Update app.py

app.py

@@ -3,13 +3,36 @@ import os
 import asyncio
 from conver import ConversationConfig, URLToAudioConverter
 from dotenv import load_dotenv
+from pydub import AudioSegment
 
 load_dotenv()
 
-def synthesize_sync(article_url, text_input, language, skip_llm):
-    return asyncio.run(synthesize(article_url, text_input, language, skip_llm))
+MUSICA_FONDO = "musica.mp3"
+TAG1 = "tag.mp3"
+TAG2 = "tag2.mp3"
 
-async def synthesize(article_url, text_input, language="en", skip_llm=False):
+def mezclar_musica_y_tags(audio_path: str) -> str:
+    podcast_audio = AudioSegment.from_file(audio_path)
+    musica_fondo = AudioSegment.from_file(MUSICA_FONDO).apply_gain(-15)
+    tag1_audio = AudioSegment.from_file(TAG1).apply_gain(-5)
+    tag2_audio = AudioSegment.from_file(TAG2).apply_gain(-5)
+
+    duracion_podcast = len(podcast_audio)
+    repeticiones = (duracion_podcast // len(musica_fondo)) + 1
+    musica_fondo_loop = musica_fondo * repeticiones
+    musica_fondo_loop = musica_fondo_loop[:duracion_podcast]
+
+    mezcla = musica_fondo_loop.overlay(podcast_audio)
+    mezcla = tag1_audio + mezcla + tag2_audio
+
+    output_path = audio_path.replace(".mp3", "_con_musica.mp3")
+    mezcla.export(output_path, format="mp3")
+    return output_path
+
+def synthesize_sync(article_url, text_input, language, skip_llm, agregar_musica):
+    return asyncio.run(synthesize(article_url, text_input, language, skip_llm, agregar_musica))
+
+async def synthesize(article_url, text_input, language="en", skip_llm=False, agregar_musica=False):
     if not article_url and not text_input:
         return "Error: Ingresa una URL o texto", None
 
@@ -17,24 +40,22 @@ async def synthesize(article_url, text_input, language="en", skip_llm=False):
         config = ConversationConfig()
         converter = URLToAudioConverter(config, llm_api_key=os.environ.get("TOGETHER_API_KEY"))
         
-        # Voces humanizadas
         voices = {
             "en": ("en-US-AvaMultilingualNeural", "en-US-AndrewMultilingualNeural"),
             "es": ("es-ES-AlvaroNeural", "es-ES-ElviraNeural")
         }
         voice1, voice2 = voices.get(language, voices["en"])
 
-        # Modo sin LLM (texto exacto)
         if skip_llm and text_input:
             output_file, conversation = await converter.raw_text_to_audio(text_input, voice1, voice2)
-            return conversation, output_file
-            
-        # Procesamiento normal (con LLM)
-        if text_input:
+        elif text_input:
             output_file, conversation = await converter.text_to_audio(text_input, voice1, voice2)
         else:
             output_file, conversation = await converter.url_to_audio(article_url, voice1, voice2)
-            
+
+        if agregar_musica:
+            output_file = mezclar_musica_y_tags(output_file)
+
         return conversation, output_file
     except Exception as e:
         return f"Error: {str(e)}", None
@@ -46,6 +67,7 @@ with gr.Blocks(theme='gstaff/sketch') as demo:
         text_input = gr.Textbox(label="Texto manual", lines=5, placeholder="Pega tu texto aquí...")
         language = gr.Dropdown(["en", "es"], label="Idioma", value="en")
         skip_llm = gr.Checkbox(label="🔴 Modo libre (sin filtros LLM)", value=False)
+        agregar_musica = gr.Checkbox(label="🎵 Agregar música de fondo y cortinillas", value=False)
         btn = gr.Button("Generar Podcast", variant="primary")
     
     with gr.Row():
@@ -54,8 +76,8 @@ with gr.Blocks(theme='gstaff/sketch') as demo:
     
     btn.click(
         synthesize_sync,
-        inputs=[text_url, text_input, language, skip_llm],
+        inputs=[text_url, text_input, language, skip_llm, agregar_musica],
         outputs=[conv_display, aud]
     )
 
-demo.launch()
+demo.launch()