Switched over to using blocks to give greater control over what's displayed

app.py

@@ -93,14 +93,26 @@ input_audio = gr.Audio(
         show_controls=False,
     ),
 )
-demo = gr.Interface(
-    fn=transcribe_audio,
-    inputs=[diction, input_audio],
-    outputs=highlighted_results,
-    title="Test your diction",
-    description=description,
-    theme="abidlabs/Lime",
-)
+
+# demo = gr.Interface(
+#     fn=transcribe_audio,
+#     inputs=[diction, input_audio],
+#     outputs=highlighted_results,
+#     title="Test your diction",
+#     description=description,
+#     theme="abidlabs/Lime",
+# )
+
+with gr.Blocks() as demo:
+    gr.HTML(description)
+    gr.HTML(test_text)
+
+    with gr.Row():
+        inp = input_audio
+        out = highlighted_results
+    btn = gr.Button("Run")
+    btn.click(fn=transcribe_audio, inputs=[diction_text, inp], outputs=out)
+
 
 if __name__ == "__main__":
     demo.launch()