ruiyun-human 1.0.9 → 1.0.11

package/README.md

@@ -0,0 +1,7 @@
+# shuzirenzujian
+
+数字人前端自助机组件
+npm 发包
+npm publish
+
+npm所有人 睿政云-陆洋

package/dist/ruiyun-human.cjs.js

@@ -385,8 +385,8 @@ var recorder = { exports: {} };
 var recorderExports = recorder.exports;
 var jsAudioRecorder = recorderExports;
 const Recorder = /* @__PURE__ */ getDefaultExportFromCjs(jsAudioRecorder);
-const baseTTSUrl = "http://127.0.0.1:8002";
-const ASR_URL = "http://127.0.0.1:9988/transcribe";
+const baseTTSUrl = "http://192.168.3.182:8002";
+const ASR_URL = "http://192.168.3.182:9988";
 const _sfc_main = {
   __name: "index",
   props: {
@@ -649,22 +649,150 @@ const _sfc_main = {
         currentAudio.value = null;
       }
     }
-    function waitWithSignal(ms, signal) {
-      return new Promise((resolve, reject) => {
-        if (signal == null ? void 0 : signal.aborted) {
-          reject(new DOMException("Aborted", "AbortError"));
-          return;
-        }
-        const timer = setTimeout(resolve, ms);
-        const onAbort = () => {
-          clearTimeout(timer);
-          reject(new DOMException("Aborted", "AbortError"));
+    const voiceFileCache = /* @__PURE__ */ new Map();
+    const FILEDATA_META = { _type: "gradio.FileData" };
+    const uploadReferenceAudio = async (blob, filename, signal) => {
+      const form = new FormData();
+      form.append("files", blob, filename);
+      const res = await fetch(`${baseTTSUrl}/gradio_api/upload`, {
+        method: "POST",
+        body: form,
+        signal
+      });
+      if (!res.ok) throw new Error(`参考音频上传失败 HTTP ${res.status}`);
+      const paths = await res.json();
+      const path = Array.isArray(paths) ? paths[0] : paths;
+      if (!path) throw new Error("参考音频上传失败：服务未返回文件路径");
+      return { path, orig_name: filename, meta: FILEDATA_META };
+    };
+    const resolveVoiceFileData = async (voiceSource, signal) => {
+      if (voiceSource && typeof voiceSource === "object" && !Array.isArray(voiceSource) && voiceSource.path) {
+        return {
+          path: voiceSource.path,
+          orig_name: voiceSource.orig_name || "reference.mp3",
+          meta: FILEDATA_META
         };
-        signal == null ? void 0 : signal.addEventListener("abort", onAbort, { once: true });
+      }
+      if (voiceSource instanceof Blob) {
+        const name = voiceSource.name || `reference_${Date.now()}.mp3`;
+        return await uploadReferenceAudio(voiceSource, name, signal);
+      }
+      if (typeof voiceSource === "string" && voiceSource) {
+        if (voiceSource.startsWith(baseTTSUrl)) {
+          const match = voiceSource.match(/\/gradio_api\/file=(.+)$/);
+          if (match) {
+            const path = decodeURIComponent(match[1]);
+            const origName = path.split(/[\\/]/).pop() || "reference.mp3";
+            return { path, orig_name: origName, meta: FILEDATA_META };
+          }
+        }
+        const res = await fetch(voiceSource, { signal });
+        if (!res.ok) throw new Error(`参考音频下载失败 HTTP ${res.status}`);
+        const blob = await res.blob();
+        const fileName = decodeURIComponent(voiceSource.split("?")[0].split("/").pop()) || "reference.mp3";
+        return await uploadReferenceAudio(blob, fileName, signal);
+      }
+      throw new Error("不支持的参考音频来源：voiceUrl 应为音频 URL 字符串、File/Blob 或 { path } 对象");
+    };
+    const submitCloneJob = async (data, signal) => {
+      const res = await fetch(`${baseTTSUrl}/gradio_api/call/_clone_fn`, {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ data }),
+        signal
       });
-    }
+      if (!res.ok) {
+        const text = await res.text().catch(() => "");
+        throw new Error(`克隆任务提交失败 HTTP ${res.status} ${text}`);
+      }
+      const json = await res.json();
+      if (!(json == null ? void 0 : json.event_id)) throw new Error("克隆任务提交失败：未返回 event_id");
+      return json.event_id;
+    };
+    const waitCloneResult = async (eventId, signal) => {
+      const res = await fetch(`${baseTTSUrl}/gradio_api/call/_clone_fn/${eventId}`, {
+        signal,
+        headers: { Accept: "text/event-stream" }
+      });
+      if (!res.ok) throw new Error(`合成结果监听失败 HTTP ${res.status}`);
+      const reader = res.body.getReader();
+      const decoder = new TextDecoder("utf-8");
+      let buffer = "";
+      let eventName = "message";
+      while (true) {
+        const { done, value } = await reader.read();
+        if (done) break;
+        buffer += decoder.decode(value, { stream: true }).replace(/\r\n/g, "\n");
+        let sep;
+        while ((sep = buffer.indexOf("\n\n")) !== -1) {
+          const frame = buffer.slice(0, sep);
+          buffer = buffer.slice(sep + 2);
+          const dataLines = [];
+          for (const line of frame.split("\n")) {
+            if (line.startsWith("event:")) eventName = line.slice(6).trim();
+            else if (line.startsWith("data:")) dataLines.push(line.slice(5).replace(/^ /, ""));
+          }
+          if (dataLines.length === 0) continue;
+          const payload = dataLines.join("\n");
+          if (eventName === "complete") {
+            return JSON.parse(payload);
+          }
+          if (eventName === "error") {
+            let msg = payload;
+            try {
+              const parsed = JSON.parse(payload);
+              msg = typeof parsed === "string" ? parsed : (parsed == null ? void 0 : parsed.message) || payload;
+            } catch {
+            }
+            throw new Error(msg || "OmniVoice 合成失败");
+          }
+          eventName = "message";
+        }
+      }
+      throw new Error("OmniVoice 合成结果流提前中断");
+    };
+    const playAudioFromUrl = async (audioUrl, requestId, signal) => {
+      const audioRes = await fetch(audioUrl, { signal });
+      if (!audioRes.ok) throw new Error(`合成音频下载失败 HTTP ${audioRes.status}`);
+      const audioBlob = await audioRes.blob();
+      const objectUrl = URL.createObjectURL(audioBlob);
+      if (signal.aborted || currentTtsRequestId.value !== requestId) {
+        URL.revokeObjectURL(objectUrl);
+        return;
+      }
+      const audio = new Audio(objectUrl);
+      currentAudio.value = audio;
+      const cleanup = () => {
+        URL.revokeObjectURL(objectUrl);
+        if (currentAudio.value === audio) currentAudio.value = null;
+      };
+      audio.onplaying = () => {
+        console.log("[OmniVoice] 音频开始播放");
+        speak();
+      };
+      audio.onended = () => {
+        console.log("[OmniVoice] 音频播放结束");
+        cleanup();
+        stopSpeaking();
+      };
+      audio.onerror = () => {
+        console.error("[OmniVoice] 音频播放失败");
+        cleanup();
+        stopSpeaking();
+      };
+      await audio.play();
+    };
     const playTTS = async (text) => {
-      var _a, _b, _c, _d, _e, _f, _g, _h;
+      var _a, _b, _c, _d;
+      console.log(props.humans);
+      let voiceUrl = "";
+      let voiceUrl2 = "http://192.168.3.182:8002/gradio_api/file=C:\\Users\\Administrator\\AppData\\Local\\Temp\\gradio\\4eef6d9157de756de02bccca2f66389382e2ac4acb0c674981aa9fe13dcc294c\\李梓萌.mp3";
+      let voiceUrl1 = "http://192.168.3.182:8002/gradio_api/file=C:\\Users\\Administrator\\AppData\\Local\\Temp\\gradio\\ff2665f538a4b0e5efbcd79f9f699236f5f0e52cad7e17f0664ddf431169be30\\guozj.mp3";
+      if (props.humans[0].speakSet.sex == "Female / 女") {
+        voiceUrl = voiceUrl2;
+      } else {
+        voiceUrl = voiceUrl1;
+      }
       if (!text || !((_a = currentHuman.value) == null ? void 0 : _a.isMuted)) return;
       abortTTS();
       ttsAbortController = new AbortController();
@@ -674,132 +802,33 @@ const _sfc_main = {
       try {
         const cleanText = text.replace(/```[\s\S]*?```/g, "").replace(/`[^`]+`/g, "").replace(/[#*_~]/g, "").replace(/\n+/g, "，").replace(/\s+/g, " ").replace(/[，。！？]{2,}/g, "，").trim();
         if (!cleanText) return;
-        const baseUrl = baseTTSUrl;
-        const sessionHash = "session_" + Date.now();
-        let speed = ((_c = (_b = currentHuman.value) == null ? void 0 : _b.speakSet) == null ? void 0 : _c.speed) || 1;
-        let age = ((_e = (_d = currentHuman.value) == null ? void 0 : _d.speakSet) == null ? void 0 : _e.age) || "Young Adult / 青年";
-        let sex = ((_g = (_f = currentHuman.value) == null ? void 0 : _f.speakSet) == null ? void 0 : _g.sex) || "Female / 女";
-        const data = [cleanText, "Auto", 32, 2, true, speed, null, true, true, sex, age, "Auto", "Auto", "Auto", "Auto"];
-        console.log("[OmniVoice] 提交请求:", { text: cleanText, length: cleanText.length, fn_index: 1 });
-        const joinRes = await fetch(`${baseUrl}/gradio_api/queue/join`, {
-          method: "POST",
-          headers: { "Content-Type": "application/json" },
-          body: JSON.stringify({ data, session_hash: sessionHash, fn_index: 1, trigger_id: 50 }),
-          signal
-        });
-        const joinData = await joinRes.json();
-        console.log("[OmniVoice] Step1 响应:", joinData);
-        console.log("[OmniVoice] Step2: 轮询等待结果");
-        let audioUrl = "";
-        let maxPolls = 60;
-        let pollInterval = 1e3;
-        while (!audioUrl && maxPolls > 0) {
-          if (signal.aborted || currentTtsRequestId.value !== requestId) return;
-          await waitWithSignal(pollInterval, signal);
-          maxPolls--;
-          try {
-            const pollRes = await fetch(`${baseUrl}/gradio_api/queue/data?session_hash=${sessionHash}`, {
-              signal
-            });
-            const pollText = await pollRes.text();
-            const lines = pollText.split("\n");
-            for (const line of lines) {
-              const trimLine = line.trim();
-              if (!trimLine.startsWith("data: ")) continue;
-              try {
-                const json = JSON.parse(trimLine.replace("data: ", ""));
-                console.log("[OmniVoice] 轮询消息:", json);
-                if (json.msg === "process_completed" && ((_h = json.output) == null ? void 0 : _h.data)) {
-                  const fileData = json.output.data[0];
-                  console.log("[OmniVoice] 获取到音频:", fileData);
-                  if (fileData == null ? void 0 : fileData.url) audioUrl = fileData.url;
-                  else if (fileData == null ? void 0 : fileData.path) audioUrl = `${baseUrl}/gradio_api/file=${fileData.path}`;
-                  else if (typeof fileData === "string") audioUrl = fileData.startsWith("http") ? fileData : `${baseUrl}/gradio_api/file=${fileData}`;
-                  else if (Array.isArray(fileData) && fileData.length >= 2) {
-                    const samplingRate = fileData[0];
-                    const waveform = fileData[1];
-                    console.log("[OmniVoice] 直接获取到音频数组:", { samplingRate, waveformLength: waveform == null ? void 0 : waveform.length });
-                    if (samplingRate && waveform && Array.isArray(waveform)) {
-                      const audioBuffer = new ArrayBuffer(waveform.length * 2);
-                      const view = new DataView(audioBuffer);
-                      for (let i = 0; i < waveform.length; i++) {
-                        view.setInt16(i * 2, waveform[i], true);
-                      }
-                      const wavBlob = createWavBlob(audioBuffer, samplingRate, 1, 16);
-                      const audioObjectUrl = URL.createObjectURL(wavBlob);
-                      if (signal.aborted || currentTtsRequestId.value !== requestId) {
-                        URL.revokeObjectURL(audioObjectUrl);
-                        return;
-                      }
-                      currentAudio.value = new Audio(audioObjectUrl);
-                      currentAudio.value.onplaying = () => {
-                        console.log("[OmniVoice] 音频开始播放");
-                        speak();
-                      };
-                      currentAudio.value.onended = () => {
-                        console.log("[OmniVoice] 音频播放结束");
-                        stopSpeaking();
-                        currentAudio.value = null;
-                        URL.revokeObjectURL(audioObjectUrl);
-                      };
-                      currentAudio.value.onerror = () => {
-                        console.error("[OmniVoice] 音频播放失败");
-                        stopSpeaking();
-                        currentAudio.value = null;
-                        URL.revokeObjectURL(audioObjectUrl);
-                      };
-                      await currentAudio.value.play();
-                      return;
-                    }
-                  }
-                  break;
-                } else if (json.msg === "process_error") {
-                  console.error("[OmniVoice] 处理错误:", json);
-                  standby();
-                  return;
-                }
-              } catch (e) {
-                console.error("[OmniVoice] 解析轮询消息失败:", e);
-              }
-            }
-          } catch (e) {
-            if ((e == null ? void 0 : e.name) === "AbortError") throw e;
-            console.warn("[OmniVoice] 轮询失败:", e.message);
-          }
-        }
-        if (audioUrl) {
-          if (audioUrl.startsWith("/")) audioUrl = baseUrl + audioUrl;
-          console.log("[OmniVoice] 下载音频:", audioUrl);
-          if (signal.aborted || currentTtsRequestId.value !== requestId) return;
-          const audioRes = await fetch(audioUrl, { signal });
-          const audioBlob = await audioRes.blob();
-          const audioObjectUrl = URL.createObjectURL(audioBlob);
-          if (signal.aborted || currentTtsRequestId.value !== requestId) {
-            URL.revokeObjectURL(audioObjectUrl);
-            return;
-          }
-          currentAudio.value = new Audio(audioObjectUrl);
-          currentAudio.value.onplaying = () => {
-            console.log("[OmniVoice] 音频开始播放");
-            speak();
-          };
-          currentAudio.value.onended = () => {
-            console.log("[OmniVoice] 音频播放结束");
-            stopSpeaking();
-            currentAudio.value = null;
-            URL.revokeObjectURL(audioObjectUrl);
-          };
-          currentAudio.value.onerror = () => {
-            console.error("[OmniVoice] 音频播放失败");
-            stopSpeaking();
-            currentAudio.value = null;
-            URL.revokeObjectURL(audioObjectUrl);
-          };
-          await currentAudio.value.play();
-        } else {
-          console.error("[OmniVoice] 未获取到音频");
+        const speed = Number((_c = (_b = currentHuman.value) == null ? void 0 : _b.speakSet) == null ? void 0 : _c.speed) || 1;
+        const voiceSource = voiceUrl || ((_d = currentHuman.value) == null ? void 0 : _d.voiceUrl);
+        if (!voiceSource) {
+          console.error("[OmniVoice] 未配置参考音频 voiceUrl，无法进行音色克隆");
           standby();
+          return;
+        }
