E2E Spoken Dialogue Demo Recipe

app.py

@@ -323,7 +323,12 @@ def transcribe(stream, new_chunk, option, asr_option):
         print("VAD: end of speech detected")
         start_time = time.time()
         if client is not None:
-            (text_str, audio_output)=client(array, orig_sr)
+            try:
+                (text_str, audio_output)=client(array, orig_sr)
+            except Exception as e:
+                text_str=""
+                audio_output=None
+                raise gr.Error(f"Error during audio streaming: {e}")
             asr_output_str=""
             latency_TTS=(time.time() - start_time)
         else: