yuanflow-cli 0.1.47 → 0.1.49

package/src/cli.js

@@ -629,19 +629,19 @@ function printHelp() {
   yuanflow-cli browser task-plan --platform xiaohongshu --task publish --account main --format agent-json
   yuanflow-cli ai qwen3-vl-plus --prompt "描述这张图" --image-url "https://example.com/image.png" --dry-run
   yuanflow-cli ai qwen3-vl-plus --prompt "总结这个视频画面" --video-url "https://example.com/video.mp4" --dry-run
-  yuanflow-cli ai qwen3-vl-plus --prompt "描述本地图片" --image-file "D:\\素材\\cover.png" --dry-run
-  yuanflow-cli ai qwen3-vl-plus --prompt "描述本地视频" --video-file "D:\\素材\\demo.mp4" --dry-run
-  yuanflow-cli voice clone --file-transfer "D:\\voice\\sample.wav" --name demo --activate --dry-run
+  yuanflow-cli ai qwen3-vl-plus --prompt "描述本地图片" --image-file "<本地图片路径>" --dry-run
+  yuanflow-cli ai qwen3-vl-plus --prompt "描述本地视频" --video-file "<本地视频路径>" --dry-run
+  yuanflow-cli voice clone --file-transfer "<本地音频路径>" --name demo --activate --dry-run
   yuanflow-cli voice list --dry-run
   yuanflow-cli voice activate --voice voice_xxx --dry-run
-  yuanflow-cli voice replicate --text "你好" --voice voice_xxx --output "D:\\voice\\replicate.mp3" --dry-run
-  yuanflow-cli ai qwen-voice-enrollment --file "D:\\voice\\sample.wav" --name demo --activate --dry-run
-  yuanflow-cli ai qwen3-tts-vc-realtime-2026-01-15 --text "你好" --voice voice_xxx --output "D:\\voice\\qwen.mp3" --dry-run
+  yuanflow-cli voice replicate --text "你好" --voice voice_xxx --output "<输出音频路径>" --dry-run
+  yuanflow-cli ai voice-enrollment --file "<本地音频路径>" --name demo --activate --dry-run
+  yuanflow-cli ai cosyvoice-v3-flash --text "你好" --voice voice_xxx --output "<输出音频路径>" --dry-run
   yuanflow-cli ai fun-asr --audio-url "https://example.com/audio.wav" --response-format verbose_json --dry-run
   yuanflow-cli ai doubao-tts voices --dry-run
   yuanflow-cli ai doubao-tts voice --voice zh_female_xiaohe_uranus_bigtts --dry-run
-  yuanflow-cli ai doubao-tts voice-download --voice zh_female_xiaohe_uranus_bigtts --output "D:\\voice\\preview.mp3" --dry-run
-  yuanflow-cli ai doubao-tts --text "你好" --voice zh_female_xiaohe_uranus_bigtts --output "D:\\voice\\doubao.mp3" --dry-run
+  yuanflow-cli ai doubao-tts voice-download --voice zh_female_xiaohe_uranus_bigtts --output "<输出音频路径>" --dry-run
+  yuanflow-cli ai doubao-tts --text "你好" --voice zh_female_xiaohe_uranus_bigtts --output "<输出音频路径>" --dry-run
   yuanflow-cli list douyin
 
 说明：
@@ -650,7 +650,7 @@ function printHelp() {
   qwen3-vl-plus 支持 --image-url、--video-url、--image-file、--video-file 四选一；视频建议最大 2GB、时长 2 秒到 1 小时。
   qwen3-vl-plus 本地图片/视频会先走 YuanFlow 文件中转，内部调用 /atomic/oss/temp-upload 后把 signed_url 提交给模型。
   browser 命令是自媒体平台专用浏览器自动化协议，只返回受控 profile/cookie/任务路径与执行计划，不用于普通网页搜索。
-  视频智能剪辑和视频制作执行链已迁移到独立项目 D:\\AI_project\\视频制作，不再由 yuanflow-cli video 命令提供；视频拆解仍由 yuanflow-skill 的“视频拆解”提供。
+  视频智能剪辑和视频制作执行链已迁移到独立项目，不再由 yuanflow-cli video 命令提供；视频拆解仍由 yuanflow-skill 的“视频拆解”提供。
   需要鉴权的请求都会使用 Authorization: Bearer <token>。
   token 优先级：--token > YUANCHUANG_API_TOKEN > 本地 config.token。
   YuanFlow-main 内置环境使用时，token 由 YuanFlow-main 内置环境注入，不需要手动配置。

package/src/voice-tools.js

@@ -7,8 +7,8 @@ const AUDIO_SPEECH_PATH = '/v1/audio/speech';
 const AUDIO_VOICES_PATH = '/v1/audio/voices';
 const YUANFLOW_FILE_TRANSFER_PATH = '/atomic/oss/temp-upload';
 
