diff --git a/backend/bgm_separation/router.py b/backend/bgm_separation/router.py
new file mode 100644
index 0000000..be6e4b1
--- /dev/null
+++ b/backend/bgm_separation/router.py
@@ -0,0 +1,61 @@
+import functools
+import numpy as np
+from fastapi import (
+    File,
+    UploadFile,
+)
+import gradio as gr
+from fastapi import APIRouter, BackgroundTasks, Depends, Response, status
+from typing import List, Dict, Tuple
+
+from modules.whisper.data_classes import *
+from modules.uvr.music_separator import MusicSeparator
+from ..util.audio import read_audio
+from ..util.schemas import QueueResponse
+from ..util.config_loader import load_server_config
+
+@functools.lru_cache
+def init_bgm_separation_inferencer() -> 'MusicSeparator':
+    config = load_server_config()["bgm_separation"]
+    inferencer = MusicSeparator()
+    inferencer.update_model(
+        model_name=config["model_size"],
+        device=config["compute_type"]
+    )
+    return inferencer
+
+bgm_separation_router = APIRouter()
+bgm_separation_inferencer = init_bgm_separation_inferencer()
+
+
+async def run_bgm_separation(
+    audio: np.ndarray,
+    params: BGMSeparationParams
+) -> Tuple[np.ndarray, np.ndarray]:
+    instrumental, vocal, filepaths = bgm_separation_inferencer.separate(
+        audio=audio,
+        model_name=params.model_size,
+        device=params.device,
+        segment_size=params.segment_size,
+        save_file=False,
+        progress=gr.Progress()
+    )
+    return instrumental, vocal
+
+
+@bgm_separation_router.post("/bgm", tags=["bgm-separation"])
+async def transcription(
+    background_tasks: BackgroundTasks,
+    file: UploadFile = File(..., description="Audio or video file to separate background music."),
+    params: TranscriptionPipelineParams = Depends()
+) -> QueueResponse:
+    if not isinstance(file, np.ndarray):
+        audio = await read_audio(file=file)
+    else:
+        audio = file
+
+    background_tasks.add_task(run_bgm_separation, audio=audio, params=params)
+
+    return QueueResponse(message="Transcription task queued")
+
+
diff --git a/backend/vad/router.py b/backend/vad/router.py
index a002e3c..64f0a02 100644
--- a/backend/vad/router.py
+++ b/backend/vad/router.py
@@ -37,7 +37,7 @@ async def run_vad(
 @vad_router.post("/vad", tags=["vad"])
 async def vad(
     background_tasks: BackgroundTasks,
-    file: UploadFile = File(..., description="Audio or video file to transcribe."),
+    file: UploadFile = File(..., description="Audio or video file to detect voices."),
     params: VadParams = Depends()
 ) -> QueueResponse:
     if not isinstance(file, np.ndarray):