display misaligned sentences only once

app.py

@@ -162,8 +162,8 @@ def format_sentence_metrics(sentence_wers, sentence_cers, average_wer, average_c
         md += "\n### Misaligned Sentences\n\n"
         for misaligned in misaligned_sentences:
             md += f"#### Sentence {misaligned['index']}\n"
-            md += f"* Reference: {misaligned['context_ref']}\n"
-            md += f"* Hypothesis: {misaligned['context_hyp']}\n"
+            md += f"* Reference: {misaligned['reference']}\n"
+            md += f"* Hypothesis: {misaligned['hypothesis']}\n"
             md += f"* Misalignment starts at position: {misaligned['misalignment_start']}\n\n"
     else:
         md += "\n### Misaligned Sentences\n\n"
@@ -171,33 +171,30 @@ def format_sentence_metrics(sentence_wers, sentence_cers, average_wer, average_c
 
     return md
 
-@spaces.GPU()
-def process_files(reference_file, hypothesis_file):
-    try:
-        with open(reference_file.name, 'r') as f:
-            reference_text = f.read()
 
-        with open(hypothesis_file.name, 'r') as f:
-            hypothesis_text = f.read()
+def process_and_display(ref_file, hyp_file):
+    result = process_files(ref_file, hyp_file)
 
-        overall_wer = calculate_wer(reference_text, hypothesis_text)
-        overall_cer = calculate_cer(reference_text, hypothesis_text)
-        sentence_metrics = calculate_sentence_metrics(reference_text, hypothesis_text)
-        misaligned = identify_misaligned_sentences(reference_text, hypothesis_text)
+    if "error" in result:
+        error_msg = result["error"]
+        return {"error": error_msg}, "", ""
 
-        return {
-            "Overall WER": overall_wer,
-            "Overall CER": overall_cer,
-            "Sentence WERs": sentence_metrics["sentence_wers"],
-            "Sentence CERs": sentence_metrics["sentence_cers"],
-            "Average WER": sentence_metrics["average_wer"],
-            "Average CER": sentence_metrics["average_cer"],
-            "Standard Deviation WER": sentence_metrics["std_dev_wer"],
-            "Standard Deviation CER": sentence_metrics["std_dev_cer"],
-            "Misaligned Sentences": misaligned
-        }
-    except Exception as e:
-        return {"error": str(e)}
+    metrics = {
+        "Overall WER": result["Overall WER"],
+        "Overall CER": result["Overall CER"]
+    }
+
+    metrics_md = format_sentence_metrics(
+        result["Sentence WERs"],
+        result["Sentence CERs"],
+        result["Average WER"],
+        result["Average CER"],
+        result["Standard Deviation WER"],
+        result["Standard Deviation CER"],
+        result["Misaligned Sentences"]
+    )
+
+    return metrics, metrics_md
 
 def process_and_display(ref_file, hyp_file):
     result = process_files(ref_file, hyp_file)
@@ -233,7 +230,6 @@ def process_and_display(ref_file, hyp_file):
 
     return metrics, metrics_md, misaligned_md
 
-
 def main():
     with gr.Blocks() as demo:
         gr.Markdown("# ASR Metrics")
@@ -249,8 +245,7 @@ def main():
         with gr.Row():
             compute_button = gr.Button("Compute Metrics")
             results_output = gr.JSON(label="Results")
-            metrics_output = gr.Markdown(label="Sentence Metrics")
-            misaligned_output = gr.Markdown(label="Misaligned Sentences")
+            metrics_output = gr.Markdown(label="Metrics")
 
         # Update previews when files are uploaded
         def update_previews(ref_file, hyp_file):
@@ -259,10 +254,10 @@ def main():
 
             if ref_file:
                 with open(ref_file.name, 'r') as f:
-                    ref_text = f.read()[:200]
+                    ref_text = f.read()[:200]  # Show first 200 characters
             if hyp_file:
                 with open(hyp_file.name, 'r') as f:
-                    hyp_text = f.read()[:200]
+                    hyp_text = f.read()[:200]  # Show first 200 characters
 
             return ref_text, hyp_text
 
@@ -280,10 +275,11 @@ def main():
         compute_button.click(
             fn=process_and_display,
             inputs=[reference_file, hypothesis_file],
-            outputs=[results_output, metrics_output, misaligned_output]
+            outputs=[results_output, metrics_output]
         )
 
     demo.launch()
 
+
 if __name__ == "__main__":
     main()