feat(audiobook): add gender field to audiobook character model and update related functionality

2026-03-10 20:23:03 +08:00
parent addb152ce1
commit 2e005b0084
9 changed files with 50 additions and 8 deletions
--- a/qwen3-tts-backend/api/audiobook.py
+++ b/qwen3-tts-backend/api/audiobook.py
@@ -291,6 +291,7 @@ async def update_character(
    char = crud.update_audiobook_character(
        db, char_id,
        name=data.name,
+        gender=data.gender,
        description=data.description,
        instruct=data.instruct,
        voice_design_id=data.voice_design_id,
--- a/qwen3-tts-backend/core/audiobook_service.py
+++ b/qwen3-tts-backend/core/audiobook_service.py
@@ -195,6 +195,7 @@ async def analyze_project(project_id: int, user: User, db: Session, turbo: bool
            name = char_data.get("name", "narrator")
            instruct = char_data.get("instruct", "")
            description = char_data.get("description", "")
+            gender = char_data.get("gender") or ("未知" if name == "narrator" else None)

            voice_design = crud.create_voice_design(
                db=db,
@@ -209,6 +210,7 @@ async def analyze_project(project_id: int, user: User, db: Session, turbo: bool
                db=db,
                project_id=project_id,
                name=name,
+                gender=gender,
                description=description,
                instruct=instruct,
                voice_design_id=voice_design.id,
--- a/qwen3-tts-backend/core/llm_service.py
+++ b/qwen3-tts-backend/core/llm_service.py
@@ -119,6 +119,8 @@ class LLMService:
    async def extract_characters(self, text_samples: list[str], on_token=None, on_sample=None, turbo: bool = False) -> list[Dict]:
        system_prompt = (
            "你是一个专业的小说分析助手兼声音导演。请分析给定的小说文本，提取所有出现的角色（包括旁白narrator）。\n"
+            "gender字段必须明确标注性别，只能取以下三个值之一：\"男\"、\"女\"、\"未知\"。\n"
+            "narrator的gender固定为\"未知\"。\n"
            "对每个角色，instruct字段必须是详细的声音导演说明，需覆盖以下六个维度，每个维度单独一句，用换行分隔：\n"
            "1. 音色信息：嗓音质感、音域、音量、气息特征（如：青年男性中低音，音色干净略带沙哑，音量偏小但稳定，情绪激动时呼吸明显）\n"
            "2. 身份背景：角色身份、职业、出身、所处时代背景对声音的影响\n"
@@ -127,7 +129,7 @@ class LLMService:
            "5. 性格特质：核心性格、情绪模式、表达习惯\n"
            "6. 叙事风格：语速节奏、停顿习惯、语气色彩、整体叙述感\n\n"
            "只输出JSON，格式如下，不要有其他文字：\n"
-            '{"characters": [{"name": "narrator", "description": "第三人称叙述者", "instruct": "音色信息：...\\n身份背景：...\\n年龄设定：...\\n外貌特征：...\\n性格特质：...\\n叙事风格：..."}, ...]}'
+            '{"characters": [{"name": "narrator", "gender": "未知", "description": "第三人称叙述者", "instruct": "音色信息：...\\n身份背景：...\\n年龄设定：...\\n外貌特征：...\\n性格特质：...\\n叙事风格：..."}, ...]}'
        )
        if turbo and len(text_samples) > 1:
            logger.info(f"Extracting characters in turbo mode: {len(text_samples)} samples concurrent")
@@ -169,11 +171,12 @@ class LLMService:
            "你是一个专业的小说角色整合助手。你收到的是从同一本书不同段落中提取的角色列表，其中可能存在重复。\n"
            "请完成以下任务：\n"
            "1. 识别并合并重复角色：通过名字完全相同或高度相似（全名与简称、不同译写）来判断。\n"
-            "2. 合并时保留最完整、最详细的 description 和 instruct 字段。\n"
-            "3. narrator 角色只保留一个。\n"
+            "2. 合并时保留最完整、最详细的 description 和 instruct 字段，gender 字段以最明确的值为准（优先选\"男\"或\"女\"，而非\"未知\"）。\n"
+            "3. narrator 角色只保留一个，其 gender 固定为\"未知\"。\n"
            "4. 去除无意义的占位角色（name 为空或仅含标点）。\n"
+            "gender 字段只能取 \"男\"、\"女\"、\"未知\" 之一。\n"
            "只输出 JSON，不要有其他文字：\n"
-            '{"characters": [{"name": "...", "description": "...", "instruct": "..."}, ...]}'
+            '{"characters": [{"name": "...", "gender": "男", "description": "...", "instruct": "..."}, ...]}'
        )
        user_message = f"请整合以下角色列表：\n\n{json.dumps(raw_characters, ensure_ascii=False, indent=2)}"
        try:
--- a/qwen3-tts-backend/db/crud.py
+++ b/qwen3-tts-backend/db/crud.py
@@ -513,6 +513,7 @@ def create_audiobook_character(
    db: Session,
    project_id: int,
    name: str,
+    gender: Optional[str] = None,
    description: Optional[str] = None,
    instruct: Optional[str] = None,
    voice_design_id: Optional[int] = None,
@@ -520,6 +521,7 @@ def create_audiobook_character(
    char = AudiobookCharacter(
        project_id=project_id,
        name=name,
+        gender=gender,
        description=description,
        instruct=instruct,
        voice_design_id=voice_design_id,
@@ -558,6 +560,7 @@ def update_audiobook_character(
    db: Session,
    char_id: int,
    name: Optional[str] = None,
+    gender: Optional[str] = None,
    description: Optional[str] = None,
    instruct: Optional[str] = None,
    voice_design_id: Optional[int] = None,
@@ -567,6 +570,8 @@ def update_audiobook_character(
        return None
    if name is not None:
        char.name = name
+    if gender is not None:
+        char.gender = gender
    if description is not None:
        char.description = description
    if instruct is not None:
--- a/qwen3-tts-backend/db/database.py
+++ b/qwen3-tts-backend/db/database.py
@@ -27,3 +27,12 @@ def get_db():

