app.py

# app.py
import gradio as gr
import jiwer
import pandas as pd

def calculate_metrics(hypothesis, reference):
    """
    Calculate WER, MER, WIL and CER metrics using jiwer
    
    Args:
        hypothesis (str): ASR output text
        reference (str): Ground truth reference text
        
    Returns:
        dict: Dictionary containing the metrics
    """
    # Calculate WER and related metrics
    transformation = jiwer.Compose([
        jiwer.ToLowerCase(),
        jiwer.RemoveMultipleSpaces(),
        jiwer.Strip(),
        jiwer.RemovePunctuation()
    ])
    
    wer = jiwer.wer(reference, hypothesis, truth_transform=transformation, hypothesis_transform=transformation)
    mer = jiwer.mer(reference, hypothesis, truth_transform=transformation, hypothesis_transform=transformation)
    wil = jiwer.wil(reference, hypothesis, truth_transform=transformation, hypothesis_transform=transformation)
    
    # Calculate CER
    cer = jiwer.cer(reference, hypothesis, truth_transform=transformation, hypothesis_transform=transformation)
    
    return {
        "WER (Word Error Rate)": f"{wer:.4f}",
        "MER (Match Error Rate)": f"{mer:.4f}",
        "WIL (Word Information Lost)": f"{wil:.4f}",
        "CER (Character Error Rate)": f"{cer:.4f}"
    }

def process_file_input(hypothesis_file, reference_file):
    """
    Process uploaded text files and calculate metrics
    
    Args:
        hypothesis_file (str): Path to ASR output file
        reference_file (str): Path to reference file
        
    Returns:
        dict: Dictionary containing the metrics
    """
    try:
        with open(hypothesis_file.name, 'r', encoding='utf-8') as f:
            hypothesis = f.read()
        with open(reference_file.name, 'r', encoding='utf-8') as f:
            reference = f.read()
            
        return calculate_metrics(hypothesis, reference)
    except Exception as e:
        return f"Error processing files: {str(e)}"

# Text input interface
text_interface = gr.Interface(
    fn=calculate_metrics,
    inputs=[
        gr.Textbox(label="ASR Output Text", placeholder="Enter ASR output text here...", lines=5),
        gr.Textbox(label="Reference Text", placeholder="Enter reference text here...", lines=5)
    ],
    outputs=gr.JSON(label="Metrics"),
    title="Speech Recognition Metrics Calculator",
    description="Calculate WER (Word Error Rate) and other metrics between ASR output and reference text.",
    examples=[
        ["the quick brown fox jumps over the lazy dog", 
         "the quick brown fox jumped over the lazy dog"],
        ["hello wrld how are you", 
         "hello world how are you"]
    ]
)

# File input interface
file_interface = gr.Interface(
    fn=process_file_input,
    inputs=[
        gr.File(label="ASR Output File (txt)"),
        gr.File(label="Reference File (txt)")
    ],
    outputs=gr.JSON(label="Metrics"),
    title="Speech Recognition Metrics Calculator (File Upload)",
    description="Calculate WER (Word Error Rate) and other metrics from text files."
)

# Create the demo
demo = gr.TabbedInterface(
    [text_interface, file_interface],
    tab_names=["Text Input", "File Upload"]
)

# Launch the demo
demo.queue().launch()