Add logs to debug

app.py

@@ -33,16 +33,15 @@ def get_audio(url):
     video = yt.streams.filter(only_audio=True).first()
     out_file = video.download(output_path=".")
     base, ext = os.path.splitext(out_file)
-    print("the extension is", ext)
     new_file = base + ".wav"
     os.rename(out_file, new_file)
     a = new_file
-    
+    print("file a is:", a)
     wav_to_cut = AudioSegment.from_wav(a)
     # pydub does things in milliseconds
     ten_seconds = 10 * 1000
     first_10_seconds = wav_to_cut[:ten_seconds]
-    
+    os.remove(new_file)
     return first_10_seconds
 
 # Credits https://huggingface.co/spaces/jeffistyping/Youtube-Whisperer
@@ -67,7 +66,6 @@ title = "Transcribe music from YouTube videos using Transformers."
 description = """
 Gradio demo for Music Transcription with Transformers. Read more in the links below.
 """
-
 article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2111.03017' target='_blank'>MT3: Multi-Task Multitrack Music Transcription</a> | <a href='https://github.com/magenta/mt3' target='_blank'>Github Repo</a></p>"
 
 # Create a block object
@@ -80,10 +78,9 @@ with demo:
                 + "</h1>")
     gr.Markdown(description)
     with gr.Box():
-        gr.Markdown("<h2>Select your model</h2>")
         gr.Markdown("""
-                    The ismir2021 model transcribes piano only, with note velocities. 
-                    The mt3 model transcribes multiple simultaneous instruments, but without velocities."
+                    Select your model: The ismir2021 model transcribes piano only, with note velocities. 
+                    The mt3 model transcribes multiple simultaneous instruments, but without velocities.
                     """)
         model = gr.Radio(
             ["mt3", "ismir2021"], label="What kind of model you want to use?", value="mt3"
@@ -96,7 +93,20 @@ with demo:
             img = gr.Image(label="Thumbnail")
         with gr.Row():
             yt_audio = gr.Audio()
+
         link.change(fn=populate_metadata, inputs=link, outputs=[img, title, yt_audio])
+        
+        with gr.Row():
+            btn = gr.Button("Transcribe music")
+        
+        audio_file = gr.File()
+        
+        btn.click(inference,
+                  inputs = [
+                      yt_audio
+                  ],
+                  outputs=audio_file)
+        
 
 demo.launch()