feat: 视频调速引擎（快进/慢放） (#294)

apps/worker/video_processing/render_audio.py

@@ -23,6 +23,7 @@ from typing import TYPE_CHECKING
 
 from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_has_audio, run_ffmpeg
 from video_processing.reverse_engine import ReverseConfig, ReverseEngine
+from video_processing.speed_engine import SpeedEngine
 
 if TYPE_CHECKING:
     from video_processing.unified_render_service import RenderLayer, ResolvedClip
@@ -225,53 +226,118 @@ def concat_main_audio(
         clip = clips[0]
         effective_duration = clip_effective_duration(clip)
         trim_start = getattr(clip, "start_time", 0) or 0
-        # 最终时长：取 clip 有效时长和视频总时长的较小值
-        # （视频总时长由主图层决定，但单 clip 场景下两者应该一致，仍做保护）
-        final_duration = effective_duration
+        speed = getattr(clip, "playback_speed", 1.0) or 1.0
+        if not isinstance(speed, (int, float)) or speed <= 0:
+            speed = 1.0
+
+        # 调速后时长
+        adjusted_duration = effective_duration / speed if abs(speed - 1.0) >= 1e-6 else effective_duration
+        # 最终时长：取调速后时长和视频总时长的较小值
+        final_duration = adjusted_duration
         if video_duration > 0 and (final_duration <= 0 or final_duration > video_duration):
             final_duration = video_duration
 
         # 音频倒放
         reverse_config = ReverseConfig.from_dict(clip.config.get("reverse"))
-        af_filters = []
-        if reverse_config.enabled and reverse_config.reverse_audio:
-            reverse_filter = ReverseEngine.build_audio_filter(reverse_config, duration=effective_duration)
-            if reverse_filter:
-                af_filters.append(reverse_filter)
+        has_reverse = reverse_config.enabled and reverse_config.reverse_audio
+        has_speed = abs(speed - 1.0) >= 1e-6
 
-        command = [
-            FFMPEG_BIN,
-            "-y",
-            "-i",
-            str(clip.local_path),
-            "-vn",
-            "-acodec",
-            "aac",
-            "-b:a",
-            "128k",
-        ]
-        if trim_start > 0:
-            command.extend(["-ss", f"{trim_start:.3f}"])
-        if af_filters:
-            command.extend(["-af", ",".join(af_filters)])
-        if final_duration > 0:
-            command.extend(["-t", f"{final_duration:.3f}"])
-        command.append(str(output_path))
-        run_ffmpeg(command)
+        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
+            command = [
+                FFMPEG_BIN,
+                "-y",
+                "-i",
+                str(clip.local_path),
+                "-vn",
+                "-acodec",
+                "aac",
+                "-b:a",
+                "128k",
+            ]
+            if trim_start > 0:
+                command.extend(["-ss", f"{trim_start:.3f}"])
+            if final_duration > 0:
+                command.extend(["-t", f"{final_duration:.3f}"])
+            command.append(str(output_path))
+            run_ffmpeg(command)
+        else:
+            # 有调速或倒放：用 filter_complex
+            speed_engine = SpeedEngine()
+            audio_filters = []
+            if effective_duration > 0:
+                audio_filters.append(f"atrim=start={trim_start:.3f}:duration={effective_duration:.3f}")
+                audio_filters.append("asetpts=PTS-STARTPTS")
+
+            # 音频调速
+            if has_speed:
+                from video_processing.speed_engine import SpeedConfig
+
+                config = SpeedConfig(speed=float(speed))
+                config.clamp()
+                atempo_filter = speed_engine.build_audio_filter(config)
+                if atempo_filter:
+                    audio_filters.append(atempo_filter)
+
+            # 音频倒放
+            if has_reverse:
+                reverse_filter = ReverseEngine.build_audio_filter(reverse_config, duration=effective_duration)
+                if reverse_filter:
+                    audio_filters.append(reverse_filter)
+
+            filter_parts: list[str] = [f"[0:a]{','.join(audio_filters)}[outa]"]
+            if video_duration > 0 and final_duration < adjusted_duration:
+                filter_parts.append(f"[outa]atrim=0:{final_duration:.3f}[final_audio]")
+                final_label = "final_audio"
+            else:
+                final_label = "outa"
+
+            filter_complex = ";".join(filter_parts)
+            command = [
+                FFMPEG_BIN,
+                "-y",
+                "-i",
+                str(clip.local_path),
+                "-filter_complex",
+                filter_complex,
+                "-map",
+                f"[{final_label}]",
+                "-acodec",
+                "aac",
+                "-b:a",
+                "128k",
+                str(output_path),
+            ]
+            run_ffmpeg(command)
         return
 
     # 多 clip，用 filter_complex concat
     input_args: list[str] = []
     filter_parts: list[str] = []
+    speed_engine = SpeedEngine()
 
     for i, clip in enumerate(clips):
         input_args.extend(["-i", str(clip.local_path)])
         effective_duration = clip_effective_duration(clip)
         trim_start = getattr(clip, "start_time", 0) or 0
+        speed = getattr(clip, "playback_speed", 1.0) or 1.0
+        if not isinstance(speed, (int, float)) or speed <= 0:
+            speed = 1.0
+
         audio_filters: list[str] = []
         if effective_duration > 0:
             audio_filters.append(f"atrim=start={trim_start:.3f}:duration={effective_duration:.3f}")
             audio_filters.append("asetpts=PTS-STARTPTS")
+
+        # 音频调速 — atempo 多级串联
+        if abs(speed - 1.0) >= 1e-6:
+            from video_processing.speed_engine import SpeedConfig
+
+            config = SpeedConfig(speed=float(speed))
+            config.clamp()
+            atempo_filter = speed_engine.build_audio_filter(config)
+            if atempo_filter:
+                audio_filters.append(atempo_filter)
         else:
             audio_filters.append("asetpts=PTS-STARTPTS")