Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,130 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import moviepy.editor as mp
|
| 3 |
+
import numpy as np
|
| 4 |
+
import librosa
|
| 5 |
+
import librosa.display
|
| 6 |
+
import matplotlib.pyplot as plt
|
| 7 |
+
import io
|
| 8 |
+
import os
|
| 9 |
+
|
| 10 |
+
# Funci贸n principal para generar el video
|
| 11 |
+
def audio_to_video(audio_file, image_file, effect_type="waveform"):
|
| 12 |
+
"""
|
| 13 |
+
Genera un video a partir de un archivo de audio y una imagen, con un efecto visual sincronizado.
|
| 14 |
+
|
| 15 |
+
Args:
|
| 16 |
+
audio_file: Ruta al archivo de audio (wav o mp3).
|
| 17 |
+
image_file: Ruta al archivo de imagen (debe ser un formato soportado por MoviePy).
|
| 18 |
+
effect_type: Tipo de efecto visual a utilizar ("waveform" por defecto, otros tipos se pueden agregar).
|
| 19 |
+
|
| 20 |
+
Returns:
|
| 21 |
+
Ruta al archivo de video generado (mp4). Si falla, retorna un mensaje de error.
|
| 22 |
+
"""
|
| 23 |
+
try:
|
| 24 |
+
# 1. Cargar el audio usando Librosa
|
| 25 |
+
y, sr = librosa.load(audio_file)
|
| 26 |
+
duration = librosa.get_duration(y=y, sr=sr)
|
| 27 |
+
|
| 28 |
+
# 2. Cargar la imagen
|
| 29 |
+
img_clip = mp.ImageClip(image_file)
|
| 30 |
+
img_clip = img_clip.set_duration(duration) # Asignar la duraci贸n del audio a la imagen
|
| 31 |
+
|
| 32 |
+
# 3. Generar el efecto visual
|
| 33 |
+
if effect_type == "waveform":
|
| 34 |
+
audio_envelope = np.abs(y) # Calculate the audio envelope
|
| 35 |
+
|
| 36 |
+
# Normalize audio envelope to image dimensions
|
| 37 |
+
audio_envelope = audio_envelope / np.max(audio_envelope)
|
| 38 |
+
audio_envelope = audio_envelope * img_clip.size[1] / 2 # Scale to half the image height
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
def make_frame(t):
|
| 42 |
+
# Create a new figure for each frame
|
| 43 |
+
fig, ax = plt.subplots(figsize=(img_clip.size[0]/100, img_clip.size[1]/100), dpi=100) # Adjust figsize for image dimensions
|
| 44 |
+
ax.set_xlim(0, duration)
|
| 45 |
+
ax.set_ylim(-img_clip.size[1] / 2, img_clip.size[1] / 2)
|
| 46 |
+
ax.axis('off') # Hide axis
|
| 47 |
+
|
| 48 |
+
# Plot waveform
|
| 49 |
+
time_index = int(t * sr)
|
| 50 |
+
wave_slice = audio_envelope[max(0,time_index - sr//10):min(len(audio_envelope), time_index + sr//10)]
|
| 51 |
+
time_slice = np.linspace(0,0.2,len(wave_slice))
|
| 52 |
+
ax.plot(np.linspace(t-0.1,t+0.1,len(wave_slice)), wave_slice-img_clip.size[1]/4, color='red')
|
| 53 |
+
ax.plot(np.linspace(t-0.1,t+0.1,len(wave_slice)), -wave_slice+img_clip.size[1]/4, color='red')
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
# Convert the Matplotlib figure to an image
|
| 57 |
+
buf = io.BytesIO()
|
| 58 |
+
fig.canvas.print_png(buf)
|
| 59 |
+
data = np.frombuffer(buf.getvalue(), dtype=np.uint8)
|
| 60 |
+
img = plt.imread(io.BytesIO(data)) #read as image
|
| 61 |
+
plt.close(fig) # Close the figure to prevent memory leaks
|
| 62 |
+
return img
|
| 63 |
+
|
| 64 |
+
|
| 65 |
+
audio_effect_clip = mp.VideoClip(make_frame, duration=duration)
|
| 66 |
+
audio_effect_clip = audio_effect_clip.set_fps(24) # Set a reasonable frame rate
|
| 67 |
+
|
| 68 |
+
else:
|
| 69 |
+
return "Error: Efecto visual no soportado."
