|
import tempfile |
|
import logging |
|
import os |
|
import asyncio |
|
from moviepy.editor import * |
|
import edge_tts |
|
import gradio as gr |
|
from pydub import AudioSegment |
|
|
|
|
|
logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s") |
|
|
|
|
|
INTRO_VIDEO = "introvideo.mp4" |
|
OUTRO_VIDEO = "outrovideo.mp4" |
|
MUSIC_BG = "musicafondo.mp3" |
|
GLITCH_SOUND = "fxsound.mp3" |
|
EJEMPLO_VIDEO = "ejemplo.mp4" |
|
|
|
|
|
for file in [INTRO_VIDEO, OUTRO_VIDEO, MUSIC_BG, GLITCH_SOUND, EJEMPLO_VIDEO]: |
|
if not os.path.exists(file): |
|
logging.error(f"Falta archivo necesario: {file}") |
|
raise FileNotFoundError(f"Falta: {file}") |
|
|
|
def eliminar_archivo_tiempo(ruta, delay=1800): |
|
def eliminar(): |
|
try: |
|
if os.path.exists(ruta): |
|
os.remove(ruta) |
|
logging.info(f"Archivo eliminado: {ruta}") |
|
except Exception as e: |
|
logging.error(f"Error al eliminar {ruta}: {e}") |
|
from threading import Timer |
|
Timer(delay, eliminar).start() |
|
|
|
async def procesar_audio(texto, voz, duracion_video, audio_original): |
|
temp_files = [] |
|
try: |
|
|
|
if not texto.strip(): |
|
raise ValueError("El texto para TTS no puede estar vac铆o.") |
|
|
|
|
|
def dividir_texto(texto, max_length=3000): |
|
return [texto[i:i + max_length] for i in range(0, len(texto), max_length)] |
|
|
|
fragmentos = dividir_texto(texto) |
|
audios_tts = [] |
|
|
|
for fragmento in fragmentos: |
|
|
|
communicate = edge_tts.Communicate(fragmento, voz) |
|
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_tts: |
|
try: |
|
await communicate.save(tmp_tts.name) |
|
except edge_tts.exceptions.NoAudioReceived as e: |
|
logging.error(f"Error en TTS: {str(e)}") |
|
raise ValueError("No se pudo generar el audio. Verifica tu conexi贸n o los par谩metros del TTS.") |
|
|
|
tts_audio = AudioFileClip(tmp_tts.name) |
|
temp_files.append(tmp_tts.name) |
|
audios_tts.append(tts_audio) |
|
|
|
|
|
tts_audio_final = concatenate_audioclips(audios_tts) |
|
|
|
|
|
if tts_audio_final.duration > duracion_video: |
|
tts_audio_final = tts_audio_final.subclip(0, duracion_video) |
|
|
|
|
|
bg_music = AudioSegment.from_mp3(MUSIC_BG) |
|
needed_ms = int(duracion_video * 1000) |
|
repeticiones = needed_ms // len(bg_music) + 1 |
|
bg_music = bg_music * repeticiones |
|
bg_music = bg_music[:needed_ms].fade_out(1000) |
|
|
|
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_bg: |
|
bg_music.export(tmp_bg.name, format="mp3") |
|
bg_audio = AudioFileClip(tmp_bg.name).volumex(0.15) |
|
temp_files.append(tmp_bg.name) |
|
|
|
|
|
audios = [bg_audio.set_duration(duracion_video)] |
|
if audio_original: |
|
audios.append(audio_original.volumex(0.7)) |
|
audios.append(tts_audio_final.volumex(0.85).set_start(0)) |
|
|
|
audio_final = CompositeAudioClip(audios).set_duration(duracion_video) |
|
return audio_final |
|
|
|
except Exception as e: |
|
logging.error(f" fallo en audio: {str(e)}") |
|
raise |
|
finally: |
|
for file in temp_files: |
|
try: |
|
os.remove(file) |
|
except Exception as e: |
|
logging.warning(f"Error limpiando {file}: {e}") |
|
|
|
def aplicar_glitch(video_clip): |
|
"""Aplica un efecto de glitch al video.""" |
|
def glitch_effect(frame): |
|
import numpy as np |
|
height, width, _ = frame.shape |
|
offset = np.random.randint(-10, 10) |
|
if offset > 0: |
|
offset = min(offset, height) |
|
if offset < 0: |
|
offset = max(offset, -height + 1) |
|
if offset!= 0 and height > 0: |
|
