generate a silent audio file if no audio file appears

Signed-off-by: David Gao <[email protected]>

Author: davidgao7

src/vllm_router/routers/main_router.py

@@ -210,7 +210,8 @@ async def get_engine_instances():
 
 @main_router.get("/health")
 async def health() -> Response:
-    """Endpoint to check the health status of various components.
+    """
+    Endpoint to check the health status of various components.
 
     This function verifies the health of the service discovery module and
     the engine stats scraper. If either component is down, it returns a
@@ -247,7 +248,7 @@ async def health() -> Response:
 
 @main_router.post("/v1/audio/transcriptions")
 async def audio_transcriptions(
-    file: UploadFile = File(...),
+    file: UploadFile | None = File(None),
     model: str = Form(...),
     prompt: str | None = Form(None),
     response_format: str | None = Form("json"),
@@ -276,7 +277,6 @@ async def audio_transcriptions(
     # logger.debug("=========files=========")
 
     data = {
-        "model": model,
         "language": language,
     }
 
@@ -300,20 +300,12 @@ async def audio_transcriptions(
     logger.debug(endpoints)
     logger.debug("==== Total endpoints ====")
 
-    # TODO: right now is skipping label check in code for local testing
-    endpoints = [
-        ep
-        for ep in endpoints
-        if model in ep.model_names  # that actually serve your model
+    # filter the endpoints url by model name for transcriptions
+    transcription_endpoints = [
+        ep for ep in endpoints
+        if model == ep.model_name and ep.model_label == "transcription"
     ]
 
-    logger.debug("==== Discovered endpoints after filtering ====")
-    logger.debug(endpoints)
-    logger.debug("==== Discovered endpoints after filtering ====")
-
-    # filter the endpoints url for transcriptions
-    transcription_endpoints = [ep for ep in endpoints if model in ep.model_names]
-
     logger.debug("====List of transcription endpoints====")
     logger.debug(transcription_endpoints)
     logger.debug("====List of transcription endpoints====")

src/vllm_router/utils.py

@@ -7,6 +7,8 @@
 import requests
 from fastapi.requests import Request
 from starlette.datastructures import MutableHeaders
+import io
+import wave
 
 from vllm_router.log import init_logger
 
@@ -77,10 +79,22 @@ def get_test_payload(model_type: str):
             case ModelType.score:
                 return {"encoding_format": "float", "text_1": "Test", "test_2": "Test2"}
             case ModelType.transcription:
+                # Generate a 0.1 second silent audio file
+                with io.BytesIO() as wav_buffer:
+                    with wave.open(wav_buffer, "wb") as wf:
+                        wf.setnchannels(1)  # mono audio channel, standard configuration
+                        wf.setsampwidth(2)  # 16 bit audio, common bit depth for wav file
+                        wf.setframerate(16000)  # 16 kHz sample rate
+                        wf.writeframes(b"\x00\x00" * 1600)  # 0.1 second of silence
+
+                    # retrieves the generated wav bytes, return
+                    wav_bytes = wav_buffer.getvalue()
+
                 return {
-                    "file": "",
+                    "file": ("empty.wav", wav_bytes, "audio/wav"),
                 }
 
+
     @staticmethod
     def get_all_fields():
         return [model_type.name for model_type in ModelType]