Fix: Follow exact story_reels workflow - create_captions, SRTParser, FFmpegUtils

modules/art_reels/router.py

@@ -189,20 +189,29 @@ async def generate_stick_figure_video(job_id: str, script: str, voice: str):
         
         update_job(job_id, "processing", 30)
         
-        # Step 2: Get timestamps with Whisper
-        logger.info(f"Transcribing audio for job {job_id}")
-        captions = await asyncio.to_thread(whisper_client.transcribe, audio_path)
+        # Step 2: Generate captions with Whisper (exact story_reels workflow)
+        logger.info(f"Generating captions with Whisper for job {job_id}")
         
-        # Step 3: Create 2-second chunks
-        from moviepy.editor import AudioFileClip
-        audio_clip = AudioFileClip(audio_path)
-        audio_duration = audio_clip.duration
-        audio_clip.close()
+        # WhisperClient.create_captions returns List[Caption]
+        captions = await asyncio.to_thread(whisper_client.create_captions, audio_path)
         
-        srt_parser = SRTParser()
-        chunks = srt_parser.create_2s_chunks(captions, audio_duration)
+        # Convert to dict format
+        captions_dict = [c.dict() for c in captions]
         
-        logger.info(f"Created {len(chunks)} 2-second chunks for job {job_id}")
+        # Generate .srt content (for video subtitles)
+        srt_content = SRTParser.generate_srt_content(captions_dict)
+        srt_path = os.path.join(temp_dir, "voice.srt")
+        with open(srt_path, "w", encoding="utf-8") as f:
+            f.write(srt_content)
+        logger.info(f"Generated .srt with {len(captions)} captions")
+        
+        # Get actual audio duration
+        from modules.video_creator.services.libraries.ffmpeg_utils import FFmpegUtils
+        audio_duration = FFmpegUtils.get_video_duration(audio_path)
+        
+        # Step 3: Create 2-second chunks (for AI scene generation)
+        chunks = SRTParser.create_2s_chunks(captions_dict, audio_duration)
+        logger.info(f"Created {len(chunks)} x 2s chunks for job {job_id}")
         
         update_job(job_id, "processing", 45)
         
@@ -239,7 +248,7 @@ async def generate_stick_figure_video(job_id: str, script: str, voice: str):
             audio_path=audio_path,
             output_name=f"stick_{job_id}.mp4",
             fps=30,
-            captions=captions  # Word-by-word captions from Whisper
+            captions=captions_dict  # Word-by-word captions from Whisper
         )
         
         update_job(job_id, "processing", 95)