Update app.py

app.py

@@ -7,6 +7,15 @@ import random
 labels = ["Real Audio 🗣️", "Cloned Audio 🤖"]
 DURATION = 2
 
+MD_DESCRIPTION = """
+# Clone Guesser
+
+Try to guess if the voice is real 🗣️ or cloned 🤖. 
+
+Each cloned voice is taken from a real voice, so the text itself is not a hint!
+
+For more information about the data, see the [dataset page](https://huggingface.co/datasets/jerpint/vox-cloned-data)
+"""
 
 def get_accuracy(score_matrix) -> str:
 
@@ -29,8 +38,10 @@ def confusion_matrix_to_markdown(matrix, labels=None):
     labels = labels or [f"Class {i}" for i in range(num_labels)]
     accuracy = get_accuracy(matrix)
 
+    markdown = "Scoreboard \n"
+
     # Header row
-    markdown = f"| {' | '.join([''] + labels)} |\n"
+    markdown += f"| {' | '.join([''] + labels)} |\n"
     markdown += f"| {' | '.join(['---'] * (num_labels + 1))} |\n"
 
     # Data rows
@@ -104,6 +115,8 @@ with gr.Blocks() as demo:
     current_audio = gr.State(get_random_audio)
     score_matrix = gr.State([[0, 0], [0, 0]])
 
+    gr.Markdown(MD_DESCRIPTION)
+
     with gr.Column():
         with gr.Row():
             audio_cmp = gr.Audio(