 def init_db():
    Base.metadata.create_all(bind=engine)
+    if "sqlite" in str(engine.url):
+        with engine.connect() as conn:
+            try:
+                conn.execute(__import__("sqlalchemy").text(
+                    "ALTER TABLE audiobook_characters ADD COLUMN gender VARCHAR(20)"
+                ))
+                conn.commit()
+            except Exception:
+                pass
--- a/qwen3-tts-backend/db/models.py
+++ b/qwen3-tts-backend/db/models.py
@@ -169,6 +169,7 @@ class AudiobookCharacter(Base):
    id = Column(Integer, primary_key=True, index=True)
    project_id = Column(Integer, ForeignKey("audiobook_projects.id"), nullable=False, index=True)
    name = Column(String(200), nullable=False)
+    gender = Column(String(20), nullable=True)
    description = Column(Text, nullable=True)
    instruct = Column(Text, nullable=True)
    voice_design_id = Column(Integer, ForeignKey("voice_designs.id"), nullable=True)
--- a/qwen3-tts-backend/schemas/audiobook.py
+++ b/qwen3-tts-backend/schemas/audiobook.py
@@ -27,6 +27,7 @@ class AudiobookCharacterResponse(BaseModel):
    id: int
    project_id: int
    name: str
+    gender: Optional[str] = None
    description: Optional[str] = None
    instruct: Optional[str] = None
    voice_design_id: Optional[int] = None
@@ -64,6 +65,7 @@ class AudiobookCharacterUpdate(BaseModel):

