Update app.py

app.py

@@ -34,36 +34,30 @@ 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):
+    def get_audio_html(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_1_5 = f'<audio controls src="{file_1_5}"></audio>' if os.path.exists(file_1_5) else "Missing Audio"
+        audio_1_4 = f'<audio controls src="{file_1_4}"></audio>' if os.path.exists(file_1_4) else "Missing Audio"
+        return audio_1_5, audio_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))
+    merged_df["Audio_1.5"], merged_df["Audio_1.4"] = zip(*merged_df["SourceText"].apply(get_audio_html))
     
-    result = merged_df[[
+    result_html = overall_summary + 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"
-    ]]
-
-    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
-    ]
+        "WhisperText_1.5", "WhisperText_1.4",
+        "Audio_1.5", "Audio_1.4"
+    ]].to_html(escape=False, index=False)
     
-    return overall_summary + result.to_html(index=False), *audio_columns
+    return result_html
 
 max_num = gr.Number(value=10)
 gr.Interface(
     fn=compare_csv_files,
     inputs=[max_num],
-    outputs=["html"] + [gr.Audio() for _ in range(len(df1))],  # Dynamically add audio outputs
+    outputs="html",
     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)"
 ).launch()