, 166, 255, 0.16)" : "rgba(63, 185, 80, 0.16)" : local === "pending" ? "rgba(88, 166, 255, 0.14)" : "rgba(139, 148, 158, 0.08)",
         cursor: "pointer",
-        padding: "4px 8px",
+        padding: "5px 10px",
         borderRadius: 8,
         display: "flex",
         alignItems: "center",
         gap: 6,
-        fontSize: 11,
+        fontSize: 12,
         fontFamily: "system-ui, sans-serif",
-        color: on ? holdMode ? "#58a6ff" : "#3fb950" : local === "pending" ? "#58a6ff" : "#8b949e",
-        transition: "background 0.15s ease, color 0.2s ease",
+        color: holding ? "#f85149" : on ? holdMode ? "#58a6ff" : "#3fb950" : local === "pending" ? "#58a6ff" : "#8b949e",
+        transition: "background 0.15s ease, color 0.2s ease, border-color 0.15s ease",
         touchAction: "none",
         // 触摸设备上让 pointer 事件独占（滑出取消可用）
-        userSelect: "none"
+        userSelect: "none",
+        WebkitUserSelect: "none",
+        // iOS Safari 前缀，防长按选中文字
+        WebkitTouchCallout: "none"
+        // iOS 长按弹出「拷贝/选择」菜单
       },
       children: [
-        /* @__PURE__ */ (0, import_jsx_runtime2.jsxs)("svg", { viewBox: "0 0 24 24", width: 14, height: 14, "aria-hidden": "true", children: [
+        /* @__PURE__ */ (0, import_jsx_runtime2.jsxs)("svg", { viewBox: "0 0 24 24", width: 16, height: 16, "aria-hidden": "true", children: [
           /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
             "path",
             {
@@ -1790,15 +3606,38 @@ function VoiceStatusBar({ bus, sessionId }) {
   }, [bus]);
   const isActive = b.active === sessionId;
   if (!isActive) return /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(import_jsx_runtime2.Fragment, {});
-  const stateText = b.ui.state === "loading-model" ? t("loadingModel") : b.ui.state === "transcribing" ? t("recognizing") : b.ui.state === "speech" ? b.ui.mode === "hold" ? t("holdDots") : t("listening") : b.ui.state === "wake" ? t("sayWake").replace("{wake}", b.ui.wakeWord || t("wakeWord")) : b.ui.mode === "hold" ? t("barHold") : t("barListening");
+  const stateText = b.ui.state === "loading-model" ? t("loadingModel") : b.ui.state === "transcribing" ? t("recognizing") : b.ui.state === "wake" ? t("sayWake").replace("{wake}", b.ui.wakeWord || t("wakeWord")) : b.ui.state === "speech" ? b.ui.mode === "hold" ? t("holdDots") : t("listening") : b.ui.playing ? t("reading") : b.ui.turn === "agent-speaking" ? t("thinking") : b.ui.mode === "hold" ? t("barHold") : t("barListening");
   const bars = Array.from({ length: WAVE_BARS }, (_, i) => b.ui.levels[i] ?? 0);
+  const telParts = [];
+  const fmt = (ms) => ms >= 1e3 ? `${(ms / 1e3).toFixed(2)}s` : `${Math.round(ms)}ms`;
+  const tel = b.ui.telemetry;
+  if (tel) {
+    for (let i = 1; i < TELEMETRY_VIEW.length; i++) {
+      const cur = tel[TELEMETRY_VIEW[i].stage];
+      const prev = tel[TELEMETRY_VIEW[i - 1].stage];
+      if (cur === void 0 || prev === void 0) continue;
+      telParts.push(`${t(TELEMETRY_VIEW[i].key)} ${fmt(cur - prev)}`);
+    }
+    const begin = tel["utterance-end"];
+    const end = tel["first-audio-played"];
+    if (begin !== void 0 && end !== void 0) telParts.push(`${t("telTotal")} ${fmt(end - begin)}`);
+  }
+  if (b.ui.interruptConfirmMs !== void 0) {
+    telParts.push(`${t("interruptConfirm")} ${fmt(b.ui.interruptConfirmMs)}`);
+  }
+  if (b.ui.echoLevels) {
+    const el = b.ui.echoLevels;
+    telParts.push(
+      `AEC delay=${Math.round(b.ui.echoDelayMs ?? 0)}ms floor=${el.floorRms.toFixed(4)} resid=${el.residualRms.toFixed(4)}`
+    );
+  }
   return /* @__PURE__ */ (0, import_jsx_runtime2.jsxs)(
     "div",
     {
       style: {
         display: "flex",
-        alignItems: "center",
-        gap: 8,
+        flexDirection: "column",
+        gap: 2,
         padding: "6px 12px",
         borderRadius: 10,
         fontSize: 12,
@@ -1809,34 +3648,84 @@ function VoiceStatusBar({ bus, sessionId }) {
         animation: "dshvm-fadein 0.2s ease"
       },
       children: [
-        /* @__PURE__ */ (0, import_jsx_runtime2.jsx)("span", { style: { display: "inline-flex", alignItems: "flex-end", gap: 2, height: 14, flexShrink: 0 }, children: bars.map((v, i) => /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
-          "span",
-          {
-            className: "dshvm-bar",
-            style: {
-              height: `${3 + v * 12}px`,
-              background: "#3fb950",
-              opacity: 0.4 + v * 0.6
+        /* @__PURE__ */ (0, import_jsx_runtime2.jsxs)("div", { style: { display: "flex", alignItems: "center", gap: 8 }, children: [
+          /* @__PURE__ */ (0, import_jsx_runtime2.jsx)("span", { style: { display: "inline-flex", alignItems: "flex-end", gap: 2, height: 14, flexShrink: 0 }, children: bars.map((v, i) => /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
+            "span",
+            {
+              className: "dshvm-bar",
+              style: {
+                height: `${3 + v * 12}px`,
+                background: "#3fb950",
+                opacity: 0.4 + v * 0.6
+              }
+            },
+            i
+          )) }),
+          /* @__PURE__ */ (0, import_jsx_runtime2.jsx)("span", { style: { overflow: "hidden", textOverflow: "ellipsis", whiteSpace: "nowrap", flexGrow: 1 }, children: b.ui.error ? b.ui.error : b.ui.state === "loading-model" || b.ui.model ? b.ui.model ? `${t("loadingModel")} ${b.ui.model.file} ${b.ui.model.percent}%` : stateText : b.ui.playing || b.ui.turn === "agent-speaking" ? stateText : b.ui.partial ? b.ui.partial : b.ui.ttsNotice ? b.ui.ttsNotice : stateText }),
+          b.ui.isSpeech === true && /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
+            "span",
+            {
+              title: t("vadDetected"),
+              style: {
+                flexShrink: 0,
+                padding: "0 6px",
+                borderRadius: 8,
+                fontSize: 10,
+                lineHeight: "16px",
+                color: "#ffa657",
+                background: "rgba(255, 166, 87, 0.15)",
+                border: "1px solid rgba(255, 166, 87, 0.35)"
+              },
+              children: t("vadDetected")
             }
-          },
-          i
-        )) }),
-        /* @__PURE__ */ (0, import_jsx_runtime2.jsx)("span", { style: { overflow: "hidden", textOverflow: "ellipsis", whiteSpace: "nowrap", flexGrow: 1 }, children: b.ui.error ? b.ui.error : b.ui.state === "loading-model" || b.ui.model ? b.ui.model ? `${t("loadingModel")} ${b.ui.model.file} ${b.ui.model.percent}%` : stateText : b.ui.partial ? b.ui.partial : b.ui.ttsNotice ? b.ui.ttsNotice : stateText }),
-        /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
-          "button",
+          ),
+          b.ui.aecOff === true && /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
+            "span",
+            {
+              title: t("aecOffHint"),
+              style: {
+                flexShrink: 0,
+                padding: "0 6px",
+                borderRadius: 8,
+                fontSize: 10,
+                lineHeight: "16px",
+                color: "#ffa657",
+                background: "rgba(255, 166, 87, 0.15)",
+                border: "1px solid rgba(255, 166, 87, 0.35)"
+              },
+              children: t("aecOff")
+            }
+          ),
+          /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
+            "button",
+            {
+              onClick: () => {
+                void bus.exit(sessionId);
+              },
+              style: {
+                border: "none",
+                background: "transparent",
+                color: "#8b949e",
+                cursor: "pointer",
+                fontSize: 12,
+                flexShrink: 0
+              },
+              children: t("exit")
+            }
+          )
+        ] }),
+        telParts.length > 0 && /* @__PURE__ */ (0, import_jsx_runtime2.jsx)(
+          "div",
           {
-            onClick: () => {
-              void bus.exit(sessionId);
-            },
             style: {
-              border: "none",
-              background: "transparent",
+              fontSize: 11,
               color: "#8b949e",
-              cursor: "pointer",
-              fontSize: 12,
-              flexShrink: 0
+              fontVariantNumeric: "tabular-nums",
+              fontFamily: "ui-monospace, SFMono-Regular, Menlo, Consolas, monospace",
+              whiteSpace: "nowrap",
+              overflowX: "auto"
             },
-            children: t("exit")
+            children: telParts.join(" \xB7 ")
           }
         )
       ]
@@ -1857,7 +3746,8 @@ function VoiceOverlay({ bus }) {
       style: {
         position: "fixed",
         right: 16,
-        bottom: 16,
+        bottom: 96,
+        // 上移，避免盖住底部输入框/麦克风按钮
         zIndex: 9999,
         display: "flex",
         alignItems: "center",
@@ -1866,7 +3756,8 @@ function VoiceOverlay({ bus }) {
         borderRadius: 999,
         fontSize: 12,
         fontFamily: "system-ui, sans-serif",
-        pointerEvents: "auto",
+        pointerEvents: "none",
+        // 浮层不挡输入框/麦克风按钮的点击（仅内部「跳过」按钮可点）
         background: "rgba(22, 24, 28, 0.85)",
         backdropFilter: "blur(14px)",
         WebkitBackdropFilter: "blur(14px)",
@@ -1904,7 +3795,9 @@ function VoiceOverlay({ bus }) {
               padding: "3px 12px",
               fontSize: 11,
               cursor: "pointer",
-              flexShrink: 0
+              flexShrink: 0,
+              pointerEvents: "auto"
+              // 仅此按钮可点
             },
             children: t("skip")
           }