 class AudiobookCharacterEdit(BaseModel):
    name: Optional[str] = None
+    gender: Optional[str] = None
    description: Optional[str] = None
    instruct: Optional[str] = None
    voice_design_id: Optional[int] = None
--- a/qwen3-tts-frontend/src/lib/api/audiobook.ts
+++ b/qwen3-tts-frontend/src/lib/api/audiobook.ts
@@ -16,6 +16,7 @@ export interface AudiobookCharacter {
  id: number
  project_id: number
  name: string
+  gender?: string
  description?: string
  instruct?: string
  voice_design_id?: number
@@ -92,7 +93,7 @@ export const audiobookApi = {
  updateCharacter: async (
    projectId: number,
    charId: number,
-    data: { name?: string; description?: string; instruct?: string; voice_design_id?: number }
+    data: { name?: string; gender?: string; description?: string; instruct?: string; voice_design_id?: number }
  ): Promise<AudiobookCharacter> => {
    const response = await apiClient.put<AudiobookCharacter>(
      `/audiobook/projects/${projectId}/characters/${charId}`,
--- a/qwen3-tts-frontend/src/pages/Audiobook.tsx
+++ b/qwen3-tts-frontend/src/pages/Audiobook.tsx
@@ -334,7 +334,7 @@ function ProjectCard({ project, onRefresh }: { project: AudiobookProject; onRefr
  const [loadingAction, setLoadingAction] = useState(false)
  const [isPolling, setIsPolling] = useState(false)
  const [editingCharId, setEditingCharId] = useState<number | null>(null)
-  const [editFields, setEditFields] = useState({ name: '', description: '', instruct: '' })
+  const [editFields, setEditFields] = useState({ name: '', gender: '', description: '', instruct: '' })
  const [sequentialPlayingId, setSequentialPlayingId] = useState<number | null>(null)
  const [charsCollapsed, setCharsCollapsed] = useState(false)
  const [chaptersCollapsed, setChaptersCollapsed] = useState(false)
@@ -520,13 +520,14 @@ function ProjectCard({ project, onRefresh }: { project: AudiobookProject; onRefr

  const startEditChar = (char: AudiobookCharacter) => {
    setEditingCharId(char.id)
-    setEditFields({ name: char.name, description: char.description || '', instruct: char.instruct || '' })
+    setEditFields({ name: char.name, gender: char.gender || '', description: char.description || '', instruct: char.instruct || '' })
  }

  const saveEditChar = async (char: AudiobookCharacter) => {
    try {
      await audiobookApi.updateCharacter(project.id, char.id, {
        name: editFields.name || char.name,
+        gender: editFields.gender || undefined,
        description: editFields.description,
        instruct: editFields.instruct,
      })
@@ -632,6 +633,16 @@ function ProjectCard({ project, onRefresh }: { project: AudiobookProject; onRefr
                          onChange={e => setEditFields(f => ({ ...f, name: e.target.value }))}
                          placeholder="角色名"
                        />
+                        <select
+                          className="w-full h-9 rounded-md border border-input bg-background px-3 py-1 text-sm"
+                          value={editFields.gender}
+                          onChange={e => setEditFields(f => ({ ...f, gender: e.target.value }))}
+                        >
+                          <option value="">性别（未设置）</option>
+                          <option value="男">男</option>
+                          <option value="女">女</option>
+                          <option value="未知">未知</option>
+                        </select>
                        <Input
                          value={editFields.instruct}
                          onChange={e => setEditFields(f => ({ ...f, instruct: e.target.value }))}
@@ -653,7 +664,14 @@ function ProjectCard({ project, onRefresh }: { project: AudiobookProject; onRefr
                      </div>
                    ) : (
                      <div className="flex flex-col gap-1 sm:flex-row sm:items-center sm:justify-between text-sm">
-                        <span className="font-medium shrink-0 truncate">{char.name}</span>
+                        <div className="flex items-center gap-1.5 shrink-0">
+                          <span className="font-medium truncate">{char.name}</span>
+                          {char.gender && (
+                            <Badge variant="outline" className={`text-xs shrink-0 ${char.gender === '男' ? 'border-blue-400/50 text-blue-400' : char.gender === '女' ? 'border-pink-400/50 text-pink-400' : 'border-muted-foreground/40 text-muted-foreground'}`}>
+                              {char.gender}
+                            </Badge>
+                          )}
+                        </div>
                        <span className="text-xs text-muted-foreground truncate sm:mx-2 sm:flex-1">{char.instruct}</span>
                        <div className="flex items-center gap-1 shrink-0">
                          {char.voice_design_id