-const MODEL_VOICE_CLONE = 'qwen-voice-enrollment';
-const MODEL_VOICE_REPLICATE = 'qwen3-tts-vc-realtime-2026-01-15';
+const MODEL_VOICE_CLONE = 'voice-enrollment';
+const MODEL_VOICE_REPLICATE = 'cosyvoice-v3-flash';
 
 export function listVoiceCommands() {
   return [
@@ -19,19 +19,21 @@ export function listVoiceCommands() {
       method: 'POST',
       apiPath: AUDIO_VOICES_PATH,
       options: [
-        option('--file', 'file', false, '本地音频文件；与 --file-transfer、--audio-url 三选一。'),
+        option('--file', 'file', false, '本地音频文件；通过 multipart 直接提交给 YuanFlow API，与 --file-transfer、--audio-url 三选一。'),
         option('--file-transfer', 'fileTransfer', false, '本地音频文件；先通过 YuanFlow 文件中转生成临时 URL，再创建声音克隆。'),
         option('--audio-url', 'audioUrl', false, '公网可访问音频 URL；与 --file、--file-transfer 三选一。'),
         option('--name', 'name', false, '声音克隆展示名。'),
         option('--preferred-name', 'preferredName', false, '偏好音色名，默认跟随 --name。'),
-        option('--text', 'text', false, '参考音频对应文本，可选。'),
-        option('--language', 'language', false, '语言代码，可选。'),
+        option('--target-model', 'targetModel', false, `后续合成模型，默认 ${MODEL_VOICE_REPLICATE}。`),
+        option('--language-hints', 'languageHints', false, '逗号分隔的样本音频语种提示，例如 zh。'),
+        option('--language', 'language', false, '兼容别名；会映射为 language_hints。'),
         option('--activate', 'activate', false, '创建后设为当前默认音色。'),
         ...commonOptions(),
       ],
       requestBody: {
         model: MODEL_VOICE_CLONE,
-        audio: '<本地音频 data URI，或通过 audio_url 传入 YuanFlow 文件中转 URL>',
+        target_model: MODEL_VOICE_REPLICATE,
+        file: '<multipart 本地音频，或通过 audio_url 传入公网音频 URL>',
       },
       returns: '返回 voice_xxx 音色对象；后续 voice replicate 可通过 --voice voice_xxx 复刻声音。',
     }),
@@ -69,8 +71,16 @@ export function listVoiceCommands() {
         option('--voice', 'voice', true, '声音克隆 ID：voice_xxx；也可传 default 使用已激活默认音色。'),
         option('--output', 'output', true, '音频保存路径；dry-run 时可不传。'),
         option('--response-format', 'responseFormat', false, 'mp3、wav、pcm 等，默认 mp3。'),
-        option('--speed', 'speed', false, '语速控制。'),
+        option('--rate', 'rate', false, '语速控制，会写入 metadata.rate。'),
+        option('--speed', 'speed', false, '兼容别名；会写入 metadata.rate。'),
+        option('--volume', 'volume', false, '音量控制，会写入 metadata.volume。'),
+        option('--pitch', 'pitch', false, '音调控制，会写入 metadata.pitch。'),
         option('--sample-rate', 'sampleRate', false, '采样率。'),
+        option('--language', 'language', false, '目标合成语言，例如 zh。'),
+        option('--language-hints', 'languageHints', false, '逗号分隔的目标合成语种提示，例如 zh。'),
+        option('--enable-ssml', 'enableSsml', false, '是否开启 SSML。'),
+        option('--word-timestamp-enabled', 'wordTimestampEnabled', false, '是否开启字级时间戳。'),
+        option('--seed', 'seed', false, '随机种子。'),
         option('--metadata', 'metadata', false, '透传给 YuanFlow API 的 metadata JSON。'),
         ...commonOptions(),
       ],
@@ -112,8 +122,10 @@ export async function runVoiceCommand({ action = 'help', options }) {
 }
 
 async function cloneVoice(options) {
-  const body = await buildVoiceCloneBody(options);
-  const response = await callJson(AUDIO_VOICES_PATH, options, body);
+  const { body, filePath } = await buildVoiceClonePayload(options);
+  const response = filePath
+    ? await callMultipartJson(AUDIO_VOICES_PATH, options, body, filePath)
+    : await callJson(AUDIO_VOICES_PATH, options, body);
   return result('voice clone', AUDIO_VOICES_PATH, body, response, { kind: 'voice-clone' });
 }
 
@@ -138,9 +150,9 @@ async function replicateVoice(options) {
   return result('voice replicate', AUDIO_SPEECH_PATH, body, response, { kind: 'voice-replicate' });
 }
 
