app.py

import pandas as pd
import gradio as gr

def compare_csv_files():
    file1 = "fish-speech-1.5.csv"
    file2 = "fish-speech-1.4.csv"
    
    df1 = pd.read_csv(file1)
    df2 = pd.read_csv(file2)
    
    merged_df = pd.merge(df1, df2, on="SourceText", suffixes=("_1.5", "_1.4"))
    merged_df["WordErrorRate_Diff"] = merged_df["WordErrorRate_1.5"] - merged_df["WordErrorRate_1.4"]
    if merged_df["WordErrorRate_Diff"] > 0: merged_df["WordErrorRate_Diff"] = f"1.5 is stronger than 1.4 ({merged_df['WordErrorRate_Diff']})"
    if merged_df["WordErrorRate_Diff"] < 0: merged_df["WordErrorRate_Diff"] = f"1.4 is stronger than 1.5 ({0 - merged_df['WordErrorRate_Diff']})"
    if merged_df["WordErrorRate_Diff"] == 0: merged_df["WordErrorRate_Diff"] = f"1.4 is the same as 1.5 (0)"
    if merged_df["CharacterErrorRate_Diff"] > 0: merged_df["CharacterErrorRate_Diff"] = f"1.5 is stronger than 1.4 ({merged_df['CharacterErrorRate_Diff']})"
    if merged_df["CharacterErrorRate_Diff"] < 0: merged_df["CharacterErrorRate_Diff"] = f"1.4 is stronger than 1.5 ({0 - merged_df['CharacterErrorRate_Diff']})"
    if merged_df["CharacterErrorRate_Diff"] == 0: merged_df["CharacterErrorRate_Diff"] = f"1.4 is the same as 1.5 (0)"
    
    comparison = merged_df[[
        "SourceText", 
        "WhisperText_1.5", "WordErrorRate_1.5", "CharacterErrorRate_1.5",
        "WhisperText_1.4", "WordErrorRate_1.4", "CharacterErrorRate_1.4",
        "WordErrorRate_Diff", "CharacterErrorRate_Diff"
    ]]
    
    return comparison.to_html(index=False)

gr.Interface(
    fn=compare_csv_files,
    inputs=None,
    outputs="html"
).launch()