feat: Enhance narrator description and instructions in LLMService and Audiobook components

qwen3-tts-backend/core/audiobook_service.py

@@ -225,8 +225,16 @@ async def analyze_project(project_id: int, user: User, db: Session, turbo: bool
         if not has_narrator:
             characters_data.insert(0, {
                 "name": "narrator",
-                "description": "旁白叙述者",
-                "instruct": "中性声音，语速平稳，叙述感强"
+                "gender": "未知",
+                "description": "第三人称旁白叙述者",
+                "instruct": (
+                    "音色信息：浑厚醇厚的男性中低音，嗓音饱满有力，带有传统说书人的磁性与感染力\n"
+                    "身份背景：中国传统说书艺人，精通评书、章回小说叙述艺术，深谙故事节奏与听众心理\n"
+                    "年龄设定：中年男性，四五十岁，声音历经岁月沉淀，成熟稳重而不失活力\n"
+                    "外貌特征：面容沉稳，气度从容，台风大气，给人以可信赖的叙述者印象\n"
+                    "性格特质：沉稳睿智，叙事冷静客观，情到深处能引发共鸣，不动声色间娓娓道来\n"
+                    "叙事风格：语速适中偏慢，抑扬顿挫，擅长铺垫悬念，停顿恰到好处，语气庄重而生动，富有画面感"
+                )
             })
 
         ps.append_line(key, f"\n\n[完成] 发现 {len(characters_data)} 个角色：{', '.join(c.get('name', '') for c in characters_data)}")