frame[offset:, :] = np.roll(frame[:-offset, :], -offset, axis=0) |
|
return frame |
|
|
|
return video_clip.fl_image(glitch_effect) |
|
|
|
async def procesar_video(video_input, texto_tts, voz_seleccionada): |
|
try: |
|
|
|
intro = VideoFileClip(INTRO_VIDEO) |
|
outro = VideoFileClip(OUTRO_VIDEO) |
|
video_original = VideoFileClip(video_input) |
|
audio_original = video_original.audio |
|
|
|
|
|
duracion_video = video_original.duration |
|
|
|
|
|
audio_final = await procesar_audio( |
|
texto_tts, |
|
voz_seleccionada, |
|
duracion_video, |
|
audio_original |
|
) |
|
|
|
|
|
target_width = 1920 |
|
target_height = 1080 |
|
|
|
|
|
intro_resized = intro.resize((target_width, target_height)) |
|
|
|
|
|
outro_resized = outro.resize((target_width, target_height)) |
|
|
|
|
|
video_resized = video_original.resize((target_width, target_height)) |
|
|
|
|
|
segment_duration = 20 |
|
overlap = 2 |
|
num_segments = int(duracion_video // (segment_duration - overlap)) + 1 |
|
segments = [] |
|
glitch_clips = [] |
|
glitch_sound = AudioFileClip(GLITCH_SOUND) |
|
|
|
start_time = 0 |
|
for i in range(num_segments): |
|
end_time = min(start_time + segment_duration, duracion_video) |
|
if start_time >= duracion_video: |
|
break |
|
|
|
|
|
segment = video_resized.subclip(start_time, end_time) |
|
|
|
|
|
if i > 0: |
|
glitch_segment = aplicar_glitch(segment.subclip(0, 0.5)) |
|
glitch_sound_clip = glitch_sound.set_start(start_time).volumex(0.5) |
|
glitch_clips.append(glitch_sound_clip) |
|
segment = concatenate_videoclips([glitch_segment, segment.subclip(0.5)], method="compose") |
|
|
|
segments.append(segment) |
|
|
|
|
|
start_time += segment_duration - overlap |
|
|
|
|
|
video_final = concatenate_videoclips(segments) |
|
|
|
|
|
audio_final = CompositeAudioClip([audio_final] + glitch_clips).set_duration(video_final.duration) |
|
|
|
|
|
video_con_audio = video_final.set_audio(audio_final) |
|
|
|
|
|
video_final = concatenate_videoclips( |
|
[intro_resized, video_con_audio, outro_resized], |
|
method="compose", |
|
padding=0 |
|
) |
|
|
|
|
|
with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as tmp: |
|
video_final.write_videofile( |
|
tmp.name, |
|
codec="libx264", |
|
audio_codec="aac", |
|
fps=video_original.fps, |
|
threads=4, |
|
ffmpeg_params=[ |
|
"-aspect", "16:9", |
|
"-vf", "scale=1920:1080" |
|
], |
|
verbose=False |
|
) |
|
eliminar_archivo_tiempo(tmp.name) |
|
return tmp.name |
|
except Exception as e: |
|
logging.error(f" fallo general: {str(e)}") |
|
raise |
|
|
|
|
|
with gr.Blocks() as demo: |
|
gr.Markdown("# Editor de Video con IA") |
|
|
|
with gr.Tab("Principal"): |
|
video_input = gr.Video(label="Subir video") |
|
texto_tts = gr.Textbox( |
|
label="Texto para TTS", |
|
lines=3, |
|
placeholder="Escribe aqu铆 tu texto..." |
|
) |
|
voz_seleccionada = gr.Dropdown( |
|
label="Voz", |
|
choices=["es-ES-AlvaroNeural", "es-MX-BeatrizNeural"], |
|
value="es-ES-AlvaroNeural" |
|
) |
|
procesar_btn = gr.Button("Generar Video") |
|
video_output = gr.Video(label="Video Procesado") |
|
|
|
with gr.Accordion("Ejemplos de Uso", open=False): |
|
gr.Examples( |
|
examples=[[EJEMPLO_VIDEO, "隆Hola! Esto es una prueba. Suscr铆bete al canal."]], |
|
inputs=[video_input, texto_tts], |
|
label="Ejemplos" |
|
) |
|
|
|
procesar_btn.click( |
|
procesar_video, |
|
inputs=[video_input, texto_tts, voz_seleccionada], |
|
outputs=video_output |
|
) |
|
|
|
if __name__ == "__main__": |
|
demo.queue().launch() |