Update app.py

app.py

@@ -1,5 +1,6 @@
 import pandas as pd
 import gradio as gr
+import os
 
 def compare_csv_files(max_num):
     df1 = pd.read_csv("fish-speech-1.5.csv")
@@ -33,20 +34,37 @@ def compare_csv_files(max_num):
     <p>Average CharacterErrorRate Difference (excluding large diffs): {f'1.5 is stronger ({avg_char_diff:.8f})' if avg_char_diff < 0 else f'1.4 is stronger ({0 - avg_char_diff:.8f})'}</p>
     """
     
+    def get_audio_files(uuid):
+        file_1_5 = os.path.join("fish-speech-1.5", f"{uuid}.wav")
+        file_1_4 = os.path.join("fish-speech-1.4", f"{uuid}.wav")
+        return file_1_5, file_1_4
+    
+    audio_files = []
+    for uuid in merged_df["SourceText"]:
+        file_1_5, file_1_4 = get_audio_files(uuid)
+        audio_files.append((file_1_5, file_1_4))
+    
     result = merged_df[[
         "SourceText",
         "WordErrorRate_1.5", "WordErrorRate_1.4", "WordErrorRate_Comparison",
         "CharacterErrorRate_1.5", "CharacterErrorRate_1.4", "CharacterErrorRate_Comparison",
         "WhisperText_1.5", "WhisperText_1.4"
     ]]
+
+    # Add audio columns to the result for Gradio interface
+    audio_columns = [
+        gr.Audio(value=file_1_5) for file_1_5, _ in audio_files
+    ] + [
+        gr.Audio(value=file_1_4) for _, file_1_4 in audio_files
+    ]
     
-    return overall_summary + result.to_html(index=False)
+    return overall_summary + result.to_html(index=False), *audio_columns
 
 max_num = gr.Number(value=10)
 gr.Interface(
     fn=compare_csv_files,
     inputs=[max_num],
-    outputs="html",
+    outputs=["html"] + [gr.Audio() for _ in range(len(df1))],  # Dynamically add audio outputs
     title="Fish Speech Benchmark",
-    description="This is a non official model performance test from Fish Speech / Whisper Base / More data will be added later (not too much)"
+    description="This is a non-official model performance test from Fish Speech / Whisper Base / More data will be added later (not too much)"
 ).launch()