change /chat endpoint to return a url

self_hosted_env/voice_agent/main.py

@@ -148,32 +148,17 @@ def create_app() -> FastAPI:
         """Fetches the agent's text and audio response with proper CORS headers."""
         try:
             print("Getting audio response...")
+            
             result = container.response_service.get_agent_response(student_id=student_id)
             
+
+
             if hasattr(result, 'status_code'):
                 # This is already a Response object from response_service
                 print(f"Response headers: {dict(result.headers)}")
+                print(f"Response audio raw bytes size: {len(result.body) if result.body else 'N/A'}")
+                print(f"Response audio first 20 bytes: {result.body[:20] if result.body else 'N/A'}")
                 return result
-            else:
-                # If it's not a Response object, create one
-                print("Creating new response object")
-                response_text = result.get('text', 'No response available')
-                
-                # Encode the response text in base64 for the header
-                encoded_text = base64.b64encode(response_text.encode('utf-8')).decode('utf-8')
-                
-                # Create response with proper headers
-                response = Response(
-                    content=result.get('audio', b''),
-                    media_type="audio/mpeg",
-                    headers={
-                        "X-Response-Text": encoded_text,
-                        "Access-Control-Expose-Headers": "X-Response-Text",
-                        "Access-Control-Allow-Origin": "*",
-                        "Access-Control-Allow-Methods": "GET, POST, OPTIONS",
-                        "Access-Control-Allow-Headers": "*"
-                    }
-                )
 
                 print(f"Created response with headers: {dict(response.headers)}")
                 return response

self_hosted_env/voice_agent/repositories/storage.py

@@ -44,3 +44,16 @@ class MinIOStorageRepository(StorageRepository):
             return "connected"
         except Exception as e:
             return f"error: {str(e)}"
+
+    def get_file_url(self, bucket: str, file_path: str, expires: int = 3600) -> str:
+        """Generate a presigned URL to share an S3 object"""
+        try:
+            url = self.client.generate_presigned_url(
+                'get_object',
+                Params={'Bucket': bucket, 'Key': file_path},
+                ExpiresIn=expires
+            )
+            return url
+        except ClientError as e:
+            print(f"Error generating presigned URL: {e}")
+            return None
\ No newline at end of file

self_hosted_env/voice_agent/services/chat_service.py

@@ -60,7 +60,7 @@ class ChatService:
             self.response_manager.store_response(
                 student_id = student_id,
                 text=agent_response_text,
-                audio_filename=audio_data.get("filename"),
+                audio_filepath=audio_data.get("filepath"),
                 audio_bytes=audio_data.get("bytes")
             )
             
@@ -71,7 +71,7 @@ class ChatService:
                 "message": "Message processed and agent response ready",
                 "student_id": student_id,
                 "agent_response": agent_response_text,
-                "audio_filename": audio_data.get("filename")
+                "audio_filepath": audio_data.get("filepath")
             }
             
         except Exception as e:
@@ -93,11 +93,11 @@ class ChatService:
 
             # 3. Determine filename and upload (same as before)
             provider = os.getenv("TTS_PROVIDER", "openai").lower()
-            file_extension = "wav" if provider == "custom" else "mp3"
-            content_type = "audio/wav" if provider == "custom" else "audio/mpeg"
+            file_extension = "wav"
+            content_type = "audio/wav"
             
             timestamp = int(time.time())
-            filename = f"agent_response_{timestamp}.{file_extension}"
+            filename = f"agent_response_{timestamp}_{student_id}.{file_extension}"
             minio_file_path = f"audio/{filename}"
 
             self.storage_repo.upload_file(
@@ -106,9 +106,14 @@ class ChatService:
                 file_path=minio_file_path
             )
             
+            full_url = self.storage_repo.get_file_url(
+                bucket=self.config.minio_bucket,
+                file_path=minio_file_path,
+                expires=3600  # 1 hour
+            )
             print(f"Successfully generated and uploaded stitched TTS audio: {filename}")
-            return {"bytes": audio_bytes, "filename": filename}
+            return {"bytes": audio_bytes, "filepath": full_url}
 
         except Exception as e:
             print(f"Error in _generate_and_upload_audio: {e}")
-            return {"bytes": None, "filename": None}
\ No newline at end of file
+            return {"bytes": None, "filepath": None}
\ No newline at end of file

self_hosted_env/voice_agent/services/openai_service.py

@@ -69,7 +69,7 @@ class OpenAIService(BaseTTSService):
                 model=Models.tts,
                 voice=voice,
                 input=text,
-                response_format="mp3"
+                response_format="wav"
             )
             
             audio_bytes = response.content

self_hosted_env/voice_agent/services/response_manager.py

@@ -22,7 +22,7 @@ class ResponseManager:
         """Creates a consistent key for Redis to avoid conflicts."""
         return f"student_response:{student_id}"
 
-    def store_response(self, student_id: str, text: str, audio_filename: Optional[str] = None, audio_bytes: Optional[bytes] = None) -> None:
+    def store_response(self, student_id: str, text: str, audio_filepath: Optional[str] = None, audio_bytes: Optional[bytes] = None) -> None:
         """Stores a response for a specific student_id in Redis."""
         key = self._get_key(student_id)
         
@@ -31,7 +31,7 @@ class ResponseManager:
         
         payload = {
             "text": text,
-            "audio_filename": audio_filename,
+            "audio_filepath": audio_filepath,
             "audio_bytes_b64": encoded_audio
         }