app.py

import gradio as gr
import numpy as np
from moviepy.editor import AudioFileClip, ImageClip, CompositeVideoClip
import tempfile
import os

def create_waveform_video(image, audio):
    # Save uploaded files temporarily
    image_path = tempfile.mktemp(suffix=".png")
    audio_path = tempfile.mktemp(suffix=".mp3")
    video_path = tempfile.mktemp(suffix=".mp4")
    
    image.save(image_path)
    audio.save(audio_path)
    
    # Load audio
    audio_clip = AudioFileClip(audio_path)
    
    # Load image and create a video clip of the same duration as the audio
    img_clip = ImageClip(image_path).set_duration(audio_clip.duration)
    img_clip = img_clip.resize(height=720)  # Resize image to a suitable video height
    
    # Generate waveform effect as a video
    waveform = audio_clip.to_soundarray(fps=22050)  # Convert audio to waveform array
    waveform = waveform.mean(axis=1)  # Convert stereo to mono by averaging channels
    
    # Normalize waveform for overlay
    waveform = np.interp(waveform, (waveform.min(), waveform.max()), (-1, 1))
    
    # Overlay waveform on image (optional customization: size and position of waveform)
    waveform_clip = ImageClip(image_path).set_duration(audio_clip.duration)
    waveform_clip = waveform_clip.fx(CompositeVideoClip, [[waveform, 'center', 0.8]]) 

    # Combine image and audio into one video
    final_clip = CompositeVideoClip([img_clip, waveform_clip.set_audio(audio_clip)])
    final_clip.write_videofile(video_path, codec="libx264", fps=24)
    
    # Cleanup temporary files
    os.remove(image_path)
    os.remove(audio_path)
    
    return video_path

iface = gr.Interface(
    fn=create_waveform_video,
    inputs=[
        gr.Image(type="pil", label="Upload Image"),
        gr.Audio(type="file", label="Upload Audio")
    ],
    outputs=gr.Video(label="Generated Video"),
    title="Image + Audio to Video with Waveform Overlay"
)

iface.launch()