small fixes

app.py

@@ -60,11 +60,7 @@ def process_uploaded_file(
     in_filename: str,
 ):
     if in_filename is None or in_filename == "":
-        return "", build_html_output(
-            "Please first upload a file and then click "
-            'the button "submit for recognition"',
-            "result_item_error",
-        )
+        return ""
 
     logging.info(f"Processing uploaded file: {in_filename}")
     try:
@@ -77,7 +73,7 @@ def process_uploaded_file(
         )
     except Exception as e:
         logging.info(str(e))
-        return "", build_html_output(str(e), "result_item_error")
+        return str(e)
 
 
 def process_microphone(
@@ -88,11 +84,10 @@ def process_microphone(
     in_filename: str,
 ):
     if in_filename is None or in_filename == "":
-        return "", build_html_output(
+        return (
             "Please first click 'Record from microphone', speak, "
             "click 'Stop recording', and then "
-            "click the button 'submit for recognition'",
-            "result_item_error",
+            "click the button 'submit for recognition'"
         )
 
     logging.info(f"Processing microphone: {in_filename}")
@@ -106,7 +101,7 @@ def process_microphone(
         )
     except Exception as e:
         logging.info(str(e))
-        return "", build_html_output(str(e), "result_item_error")
+        return str(e)
 
 
 @torch.no_grad()
@@ -153,16 +148,20 @@ def process(
     logging.info(f"Finished at {date_time} s. Elapsed: {end - start: .3f} s")
 
     info = f"""
-    Result: {text}
     Wave duration  : {duration: .3f} s <br/>
     Processing time: {end - start: .3f} s <br/>
     RTF: {end - start: .3f}/{duration: .3f} = {rtf:.3f} <br/>
     """
+    if rtf > 1:
+        info += (
+            "<br/>We are loading the model for the first run. "
+            "Please run again to measure the real RTF.<br/>"
+        )
 
     logging.info(info)
     logging.info(f"\nrepo_id: {repo_id}\nhyp: {text}")
 
-    return build_html_output(info)
+    return text
 
 
 title = "# Automatic Speech Recognition with Next-gen Kaldi"
@@ -246,7 +245,7 @@ with demo:
                 label="Upload from disk",
             )
             upload_button = gr.Button("Submit for recognition")
-            uploaded_html_info = gr.HTML(label="Info")
+            uploaded_output = gr.Textbox(label="Recognized speech from uploaded file")
 
             gr.Examples(
                 examples=examples,
@@ -257,7 +256,7 @@ with demo:
                     num_active_paths_slider,
                     uploaded_file,
                 ],
-                outputs=[uploaded_html_info],
+                outputs=[uploaded_output],
                 fn=process_uploaded_file,
             )
 
@@ -270,7 +269,7 @@ with demo:
             )
 
             record_button = gr.Button("Submit for recognition")
-            recorded_html_info = gr.HTML(label="Info")
+            recorded_output = gr.Textbox(label="Recognized speech from recordings")
 
             gr.Examples(
                 examples=examples,
@@ -281,7 +280,7 @@ with demo:
                     num_active_paths_slider,
                     microphone,
                 ],
-                outputs=[recorded_html_info],
+                outputs=[recorded_output],
                 fn=process_microphone,
             )
 
@@ -294,7 +293,7 @@ with demo:
                 num_active_paths_slider,
                 uploaded_file,
             ],
-            outputs=[uploaded_html_info],
+            outputs=[uploaded_output],
         )
 
         record_button.click(
@@ -306,7 +305,7 @@ with demo:
                 num_active_paths_slider,
                 microphone,
             ],
-            outputs=[recorded_html_info],
+            outputs=[recorded_output],
         )
     gr.Markdown(description)