Update app.py

app.py

@@ -1,115 +1,81 @@
+import os
+import gradio as gr
 import numpy as np
-from pydub import AudioSegment
-
+from magenta.music import melodies, midi_io
+from moviepy.editor import ImageSequenceClip, AudioFileClip
+from transformers import GPT2LMHeadModel, GPT2Tokenizer
+from PIL import Image
+import requests
+import torch
+
+# Load pre-trained GPT-Neo model from Hugging Face
+model_name = "EleutherAI/gpt-neo-125M"  # Open-source GPT-Neo
+tokenizer = GPT2Tokenizer.from_pretrained(model_name)
+model = GPT2LMHeadModel.from_pretrained(model_name)
+
+# Function to generate kids' music using Magenta's MelodyGenerator
 def generate_kids_music(theme):
-    # Define basic parameters
-    duration = 500  # milliseconds for each note
-    sample_rate = 44100  # Sample rate for the audio
-
-    # Define a simple scale of frequencies for the melody
-    frequencies = [261.63, 293.66, 329.63, 349.23, 392.00, 440.00, 493.88]  # C4, D4, E4, F4, G4, A4, B4
-    melody = []
-
-    # Create a simple repetitive melody pattern
-    pattern = [0, 1, 2, 3, 4, 0]  # Indices corresponding to the frequencies above
-
-    for note in pattern:
-        # Generate the sine wave for the note
-        t = np.linspace(0, duration / 1000.0, int(sample_rate * duration / 1000.0), False)
-        sine_wave = 0.5 * np.sin(2 * np.pi * frequencies[note] * t)
-        
-        # Append to melody
-        melody.append(sine_wave)
-
-    # Concatenate all notes into a single array
-    melody = np.concatenate(melody)
-
-    # Convert to an AudioSegment
-    audio_segment = AudioSegment(
-        (melody * 32767).astype(np.int16), 
-        frame_rate=sample_rate,
-        sample_width=2,
-        channels=1
-    )
-
-    # Save the audio to a file
-    audio_file = "kids_music.wav"
-    audio_segment.export(audio_file, format="wav")
-
-    # Create simple lyrics
-    lyrics = f"{theme} shark, doo doo doo doo doo doo\n" * 3 + \
-             f"{theme} shark, doo doo doo doo doo doo\n" + \
-             "It's time to play!\n" * 2
-
-    return audio_file, lyrics
-
-
-
-from transformers import pipeline
-
-# Load the GPT-2 model for text generation
-comedy_generator = pipeline('text-generation', model='gpt2')
-
-def generate_comedy_script(prompt):
-    result = comedy_generator(prompt, max_length=150, num_return_sequences=1)
-    return result[0]['generated_text']
-
-from gtts import gTTS
-
-def script_to_audio(script):
-    tts = gTTS(script, lang='en')
-    audio_file = "comedy_script.mp3"
-    tts.save(audio_file)
-    return audio_file
-
-from moviepy.editor import TextClip, AudioFileClip, CompositeVideoClip
-
-def create_animated_video(script, audio_file):
-    # Create text clips for each sentence in the script
-    sentences = script.split('. ')
-    clips = []
+    melody = melodies.MelodyGenerator()  # Create melody generator instance
+    generated_melody = melody.generate()  # Generate a random melody
     
-    # Create an animation for each sentence
-    for i, sentence in enumerate(sentences):
-        text_clip = TextClip(sentence, fontsize=40, color='white', bg_color='black', size=(640, 480), print_cmd=True)
-        text_clip = text_clip.set_duration(3).set_position('center').set_start(i * 3)  # Each sentence appears for 3 seconds
-        clips.append(text_clip)
-
-    # Load the audio file
-    audio_clip = AudioFileClip(audio_file)
-
-    # Set the duration of the video based on the audio
-    total_duration = audio_clip.duration
-    final_video = CompositeVideoClip(clips)
-    final_video = final_video.set_duration(total_duration).set_audio(audio_clip)
-
-    # Write the result to a file
-    video_file = "comedy_animation.mp4"
-    final_video.write_videofile(video_file, fps=24)
-    return video_file
-
-import gradio as gr
-
-def generate_music_and_comedy(theme, prompt):
-    # Generate kids music and lyrics
-    music_file, lyrics = generate_kids_music(theme)
-
-    script = generate_comedy_script(prompt)
-
-    # Generate audio from the comedic script
-    audio_file = generate_audio_from_script(script)
-
-    # Create an animated video from the comedic script and audio
-    video_file = create_animated_video(script, audio_file)
-
-    return video_file, music_file, lyrics
-
+    # Save the generated melody to a MIDI file
+    midi_file_path = 'output_music.mid'
+    midi_io.sequence_proto_to_midi_file(generated_melody, midi_file_path)
+    
+    return midi_file_path
+
+# Function to generate a comedic script
+def generate_script(theme):
+    prompt = f"Create a funny script for kids based on the theme: {theme}."
+    inputs = tokenizer.encode(prompt, return_tensors="pt")
+    outputs = model.generate(inputs, max_length=150, num_return_sequences=1)
+    script = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return script
+
+# Function to generate image frames based on script
+def generate_frames_from_script(script):
+    frames = []
+    for line in script.split('.'):
+        line = line.strip()
+        if line:
+            image = generate_image(line)  # Placeholder image generation
+            frames.append(image)
+    return frames
+
+# Function to generate placeholder images (replace with DALL-E Mini if needed)
+def generate_image(text):
+    image = Image.new('RGB', (1024, 1024), color=(255, 255, 255))  # Blank image
+    return image
+
+# Create a video based on the generated frames
+def create_video_from_frames(frames):
+    images = []
+    for frame in frames:
+        images.append(frame)  # Placeholder images
+    video_file_path = 'output_video.mp4'
+    clip = ImageSequenceClip(images, fps=24)
+    clip.write_videofile(video_file_path, codec='libx264')
+    return video_file_path
+
+# Main function to handle both music and comedy generation
+def generate_music_and_comedy(theme):
+    script = generate_script(theme)  # Generate a comedic script
+    music_file = generate_kids_music(theme)  # Generate kid-friendly music
+    frames = generate_frames_from_script(script)  # Generate frames from script
+    video_file = create_video_from_frames(frames)  # Create video from frames
+
+    return video_file, music_file
+
+# Gradio UI setup
 interface = gr.Interface(
     fn=generate_music_and_comedy,
-    inputs=["text", "text"],
-    outputs=["video", "audio", "text"],
-    title="AI Music and Comedy Animation Generation",
-    description="Generate unique music tracks for kids and animated comedic videos."
+    inputs=gr.Textbox(label="Theme for Kids Music & Comedy"),
+    outputs=[gr.Video(label="Generated Animated Video"), gr.File(label="Generated Music")],
+    title="AI Kids Music & Comedy Generator",
+    description="Generate unique music and animated videos for kids based on a theme."
 )
 
-interface.launch()
+# Launch the interface
+if __name__ == "__main__":
+    interface.launch()
+