Update app.py

app.py

@@ -62,10 +62,13 @@ with gr.Blocks() as demo:
     with gr.Column():
         gr.Markdown(f"# Realtime Whisper Large V3 Turbo: \n Transcribe Audio in Realtime. This Demo uses the Checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers.\n Note: The first token takes about 5 seconds. After that, it works flawlessly.")
         with gr.Row():
-            input_audio_microphone = gr.Audio(streaming=True).stream(transcribe, [input_audio_microphone, output], [output, latency_textbox], time_limit=45, stream_every=2, concurrency_limit=None)
+            input_audio_microphone = gr.Audio(streaming=True)
             output = gr.Textbox(label="Transcription", value="")
             latency_textbox = gr.Textbox(label="Latency (seconds)", value="0.0", scale=0)
         with gr.Row():
-            clear_button = gr.Button("Clear Output").click(clear, outputs=[output])
+            clear_button = gr.Button("Clear Output")
+
+        input_audio_microphone.stream(transcribe, [input_audio_microphone, output], [output, latency_textbox], time_limit=45, stream_every=2, concurrency_limit=None)
+        clear_button.click(clear, outputs=[output])
 
 demo.launch()