+        let refFile = voiceFileCache.get(voiceSource);
+        if (!refFile) {
+          console.log("[OmniVoice] 准备参考音频:", typeof voiceSource === "string" ? voiceSource : "Blob/File");
+          refFile = await resolveVoiceFileData(voiceSource, signal);
+          voiceFileCache.set(voiceSource, refFile);
         }
+        console.log("[OmniVoice] 参考音频就绪:", refFile);
+        const data = [cleanText, "Auto", refFile, "", null, 32, 2, true, speed, null, true, true];
+        console.log("[OmniVoice] 提交克隆请求:", { text: cleanText, length: cleanText.length });
+        const eventId = await submitCloneJob(data, signal);
+        const outputs = await waitCloneResult(eventId, signal);
+        const audioFile = outputs == null ? void 0 : outputs[0];
+        const statusText = outputs == null ? void 0 : outputs[1];
+        console.log("[OmniVoice] 合成结果:", { audioFile, statusText });
+        if (signal.aborted || currentTtsRequestId.value !== requestId) return;
+        if (!audioFile || !audioFile.path && !audioFile.url) {
+          throw new Error(statusText || "OmniVoice 未返回合成音频");
+        }
+        const audioUrl = audioFile.url && audioFile.url.startsWith("http") ? audioFile.url : `${baseTTSUrl}/gradio_api/file=${audioFile.path}`;
+        await playAudioFromUrl(audioUrl, requestId, signal);
       } catch (e) {
         if ((e == null ? void 0 : e.name) === "AbortError") {
           console.log("[OmniVoice] TTS已取消");
@@ -836,7 +865,7 @@ const _sfc_main = {
       const formData = new FormData();
       formData.append("file", blob, "recorded_audio.wav");
       try {
-        const res = await fetch(ASR_URL, {
+        const res = await fetch(ASR_URL + "/transcribe", {
           method: "POST",
           body: formData
         });
@@ -849,42 +878,6 @@ const _sfc_main = {
         return { status: e, text: "" };
       }
     };
-    const createWavBlob = (audioBuffer, sampleRate, channels, bitsPerSample) => {
-      const bytesPerSample = bitsPerSample / 8;
-      const blockAlign = channels * bytesPerSample;
-      const dataSize = audioBuffer.byteLength;
-      const buffer = new ArrayBuffer(44 + dataSize);
-      const view = new DataView(buffer);
-      view.setUint8(0, 82);
-      view.setUint8(1, 73);
-      view.setUint8(2, 70);
-      view.setUint8(3, 70);
-      view.setUint32(4, 36 + dataSize, true);
-      view.setUint8(8, 87);
-      view.setUint8(9, 65);
-      view.setUint8(10, 86);
-      view.setUint8(11, 69);
-      view.setUint8(12, 102);
-      view.setUint8(13, 109);
-      view.setUint8(14, 116);
-      view.setUint8(15, 32);
-      view.setUint32(16, 16, true);
-      view.setUint16(20, 1, true);
-      view.setUint16(22, channels, true);
-      view.setUint32(24, sampleRate, true);
-      view.setUint32(28, sampleRate * blockAlign, true);
-      view.setUint16(32, blockAlign, true);
-      view.setUint16(34, bitsPerSample, true);
-      view.setUint8(36, 100);
-      view.setUint8(37, 97);
-      view.setUint8(38, 116);
-      view.setUint8(39, 97);
-      view.setUint32(40, dataSize, true);
-      const audioData = new Uint8Array(audioBuffer);
-      const destData = new Uint8Array(buffer, 44);
-      destData.set(audioData);
-      return new Blob([buffer], { type: "audio/wav" });
-    };
     let recorder2 = vue.ref(null);
     const initRecorder = () => {
       console.log("initRecorder");
@@ -936,7 +929,7 @@ const _sfc_main = {
     };
   }
 };
-const DigitalHuman = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-711ac421"]]);
+const DigitalHuman = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-9c25016c"]]);
 DigitalHuman.install = (app) => app.component("DigitalHuman", DigitalHuman);
 const components = [DemoButton, DemoInput, DigitalHuman];
 const install = (app) => {

package/dist/ruiyun-human.es.js

@@ -383,8 +383,8 @@ var recorder = { exports: {} };
 var recorderExports = recorder.exports;
 var jsAudioRecorder = recorderExports;
 const Recorder = /* @__PURE__ */ getDefaultExportFromCjs(jsAudioRecorder);
-const baseTTSUrl = "http://127.0.0.1:8002";
-const ASR_URL = "http://127.0.0.1:9988/transcribe";
+const baseTTSUrl = "http://192.168.3.182:8002";
+const ASR_URL = "http://192.168.3.182:9988";
 const _sfc_main = {
   __name: "index",
   props: {
@@ -647,22 +647,150 @@ const _sfc_main = {
         currentAudio.value = null;
       }
     }
-    function waitWithSignal(ms, signal) {
-      return new Promise((resolve, reject) => {
-        if (signal == null ? void 0 : signal.aborted) {
-          reject(new DOMException("Aborted", "AbortError"));
-          return;
-        }
-        const timer = setTimeout(resolve, ms);
-        const onAbort = () => {
-          clearTimeout(timer);
-          reject(new DOMException("Aborted", "AbortError"));
+    const voiceFileCache = /* @__PURE__ */ new Map();
+    const FILEDATA_META = { _type: "gradio.FileData" };
