diff --git a/apps/worker/video_processing/render_audio.py b/apps/worker/video_processing/render_audio.py
index 9de62265f..bd0388d2c 100755
--- a/apps/worker/video_processing/render_audio.py
+++ b/apps/worker/video_processing/render_audio.py
@@ -278,7 +278,9 @@ def concat_main_audio(
         has_speed = abs(speed - 1.0) >= 1e-6
 
         if not has_speed and not has_reverse:
-            # 无调速无倒放：简单命令行，-ss 裁剪更高效
+            # 无调速无倒放：简单命令行，-ss 裁剪更高效。
+            # 注意：有调速或倒放时走下方 else 分支（filter_complex），那里正确处理了
+            # atrim/atempo/reverse/volume/aformat，单 clip 简单路径不需要这些。
             command = [
                 FFMPEG_BIN,
                 "-y",
diff --git a/apps/worker/video_processing/unified_render_service.py b/apps/worker/video_processing/unified_render_service.py
index 313b30e02..fc3286bdc 100755
--- a/apps/worker/video_processing/unified_render_service.py
+++ b/apps/worker/video_processing/unified_render_service.py
@@ -1290,11 +1290,7 @@ class UnifiedRenderService:
         if has_audio:
             af_parts: list[str] = []
 
-            # 片段音量（原声=1.0，静音已在上层排除）
-            if abs(clip_volume - 1.0) >= 1e-6:
-                af_parts.append(f"volume={clip_volume:.4f}")
-
-            # 音频降噪
+            # 音频降噪（必须先降噪，再调整音量，否则音量放大会增加降噪难度）
             try:
                 from video_processing.noise_reduction_engine import NoiseReductionConfig, NoiseReductionEngine
 
@@ -1308,6 +1304,10 @@ class UnifiedRenderService:
             except Exception as e:
                 logger.warning("[unified-render] 直通模式音频降噪应用失败，跳过: %s", e)
 
+            # 片段音量（原声=1.0，静音已在上层排除；降噪后再调音量）
+            if abs(clip_volume - 1.0) >= 1e-6:
+                af_parts.append(f"volume={clip_volume:.4f}")
+
             # 音频调速（与视频setpts对应，保持音画同步）
             speed = UnifiedRenderService._clip_speed(clip)
             if abs(speed - 1.0) >= 1e-6: