Update app.py

app.py

@@ -4,7 +4,6 @@ from fastapi import FastAPI, HTTPException
 from fastapi.responses import StreamingResponse
 from pydantic import BaseModel
 from transformers import (
-    AutoConfig,
     AutoModelForCausalLM,
     AutoTokenizer,
     GenerationConfig,
@@ -59,7 +58,7 @@ class S3ModelLoader:
         )
 
     def _get_s3_uri(self, model_name):
-        return f"s3://{self.bucket_name}/{model_name}"
+        return f"s3://{self.bucket_name}/{model_name.replace('/', '-')}"
     
     def load_model_and_tokenizer(self, model_name):
         if model_name in token_dict:
@@ -224,7 +223,7 @@ async def generate_text_to_speech(request: GenerateRequest):
         audio = audio_generator(validated_body.input_text)[0]
 
         audio_byte_arr = BytesIO()
-        audio.save(audio_byte_arr)
+        audio.save(audio_byte_arr, format="WAV")
         audio_byte_arr.seek(0)
 
         return StreamingResponse(audio_byte_arr, media_type="audio/wav")
@@ -232,24 +231,6 @@ async def generate_text_to_speech(request: GenerateRequest):
     except Exception as e:
         raise HTTPException(status_code=500, detail=f"Internal server error: {str(e)}")
 
-# Endpoint para generar video
-@app.post("/generate-video")
-async def generate_video(request: GenerateRequest):
-    try:
-        validated_body = request
-        device = "cuda" if torch.cuda.is_available() else "cpu"
-        video_generator = pipeline("text-to-video", model=validated_body.model_name, device=device)
-        video = video_generator(validated_body.input_text)[0]
-
-        video_byte_arr = BytesIO()
-        video.save(video_byte_arr, format="MP4")
-        video_byte_arr.seek(0)
-
-        return StreamingResponse(video_byte_arr, media_type="video/mp4")
-
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=f"Internal server error: {str(e)}")
-
 # Ejecutar el servidor FastAPI con Uvicorn
 if __name__ == "__main__":
     import uvicorn