+    const uploadReferenceAudio = async (blob, filename, signal) => {
+      const form = new FormData();
+      form.append("files", blob, filename);
+      const res = await fetch(`${baseTTSUrl}/gradio_api/upload`, {
+        method: "POST",
+        body: form,
+        signal
+      });
+      if (!res.ok) throw new Error(`参考音频上传失败 HTTP ${res.status}`);
+      const paths = await res.json();
+      const path = Array.isArray(paths) ? paths[0] : paths;
+      if (!path) throw new Error("参考音频上传失败：服务未返回文件路径");
+      return { path, orig_name: filename, meta: FILEDATA_META };
+    };
+    const resolveVoiceFileData = async (voiceSource, signal) => {
+      if (voiceSource && typeof voiceSource === "object" && !Array.isArray(voiceSource) && voiceSource.path) {
+        return {
+          path: voiceSource.path,
+          orig_name: voiceSource.orig_name || "reference.mp3",
+          meta: FILEDATA_META
         };
-        signal == null ? void 0 : signal.addEventListener("abort", onAbort, { once: true });
+      }
+      if (voiceSource instanceof Blob) {
+        const name = voiceSource.name || `reference_${Date.now()}.mp3`;
+        return await uploadReferenceAudio(voiceSource, name, signal);
+      }
+      if (typeof voiceSource === "string" && voiceSource) {
+        if (voiceSource.startsWith(baseTTSUrl)) {
+          const match = voiceSource.match(/\/gradio_api\/file=(.+)$/);
+          if (match) {
+            const path = decodeURIComponent(match[1]);
+            const origName = path.split(/[\\/]/).pop() || "reference.mp3";
+            return { path, orig_name: origName, meta: FILEDATA_META };
+          }
+        }
+        const res = await fetch(voiceSource, { signal });
+        if (!res.ok) throw new Error(`参考音频下载失败 HTTP ${res.status}`);
+        const blob = await res.blob();
+        const fileName = decodeURIComponent(voiceSource.split("?")[0].split("/").pop()) || "reference.mp3";
+        return await uploadReferenceAudio(blob, fileName, signal);
+      }
+      throw new Error("不支持的参考音频来源：voiceUrl 应为音频 URL 字符串、File/Blob 或 { path } 对象");
+    };
+    const submitCloneJob = async (data, signal) => {
+      const res = await fetch(`${baseTTSUrl}/gradio_api/call/_clone_fn`, {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ data }),
+        signal
       });
-    }
+      if (!res.ok) {
+        const text = await res.text().catch(() => "");
+        throw new Error(`克隆任务提交失败 HTTP ${res.status} ${text}`);
+      }
+      const json = await res.json();
+      if (!(json == null ? void 0 : json.event_id)) throw new Error("克隆任务提交失败：未返回 event_id");
+      return json.event_id;
+    };
+    const waitCloneResult = async (eventId, signal) => {
+      const res = await fetch(`${baseTTSUrl}/gradio_api/call/_clone_fn/${eventId}`, {
+        signal,
+        headers: { Accept: "text/event-stream" }
+      });
+      if (!res.ok) throw new Error(`合成结果监听失败 HTTP ${res.status}`);
+      const reader = res.body.getReader();
+      const decoder = new TextDecoder("utf-8");
+      let buffer = "";
+      let eventName = "message";
+      while (true) {
+        const { done, value } = await reader.read();
+        if (done) break;
+        buffer += decoder.decode(value, { stream: true }).replace(/\r\n/g, "\n");
+        let sep;
+        while ((sep = buffer.indexOf("\n\n")) !== -1) {
+          const frame = buffer.slice(0, sep);
+          buffer = buffer.slice(sep + 2);
+          const dataLines = [];
+          for (const line of frame.split("\n")) {
+            if (line.startsWith("event:")) eventName = line.slice(6).trim();
+            else if (line.startsWith("data:")) dataLines.push(line.slice(5).replace(/^ /, ""));
+          }
+          if (dataLines.length === 0) continue;
+          const payload = dataLines.join("\n");
+          if (eventName === "complete") {
+            return JSON.parse(payload);
+          }
+          if (eventName === "error") {
+            let msg = payload;
+            try {
+              const parsed = JSON.parse(payload);
+              msg = typeof parsed === "string" ? parsed : (parsed == null ? void 0 : parsed.message) || payload;
+            } catch {
+            }
+            throw new Error(msg || "OmniVoice 合成失败");
+          }
+          eventName = "message";
+        }
+      }
+      throw new Error("OmniVoice 合成结果流提前中断");
+    };
+    const playAudioFromUrl = async (audioUrl, requestId, signal) => {
+      const audioRes = await fetch(audioUrl, { signal });
+      if (!audioRes.ok) throw new Error(`合成音频下载失败 HTTP ${audioRes.status}`);
+      const audioBlob = await audioRes.blob();
+      const objectUrl = URL.createObjectURL(audioBlob);
+      if (signal.aborted || currentTtsRequestId.value !== requestId) {
+        URL.revokeObjectURL(objectUrl);
+        return;
+      }
+      const audio = new Audio(objectUrl);
+      currentAudio.value = audio;
+      const cleanup = () => {
