Upload 2 files

app.py

@@ -0,0 +1,67 @@
+import gradio as gr
+import whisper
+import torch
+import os
+import ffmpeg
+from flask import send_file
+import time
+from deep_translator import GoogleTranslator  
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+print("Usando dispositivo:", device)
+
+model = whisper.load_model("medium").to(device)  
+
+def extract_audio(video_path, output_audio="temp_audio.wav"):
+    ffmpeg.input(video_path).output(output_audio, format="wav", acodec="pcm_s16le", ac=1, ar="16k").run(overwrite_output=True, quiet=True)
+    return output_audio
+
+def translate_text(text, target_lang="es"):
+    return GoogleTranslator(source="auto", target=target_lang).translate(text)
+
+def transcribe_audio(audio_path, target_language="es"):
+    start_time = time.time()
+    result = model.transcribe(audio_path, fp16=(device == "cuda"), task="translate", beam_size=5)
+    translated_text = translate_text(result["text"], target_language)  
+    end_time = time.time()
+    print(f"Tiempo de transcripción y traducción: {end_time - start_time:.2f} segundos")
+    return translated_text
+
+def save_transcription(text, format, output_name="translation"):
+    file_path = f"{output_name}.{format.lower()}"
+    with open(file_path, "w", encoding="utf-8") as f:
+        f.write(text)
+    return file_path  
+
+def process_media(file, format, target_language):
+    if not os.path.exists(file):
+        return "Error: No se encontró el archivo."
+    extension = file.split(".")[-1].lower()
+    audio_extensions = ["mp3", "wav", "flac", "ogg", "m4a"]
+    video_extensions = ["mp4", "avi", "mov", "mkv"]
+
+    if extension in video_extensions:
+        audio_path = extract_audio(file)
+    elif extension in audio_extensions:
+        audio_path = file
+    else:
+        return "Error: Formato de archivo no compatible."
+
+    transcription = transcribe_audio(audio_path, target_language)
+    file_path = save_transcription(transcription, format)
+    
+    return transcription, file_path
+
+gui = gr.Interface(
+    fn=process_media,
+    inputs=[
+        gr.File(label="Sube un archivo de audio o vídeo"),
+        gr.Radio(["TXT", "DOCX", "SRT", "JSON"], label="Formato de salida"),
+        gr.Dropdown(["es", "en", "fr", "de", "it"], label="Idioma de traducción", value="es")
+    ],
+    outputs=["text", "file"],  
+    title="Transcriptor y Traductor de Audio y Vídeo",
+    description="Sube un archivo, transcribe y traduce a cualquier idioma con alta precisión."
+)
+
+gui.launch()

requirements.txt

@@ -0,0 +1,8 @@
+torch
+torchaudio
+torchvision
+openai-whisper
+deep-translator
+ffmpeg-python
+gradio
+flask