lang param.

src/runners/audio_transcriber.py

@@ -19,12 +19,17 @@ def __init__(self):
         self.model = WhisperModel(model, device=device, compute_type=compute_type)
 
     @bentoml.Runnable.method(batchable=False)
-    def transcribe_audio(self, audio):
+    def transcribe_audio(self, audio, lang):
+        if len(lang) < 2:
+            lang = "en"
+        else:
+            lang = lang[0:2]
+
         segments, info = self.model.transcribe(
             audio,
             vad_filter=True,
             vad_parameters=dict(min_silence_duration_ms=500),
-            language="en",
+            language=lang,
         )
 
         text = ""

src/stt_api.py

@@ -41,7 +41,10 @@ async def handleSticky():
 
 @app.post("/up")
 async def handleUpstream(
-    pair: str, request: Request, is_valid_brave_key=Depends(check_stt_request)
+    pair: str,
+    request: Request,
+    lang: str = "en",
+    is_valid_brave_key=Depends(check_stt_request),
 ):
     if not is_valid_brave_key:
         return JSONResponse(
@@ -57,7 +60,9 @@ async def handleUpstream(
                     if len(chunk) == 0:
                         break
                     mic_data += chunk
-                    transciption = await runner_audio_transcriber.async_run(io.BytesIO(mic_data))
+                    transciption = await runner_audio_transcriber.async_run(
+                        io.BytesIO(mic_data), lang
+                    )
                     text = transciption["text"]
                     if text:
                         await pipe.push(ipc.messages.Text(text, False))