+        URL.revokeObjectURL(objectUrl);
+        if (currentAudio.value === audio) currentAudio.value = null;
+      };
+      audio.onplaying = () => {
+        console.log("[OmniVoice] 音频开始播放");
+        speak();
+      };
+      audio.onended = () => {
+        console.log("[OmniVoice] 音频播放结束");
+        cleanup();
+        stopSpeaking();
+      };
+      audio.onerror = () => {
+        console.error("[OmniVoice] 音频播放失败");
+        cleanup();
+        stopSpeaking();
+      };
+      await audio.play();
+    };
     const playTTS = async (text) => {
-      var _a, _b, _c, _d, _e, _f, _g, _h;
+      var _a, _b, _c, _d;
+      console.log(props.humans);
+      let voiceUrl = "";
+      let voiceUrl2 = "http://192.168.3.182:8002/gradio_api/file=C:\\Users\\Administrator\\AppData\\Local\\Temp\\gradio\\4eef6d9157de756de02bccca2f66389382e2ac4acb0c674981aa9fe13dcc294c\\李梓萌.mp3";
+      let voiceUrl1 = "http://192.168.3.182:8002/gradio_api/file=C:\\Users\\Administrator\\AppData\\Local\\Temp\\gradio\\ff2665f538a4b0e5efbcd79f9f699236f5f0e52cad7e17f0664ddf431169be30\\guozj.mp3";
+      if (props.humans[0].speakSet.sex == "Female / 女") {
+        voiceUrl = voiceUrl2;
+      } else {
+        voiceUrl = voiceUrl1;
+      }
       if (!text || !((_a = currentHuman.value) == null ? void 0 : _a.isMuted)) return;
       abortTTS();
       ttsAbortController = new AbortController();
@@ -672,132 +800,33 @@ const _sfc_main = {
       try {
         const cleanText = text.replace(/```[\s\S]*?```/g, "").replace(/`[^`]+`/g, "").replace(/[#*_~]/g, "").replace(/\n+/g, "，").replace(/\s+/g, " ").replace(/[，。！？]{2,}/g, "，").trim();
         if (!cleanText) return;
-        const baseUrl = baseTTSUrl;
-        const sessionHash = "session_" + Date.now();
-        let speed = ((_c = (_b = currentHuman.value) == null ? void 0 : _b.speakSet) == null ? void 0 : _c.speed) || 1;
-        let age = ((_e = (_d = currentHuman.value) == null ? void 0 : _d.speakSet) == null ? void 0 : _e.age) || "Young Adult / 青年";
-        let sex = ((_g = (_f = currentHuman.value) == null ? void 0 : _f.speakSet) == null ? void 0 : _g.sex) || "Female / 女";
-        const data = [cleanText, "Auto", 32, 2, true, speed, null, true, true, sex, age, "Auto", "Auto", "Auto", "Auto"];
-        console.log("[OmniVoice] 提交请求:", { text: cleanText, length: cleanText.length, fn_index: 1 });
-        const joinRes = await fetch(`${baseUrl}/gradio_api/queue/join`, {
-          method: "POST",
-          headers: { "Content-Type": "application/json" },
-          body: JSON.stringify({ data, session_hash: sessionHash, fn_index: 1, trigger_id: 50 }),
-          signal
-        });
-        const joinData = await joinRes.json();
-        console.log("[OmniVoice] Step1 响应:", joinData);
-        console.log("[OmniVoice] Step2: 轮询等待结果");
-        let audioUrl = "";
-        let maxPolls = 60;
-        let pollInterval = 1e3;
-        while (!audioUrl && maxPolls > 0) {
-          if (signal.aborted || currentTtsRequestId.value !== requestId) return;
-          await waitWithSignal(pollInterval, signal);
-          maxPolls--;
-          try {
-            const pollRes = await fetch(`${baseUrl}/gradio_api/queue/data?session_hash=${sessionHash}`, {
-              signal
-            });
-            const pollText = await pollRes.text();
-            const lines = pollText.split("\n");
-            for (const line of lines) {
-              const trimLine = line.trim();
-              if (!trimLine.startsWith("data: ")) continue;
-              try {
-                const json = JSON.parse(trimLine.replace("data: ", ""));
-                console.log("[OmniVoice] 轮询消息:", json);
-                if (json.msg === "process_completed" && ((_h = json.output) == null ? void 0 : _h.data)) {
-                  const fileData = json.output.data[0];
-                  console.log("[OmniVoice] 获取到音频:", fileData);
-                  if (fileData == null ? void 0 : fileData.url) audioUrl = fileData.url;
-                  else if (fileData == null ? void 0 : fileData.path) audioUrl = `${baseUrl}/gradio_api/file=${fileData.path}`;
-                  else if (typeof fileData === "string") audioUrl = fileData.startsWith("http") ? fileData : `${baseUrl}/gradio_api/file=${fileData}`;
-                  else if (Array.isArray(fileData) && fileData.length >= 2) {
-                    const samplingRate = fileData[0];
-                    const waveform = fileData[1];
-                    console.log("[OmniVoice] 直接获取到音频数组:", { samplingRate, waveformLength: waveform == null ? void 0 : waveform.length });
-                    if (samplingRate && waveform && Array.isArray(waveform)) {
-                      const audioBuffer = new ArrayBuffer(waveform.length * 2);
-                      const view = new DataView(audioBuffer);
-                      for (let i = 0; i < waveform.length; i++) {
-                        view.setInt16(i * 2, waveform[i], true);