|
| 70 |
+
|
| 71 |
+
# 4. Overlay effect onto image
|
| 72 |
+
final_clip = mp.CompositeVideoClip([img_clip, audio_effect_clip.set_pos("center")])
|
| 73 |
+
|
| 74 |
+
|
| 75 |
+
# 5. Agregar el audio al video
|
| 76 |
+
audio_clip = mp.AudioFileClip(audio_file)
|
| 77 |
+
final_clip = final_clip.set_audio(audio_clip)
|
| 78 |
+
|
| 79 |
+
|
| 80 |
+
# 6. Guardar el video
|
| 81 |
+
output_video_path = "output.mp4"
|
| 82 |
+
final_clip.write_videofile(output_video_path, fps=24, codec="libx264", audio_codec="aac") # Ajustar los par谩metros de codificaci贸n seg煤n sea necesario
|
| 83 |
+
return output_video_path
|
| 84 |
+
|
| 85 |
+
except Exception as e:
|
| 86 |
+
return f"Error: {str(e)}"
|
| 87 |
+
|
| 88 |
+
|
| 89 |
+
# ----------------------------------
|
| 90 |
+
# Gradio Interface
|
| 91 |
+
# ----------------------------------
|
| 92 |
+
|
| 93 |
+
iface = gr.Interface(
|
| 94 |
+
fn=audio_to_video,
|
| 95 |
+
inputs=[
|
| 96 |
+
gr.Audio(source="upload", type="filepath", label="Subir Archivo de Audio (WAV o MP3)"),
|
| 97 |
+
gr.Image(source="upload", type="filepath", label="Subir Imagen"),
|
| 98 |
+
gr.Radio(["waveform"], value="waveform", label="Tipo de Efecto Visual (Waveform es el 煤nico soportado por ahora)")
|
| 99 |
+
],
|
| 100 |
+
outputs="video",
|
| 101 |
+
title="Audio to Video Generator",
|
| 102 |
+
description="Sube un archivo de audio y una imagen para crear un video con un efecto visual sincronizado con la m煤sica. Por ahora solo funciona el efecto 'waveform'.",
|
| 103 |
+
examples=[["audio_example.wav", "image_example.jpg", "waveform"]] # Add example files if you want.
|
| 104 |
+
)
|
| 105 |
+
|
| 106 |
+
|
| 107 |
+
# ----------------------------------
|
| 108 |
+
# Example files (optional). Create these files
|
| 109 |
+
# or remove the 'examples' line above.
|
| 110 |
+
# ----------------------------------
|
| 111 |
+
# Create dummy audio and image for example purposes if they don't exist
|
| 112 |
+
if not os.path.exists("audio_example.wav"):
|
| 113 |
+
sr = 22050
|
| 114 |
+
T = 5
|
| 115 |
+
t = np.linspace(0, T, int(T*sr), endpoint=False)
|
| 116 |
+
x = 0.5*np.sin(2*np.pi*440*t) # A4 frequency
|
| 117 |
+
librosa.output.write_wav("audio_example.wav", x, sr)
|
| 118 |
+
|
| 119 |
+
if not os.path.exists("image_example.jpg"):
|
| 120 |
+
# Create a simple placeholder image
|
| 121 |
+
import matplotlib.pyplot as plt
|
| 122 |
+
fig, ax = plt.subplots(figsize=(6,4))
|
| 123 |
+
ax.text(0.5, 0.5, "Placeholder Image", ha="center", va="center")
|
| 124 |
+
ax.axis("off")
|
| 125 |
+
fig.savefig("image_example.jpg")
|
| 126 |
+
plt.close(fig)
|
| 127 |
+
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
iface.launch()
|