feat: BGM音轨混音能力（音量/淡入淡出/人声闪避/预设BGM库） (#291)

apps/worker/video_processing/unified_render_service.py

@@ -165,6 +165,7 @@ class UnifiedRenderService:
         output_fps: int = DEFAULT_FPS,
         transition_duration: float = DEFAULT_TRANSITION_DURATION,
         asr_service: Any = None,  # ASRService 实例，用于自动生成字幕
+        bgm_path: str | None = None,  # BGM 本地文件路径
     ):
         self.plan = plan
         self.clips = clips
@@ -175,6 +176,7 @@ class UnifiedRenderService:
         self.output_fps = output_fps
         self.transition_duration = transition_duration
         self.asr_service = asr_service
+        self.bgm_path = bgm_path
 
     def render(self) -> RenderResult:
         """执行渲染，返回 RenderResult.
@@ -288,9 +290,55 @@ class UnifiedRenderService:
         if is_pass_through:
             # 直通场景已在一次调用中完成视频+音频
             has_audio = pass_through_has_audio
+            # 直通模式下也支持 BGM 混音：提取音频 → 混 BGM → 合并回视频
+            if self.bgm_path and pass_through_has_audio:
+                config = self.plan.config or {}
+                bgm_config = config.get("bgm", {}) or {}
+                if bgm_config.get("enabled", False):
+                    ctx = RenderContext(work_dir=self.work_dir, plan_id=self.plan.id)
+                    from video_processing.bgm_mixer import BGMConfig, mix_bgm_with_main
+
+                    bgm_cfg = BGMConfig.from_config_dict(self.bgm_path, bgm_config)
+                    # 从直通输出中提取音频
+                    main_audio_path = self.work_dir / f"pass_through_audio_{self.plan.id}.aac"
+                    extract_cmd = [
+                        FFMPEG_BIN,
+                        "-y",
+                        "-i",
+                        str(output_path),
+                        "-vn",
+                        "-acodec",
+                        "aac",
+                        "-b:a",
+                        "128k",
+                        str(main_audio_path),
+                    ]
+                    try:
+                        from video_processing.ffmpeg_utils import run_ffmpeg
+
+                        run_ffmpeg(extract_cmd)
+                        final_audio = mix_bgm_with_main(ctx, main_audio_path, bgm_cfg, video_duration)
+                        # 合并回视频
+
+                        bgm_output = self.work_dir / f"rendered_{self.plan.id}_bgm.mp4"
+                        merge_audio_video(ctx, output_path, final_audio, bgm_output)
+                        output_path = bgm_output
+                        logger.info("[unified-render] pass-through BGM mix done: plan_id=%s", self.plan.id)
+                    except Exception:
+                        logger.exception(
+                            "[unified-render] pass-through BGM mix failed, skipping: plan_id=%s", self.plan.id
+                        )
         else:
             ctx = RenderContext(work_dir=self.work_dir, plan_id=self.plan.id)
-            audio_path = mix_audio(ctx, layers, video_duration)
+            config = self.plan.config or {}
+            bgm_config = config.get("bgm", {}) or {}
+            audio_path = mix_audio(
+                ctx,
+                layers,
+                video_duration,
+                bgm_path=self.bgm_path,
+                bgm_config=bgm_config,
+            )
             t_audio_end = time.time()
             audio_mix_ms = int((t_audio_end - t_audio_start) * 1000)
             has_audio = audio_path is not None