-                      }
-                      const wavBlob = createWavBlob(audioBuffer, samplingRate, 1, 16);
-                      const audioObjectUrl = URL.createObjectURL(wavBlob);
-                      if (signal.aborted || currentTtsRequestId.value !== requestId) {
-                        URL.revokeObjectURL(audioObjectUrl);
-                        return;
-                      }
-                      currentAudio.value = new Audio(audioObjectUrl);
-                      currentAudio.value.onplaying = () => {
-                        console.log("[OmniVoice] 音频开始播放");
-                        speak();
-                      };
-                      currentAudio.value.onended = () => {
-                        console.log("[OmniVoice] 音频播放结束");
-                        stopSpeaking();
-                        currentAudio.value = null;
-                        URL.revokeObjectURL(audioObjectUrl);
-                      };
-                      currentAudio.value.onerror = () => {
-                        console.error("[OmniVoice] 音频播放失败");
-                        stopSpeaking();
-                        currentAudio.value = null;
-                        URL.revokeObjectURL(audioObjectUrl);
-                      };
-                      await currentAudio.value.play();
-                      return;
-                    }
-                  }
-                  break;
-                } else if (json.msg === "process_error") {
-                  console.error("[OmniVoice] 处理错误:", json);
-                  standby();
-                  return;
-                }
-              } catch (e) {
-                console.error("[OmniVoice] 解析轮询消息失败:", e);
-              }
-            }
-          } catch (e) {
-            if ((e == null ? void 0 : e.name) === "AbortError") throw e;
-            console.warn("[OmniVoice] 轮询失败:", e.message);
-          }
-        }
-        if (audioUrl) {
-          if (audioUrl.startsWith("/")) audioUrl = baseUrl + audioUrl;
-          console.log("[OmniVoice] 下载音频:", audioUrl);
-          if (signal.aborted || currentTtsRequestId.value !== requestId) return;
-          const audioRes = await fetch(audioUrl, { signal });
-          const audioBlob = await audioRes.blob();
-          const audioObjectUrl = URL.createObjectURL(audioBlob);
-          if (signal.aborted || currentTtsRequestId.value !== requestId) {
-            URL.revokeObjectURL(audioObjectUrl);
-            return;
-          }
-          currentAudio.value = new Audio(audioObjectUrl);
-          currentAudio.value.onplaying = () => {
-            console.log("[OmniVoice] 音频开始播放");
-            speak();
-          };
-          currentAudio.value.onended = () => {
-            console.log("[OmniVoice] 音频播放结束");
-            stopSpeaking();
-            currentAudio.value = null;
-            URL.revokeObjectURL(audioObjectUrl);
-          };
-          currentAudio.value.onerror = () => {
-            console.error("[OmniVoice] 音频播放失败");
-            stopSpeaking();
-            currentAudio.value = null;
-            URL.revokeObjectURL(audioObjectUrl);
-          };
-          await currentAudio.value.play();
-        } else {
-          console.error("[OmniVoice] 未获取到音频");
+        const speed = Number((_c = (_b = currentHuman.value) == null ? void 0 : _b.speakSet) == null ? void 0 : _c.speed) || 1;
+        const voiceSource = voiceUrl || ((_d = currentHuman.value) == null ? void 0 : _d.voiceUrl);
+        if (!voiceSource) {
+          console.error("[OmniVoice] 未配置参考音频 voiceUrl，无法进行音色克隆");
           standby();
+          return;
+        }
+        let refFile = voiceFileCache.get(voiceSource);
+        if (!refFile) {
+          console.log("[OmniVoice] 准备参考音频:", typeof voiceSource === "string" ? voiceSource : "Blob/File");
+          refFile = await resolveVoiceFileData(voiceSource, signal);
+          voiceFileCache.set(voiceSource, refFile);
         }
+        console.log("[OmniVoice] 参考音频就绪:", refFile);
+        const data = [cleanText, "Auto", refFile, "", null, 32, 2, true, speed, null, true, true];
+        console.log("[OmniVoice] 提交克隆请求:", { text: cleanText, length: cleanText.length });
+        const eventId = await submitCloneJob(data, signal);
+        const outputs = await waitCloneResult(eventId, signal);
+        const audioFile = outputs == null ? void 0 : outputs[0];
+        const statusText = outputs == null ? void 0 : outputs[1];
+        console.log("[OmniVoice] 合成结果:", { audioFile, statusText });
+        if (signal.aborted || currentTtsRequestId.value !== requestId) return;
+        if (!audioFile || !audioFile.path && !audioFile.url) {
+          throw new Error(statusText || "OmniVoice 未返回合成音频");
+        }
+        const audioUrl = audioFile.url && audioFile.url.startsWith("http") ? audioFile.url : `${baseTTSUrl}/gradio_api/file=${audioFile.path}`;