-async function buildVoiceCloneBody(options) {
+async function buildVoiceClonePayload(options) {
   if (options.json) {
-    return JSON.parse(options.json);
+    return { body: JSON.parse(options.json) };
   }
   const filePath = cleanOptional(options.file);
   const fileTransferPath = cleanOptional(options.named?.['file-transfer']);
@@ -155,20 +167,24 @@ async function buildVoiceCloneBody(options) {
 
   const body = {
     model: MODEL_VOICE_CLONE,
+    target_model: cleanOptional(options.named?.['target-model']) || MODEL_VOICE_REPLICATE,
     ...optionalField('name', options.named?.name),
     ...optionalField('preferred_name', options.named?.['preferred-name']),
-    ...optionalField('text', options.named?.text),
-    ...optionalField('language', options.named?.language),
     ...optionalBooleanField('activate', options.named?.activate),
   };
+  const languageHints = splitList(options.named?.['language-hints'] || options.named?.language);
+  if (languageHints.length > 0) {
+    body.language_hints = languageHints;
+  }
   if (audioUrl) {
     body.audio_url = audioUrl;
   } else if (fileTransferPath) {
     body.audio_url = await resolveYuanFlowAudioFile(fileTransferPath, options);
   } else {
-    body.audio = options.dryRun ? '<data URI omitted in dry-run>' : await fileToDataUri(filePath);
+    body.file = '<file omitted>';
+    return { body, filePath };
   }
-  return body;
+  return { body };
 }
 
 function buildVoiceReplicateBody(options) {
@@ -187,9 +203,19 @@ function buildVoiceReplicateBody(options) {
     response_format: cleanOptional(options.named?.['response-format']) || 'mp3',
     ...optionalField('instructions', options.named?.instructions),
   };
-  addNumber(body, 'speed', options.named?.speed);
   const metadata = parseJsonObject(options.named?.metadata);
   addNumber(metadata, 'sample_rate', options.named?.['sample-rate']);
+  addNumber(metadata, 'volume', options.named?.volume);
+  addNumber(metadata, 'rate', options.named?.rate || options.named?.speed);
+  addNumber(metadata, 'pitch', options.named?.pitch);
+  addString(metadata, 'language', options.named?.language);
+  addNumber(metadata, 'seed', options.named?.seed);
+  addBoolean(metadata, 'enable_ssml', options.named?.['enable-ssml']);
+  addBoolean(metadata, 'word_timestamp_enabled', options.named?.['word-timestamp-enabled']);
+  const hints = splitList(options.named?.['language-hints']);
+  if (hints.length > 0) {
+    metadata.language_hints = hints;
+  }
   if (Object.keys(metadata).length > 0) {
     body.metadata = metadata;
   }
@@ -236,6 +262,28 @@ async function callJson(apiPath, options, body) {
   return readJsonResponse(response);
 }
 
+async function callMultipartJson(apiPath, options, body, filePath) {
+  const request = await buildRequest(apiPath, options, 'POST', body);
+  if (request.dryRun) {
+    return request;
+  }
+  const form = new FormData();
+  for (const [key, value] of Object.entries(body || {})) {
+    if (key === 'file') {
+      continue;
+    }
+    form.set(key, Array.isArray(value) ? value.join(',') : String(value));
+  }
+  const file = new Blob([await readFile(filePath)], { type: inferAudioMimeType(filePath) });
+  form.set('file', file, path.basename(filePath));
+  const response = await fetch(request.url, {
+    method: 'POST',
+    headers: request.headers,
+    body: form,
+  });
+  return readJsonResponse(response);
+}
+
 async function callGetJson(apiPath, options) {
   const request = await buildRequest(apiPath, options, 'GET');
   if (request.dryRun) {
@@ -364,11 +412,6 @@ function option(flag, name, required, label) {
   return { flag, name, required, label };
 }
 
-async function fileToDataUri(filePath) {
-  const data = await readFile(filePath);
-  return `data:${inferAudioMimeType(filePath)};base64,${data.toString('base64')}`;
-}
-
 function inferAudioMimeType(filePath) {
   switch (path.extname(filePath).toLowerCase()) {
     case '.mp3':
@@ -418,6 +461,20 @@ function addNumber(target, name, value) {
   }
 }
 
+function addString(target, name, value) {
+  const cleaned = cleanOptional(value);
+  if (cleaned !== undefined) {
+    target[name] = cleaned;
+  }
+}
+
+function addBoolean(target, name, value) {
+  const parsed = parseBoolean(value);
+  if (parsed !== undefined) {
+    target[name] = parsed;
+  }
+}
+
 function parseBoolean(value) {
   const cleaned = cleanOptional(value);
   if (cleaned === undefined) {
@@ -429,6 +486,20 @@ function parseBoolean(value) {
   return ['1', 'true', 'yes', 'on'].includes(String(cleaned).toLowerCase());
 }
 
+function splitList(value) {
+  const cleaned = cleanOptional(value);
+  if (!cleaned) {
+    return [];
+  }
+  if (Array.isArray(cleaned)) {
+    return cleaned.map((item) => String(item).trim()).filter(Boolean);
+  }
+  return String(cleaned)
+    .split(',')
+    .map((item) => item.trim())
+    .filter(Boolean);
+}
+
 function cleanOptional(value) {
   if (value === undefined || value === null) return undefined;
   if (typeof value === 'string') {