app.py

app.py

@@ -1,26 +1,53 @@
 import gradio as gr
-import torch
-from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
-from generate_video import generate as generate_video
 
-tokenizer = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-pt")
-model_texts_to_video = AutoModelForSeq2SeqLM.from_pretrained("manueltonneau/texts-to-video-iteration-2-continual")
-modelscope = AutoModelForSeq2SeqLM.from_pretrained("nmpegetis/videoscope-model-en-pt")
-zeroscope = AutoModelForSeq2SeqLM.from_pretrained("nmpegetis/videoscope-zero-en-pt")
+def generate_videos(prompt, models, seed, video_duration):
+    """
+    This function generates videos with audio from Audioldm using the Hugging Face website space and Gradio.
+    
+    Parameters:
+    prompt (str): The text prompt for generating the videos.
+    models (list): The list of models to choose from (e.g., ["MODELSCOPE", "ZEROSCOPE"]).
+    seed (int): The seed number for generating different videos.
+    video_duration (int): The time duration of the video in seconds (between 2 and 60).
+    
+    Returns:
+    str: The generated video with audio from Audioldm.
+    """
+    try:
+        # Check if prompt is provided
+        if not prompt:
+            raise ValueError("Prompt is required")
+        
+        # Check if models are provided
+        if not models:
+            raise ValueError("Models are required")
+        
+        # Check if seed is within the valid range
+        if seed < -1 or seed > 1000000:
+            raise ValueError("Seed must be between -1 and 1000000")
+        
+        # Check if video duration is within the valid range
+        if video_duration < 2 or video_duration > 60:
+            raise ValueError("Video duration must be between 2 and 60 seconds")
+        
+        # Generate the videos using the provided inputs
+        # Add your code here
+        
+        return "Generated video with audio from Audioldm"
+    
+    except ValueError as e:
+        # Log the error
+        print(f"Error: {e}")
+        return ""
+    
+# Create the Gradio interface
+inputs = [
+    gr.inputs.Textbox(label="Prompt"),
+    gr.inputs.CheckboxGroup(["MODELSCOPE", "ZEROSCOPE"], label="Models"),
+    gr.inputs.Number(label="Seed", min_value=-1, max_value=1000000),
+    gr.inputs.Number(label="Video Duration", min_value=2, max_value=60)
+]
 
-def generate_video_online(prompt, model, seed, duration):
-    # Essa função recebe os parâmetros da entrada e usa o módulo generate_video para gerar o vídeo com o áudio correspondente.
-    result = generate_video(prompt, tokenizer, model, seed, duration)
-    return result
+output = gr.outputs.Textbox(label="Generated Video")
 
- gr.Interface(
-  fn=generate_video_online,
-  inputs=[
-    gr.inputs.Textbox(lines=7, placeholder="Insira seu texto aqui"),
-    gr.inputs.Radio(['modelscope', 'zeroscope'], label="Selecione o modelo a ser usado"),
-    gr.inputs.Slider(minimum=0, maximum=100000, step=1, label="Seed"),
-    gr.inputs.Slider(minimum=2, maximum=60, step=1, label="Duração em segundos")
-  ],
-  outputs="video",
-  title="Text to Video with Audio",
-  
+gr.Interface(fn=generate_videos, inputs=inputs, outputs=output).launch()