+        await playAudioFromUrl(audioUrl, requestId, signal);
       } catch (e) {
         if ((e == null ? void 0 : e.name) === "AbortError") {
           console.log("[OmniVoice] TTS已取消");
@@ -834,7 +863,7 @@ const _sfc_main = {
       const formData = new FormData();
       formData.append("file", blob, "recorded_audio.wav");
       try {
-        const res = await fetch(ASR_URL, {
+        const res = await fetch(ASR_URL + "/transcribe", {
           method: "POST",
           body: formData
         });
@@ -847,42 +876,6 @@ const _sfc_main = {
         return { status: e, text: "" };
       }
     };
-    const createWavBlob = (audioBuffer, sampleRate, channels, bitsPerSample) => {
-      const bytesPerSample = bitsPerSample / 8;
-      const blockAlign = channels * bytesPerSample;
-      const dataSize = audioBuffer.byteLength;
-      const buffer = new ArrayBuffer(44 + dataSize);
-      const view = new DataView(buffer);
-      view.setUint8(0, 82);
-      view.setUint8(1, 73);
-      view.setUint8(2, 70);
-      view.setUint8(3, 70);
-      view.setUint32(4, 36 + dataSize, true);
-      view.setUint8(8, 87);
-      view.setUint8(9, 65);
-      view.setUint8(10, 86);
-      view.setUint8(11, 69);
-      view.setUint8(12, 102);
-      view.setUint8(13, 109);
-      view.setUint8(14, 116);
-      view.setUint8(15, 32);
-      view.setUint32(16, 16, true);
-      view.setUint16(20, 1, true);
-      view.setUint16(22, channels, true);
-      view.setUint32(24, sampleRate, true);
-      view.setUint32(28, sampleRate * blockAlign, true);
-      view.setUint16(32, blockAlign, true);
-      view.setUint16(34, bitsPerSample, true);
-      view.setUint8(36, 100);
-      view.setUint8(37, 97);
-      view.setUint8(38, 116);
-      view.setUint8(39, 97);
-      view.setUint32(40, dataSize, true);
-      const audioData = new Uint8Array(audioBuffer);
-      const destData = new Uint8Array(buffer, 44);
-      destData.set(audioData);
-      return new Blob([buffer], { type: "audio/wav" });
-    };
     let recorder2 = ref(null);
     const initRecorder = () => {
       console.log("initRecorder");
@@ -934,7 +927,7 @@ const _sfc_main = {
     };
   }
 };
-const DigitalHuman = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-711ac421"]]);
+const DigitalHuman = /* @__PURE__ */ _export_sfc(_sfc_main, [["__scopeId", "data-v-9c25016c"]]);
 DigitalHuman.install = (app) => app.component("DigitalHuman", DigitalHuman);
 const components = [DemoButton, DemoInput, DigitalHuman];
 const install = (app) => {

package/dist/style.css

@@ -1 +1 @@
-.demo-btn[data-v-9a48516a]{padding:6px 14px;border-radius:4px;border:none;cursor:pointer;font-size:14px}.primary[data-v-9a48516a]{background:#409eff;color:#fff}.success[data-v-9a48516a]{background:#67c23a;color:#fff}.warning[data-v-9a48516a]{background:#e6a23c;color:#fff}.demo-input-wrap[data-v-28a3493c]{display:inline-flex;align-items:center;border:1px solid #dcdcdc;border-radius:4px;padding:0 10px}.demo-input-wrap[data-v-28a3493c]:focus-within{border-color:#409eff}.demo-input[data-v-28a3493c]{border:none;outline:none;padding:8px 6px;flex:1}.prefix[data-v-28a3493c],.suffix[data-v-28a3493c]{color:#909399}.digital-human[data-v-711ac421]{width:100%;height:100%;display:flex;align-items:center;justify-content:center}.digital-human canvas[data-v-711ac421]{max-width:100%;max-height:100%}
+.demo-btn[data-v-9a48516a]{padding:6px 14px;border-radius:4px;border:none;cursor:pointer;font-size:14px}.primary[data-v-9a48516a]{background:#409eff;color:#fff}.success[data-v-9a48516a]{background:#67c23a;color:#fff}.warning[data-v-9a48516a]{background:#e6a23c;color:#fff}.demo-input-wrap[data-v-28a3493c]{display:inline-flex;align-items:center;border:1px solid #dcdcdc;border-radius:4px;padding:0 10px}.demo-input-wrap[data-v-28a3493c]:focus-within{border-color:#409eff}.demo-input[data-v-28a3493c]{border:none;outline:none;padding:8px 6px;flex:1}.prefix[data-v-28a3493c],.suffix[data-v-28a3493c]{color:#909399}.digital-human[data-v-9c25016c]{width:100%;height:100%;display:flex;align-items:center;justify-content:center}.digital-human canvas[data-v-9c25016c]{max-width:100%;max-height:100%}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ruiyun-human",
-  "version": "1.0.9",
+  "version": "1.0.11",
   "type": "module",
   "description": "数字人初步实现",
   "main": "./dist/ruiyun-human.cjs.js",
@@ -29,11 +29,11 @@
     "postcss": "^8.5.20",
     "tailwindcss": "^4.3.3",
     "terser": "^5.49.0",
-    "vite": "^5.1.4",
-    "vue": "^3.4.21"
+    "vite": "^5.1.4"
   },
   "dependencies": {
     "js-audio-recorder": "^1.0.7",
-    "vue-router": "^4.6.4"
+    "vue-router": "^4.6.4",
+    "vue": "^3.4.21"
   }
 }

package/dist/favicon.svg

@@ -1,4 +0,0 @@
-<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 100 100">
-  <rect width="100" height="100" rx="20" fill="#42b883"/>
-  <text y=".9em" font-size="90" fill="white" text-anchor="middle" x="50">V</text>
-</svg>