Add VibeVoice-ASR backend

Third ASR backend option alongside faster-whisper and qwen3. Builds
microsoft/VibeVoice from source (pinned to a specific commit, since it's
custom modeling code not in transformers' Auto* registry) into its own venv,
inheriting the base image's torch/CUDA. Comes with built-in speaker
diarization (VibeVoiceASRProcessor.post_process_transcription returns
per-segment speaker ids directly, no separate pyannote pass needed).

Verified end-to-end against a real recording: 200 OK, correct Korean
transcription, speaker labels populated.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

app/asr/router.py

@@ -18,6 +18,7 @@ from asr.config import (
     DEFAULT_MODEL,
     FASTER_WHISPER_URL,
     QWEN3_URL,
+    VIBEVOICE_URL,
     RESULT_DIR,
     UPLOAD_DIR,
     ensure_runtime_dirs,
@@ -29,6 +30,7 @@ ws_router = APIRouter()
 BACKENDS = {
     "faster-whisper": FASTER_WHISPER_URL,
     "qwen3": QWEN3_URL,
+    "vibevoice": VIBEVOICE_URL,
 }
 
 REALTIME_SAMPLE_RATE = 16000
@@ -50,6 +52,9 @@ def config() -> Dict[str, Any]:
             "qwen3": {
                 "models": ["Qwen/Qwen3-ASR-0.6B-hf", "Qwen/Qwen3-ASR-1.7B-hf"],
             },
+            "vibevoice": {
+                "models": ["microsoft/VibeVoice-ASR"],
+            },
         },
     }
 
@@ -168,6 +173,10 @@ async def transcribe(
                 "language": effective_language,
                 "task": task,
             }
+        elif backend == "vibevoice":
+            data = {
+                "model": model,
+            }
         else:
             # faster-whisper
             data = {