Update app.py

app.py

@@ -26,11 +26,6 @@ from threading import Timer
 from flask import Flask, render_template, request, url_for, send_from_directory
 from google import genai
 
-# New imports for audio generation and handling
-from kokoro import KPipeline
-import soundfile as sf
-import numpy as np
-
 app = Flask(__name__)
 
 # Load API key from environment variable
@@ -67,11 +62,7 @@ Keep the scene uncluttered and aesthetically pleasing.
 Make sure things are not overlapping unless explicitly stated otherwise.
 It is crucial that the script works correctly on the first try, so make sure to think about the layout and storyboard and stuff of the scene.
 Make sure to think through what you are going to do and think about the topic before you write the code.
-In addition, write a commentary script inside of ```script``` codeblock. This should be short and fit the content and align with the timing of the scene. Use "..." if needed to add a bit of a pause.
-Don't describe music or sound effects in your script, only what the text to speech model will say.
-Make sure that the script is short and don't do an intro and outro, simply start it. Additionally, keep it short enough so that it aligns with the video timing.
-You can always adjust the self.wait(x) or run_time=x inside the code to wait for the speech if needed.
-Make sure to think about the script timing before you write the code so that it doesnt cut off the speech or anything and the video and speech are synced
+Use self.wait at the end to give the user enough time to process the frame.
 You got this!! <3
 """
                 )
@@ -82,13 +73,7 @@ You got this!! <3
                 if not code_match:
                     raise Exception("No python code block found in the AI response.")
                 code = code_match.group(1)
-                
-                # Extract the commentary script from the AI response
-                script_pattern = r"```script\s*(.*?)\s*```"
-                script_match = re.search(script_pattern, ai_response.text, re.DOTALL)
-                if not script_match:
-                    raise Exception("No script block found in the AI response.")
-                script = script_match.group(1)
+
                 
                 # Determine the scene class name from the generated code
                 scene_match = re.search(r"class\s+(\w+)\(.*Scene.*\):", code)
@@ -103,24 +88,6 @@ You got this!! <3
                 with open(code_filepath, "w") as f:
                     f.write(code)
                 
-                # === Generate Commentary Audio via Kokoro ===
-                # Initialize the Kokoro pipeline (adjust lang_code if needed)
-                audio_pipeline = KPipeline(lang_code='a')
-                # Feed in the commentary script; here we split the text by one or more newlines.
-                audio_generator = audio_pipeline(script, voice='af_heart', speed=1, split_pattern=r'\n+')
-                
-                audio_segments = []
-                for _, _, audio in audio_generator:
-                    audio_segments.append(audio)
-                
-                if not audio_segments:
-                    raise Exception("No audio segments were generated from the commentary script.")
-                
-                # Concatenate all audio segments into one audio track
-                full_audio = np.concatenate(audio_segments)
-                commentary_audio_filename = f"commentary_{uuid.uuid4().hex}.wav"
-                commentary_audio_path = os.path.join("/tmp", commentary_audio_filename)
-                sf.write(commentary_audio_path, full_audio, 24000)
                 
                 # === Run Manim to generate the silent video ===
                 # Prepare the Manim command with the --media_dir flag
@@ -147,26 +114,7 @@ You got this!! <3
                 # Move the video file to /tmp (to serve it from there)
                 tmp_video_path = os.path.join("/tmp", video_filename)
                 shutil.move(video_path_in_media, tmp_video_path)
-                
-                # === Combine Video with Commentary Audio using FFmpeg ===
-                final_video_filename = f"final_video_{uuid.uuid4().hex}.mp4"
-                final_video_path = os.path.join("/tmp", final_video_filename)
-                
-                ffmpeg_cmd = [
-                    "ffmpeg", "-y",
-                    "-i", tmp_video_path,
-                    "-i", commentary_audio_path,
-                    "-c:v", "copy",
-                    "-c:a", "aac",
-                    "-shortest",
-                    final_video_path
-                ]
-                try:
-                    subprocess.run(ffmpeg_cmd, check=True, capture_output=True, text=True)
-                except subprocess.CalledProcessError as cpe:
-                    app.logger.error("FFmpeg error output: %s", cpe.stderr)
-                    raise Exception(f"FFmpeg failed: {cpe.stderr}")
-                
+
                 # Schedule deletion of all temporary files after 10 minutes (600 seconds)
                 def remove_files():
                     for fpath in [tmp_video_path, code_filepath, commentary_audio_path, final_video_path]:
@@ -179,7 +127,7 @@ You got this!! <3
                 Timer(600, remove_files).start()
                 
                 # Use the final combined video for display
-                video_url = url_for('get_video', filename=final_video_filename)
+                video_url = url_for('get_video', filename=tmp_video_path)
                 return render_template("result.html", video_url=video_url)
             
             except Exception as e: