improve diarized transcript format

app.py

@@ -57,23 +57,36 @@ sense_models = {}
 dar_pipe = None
 
 converter = opencc.OpenCC('s2t')
- # —————— Helpers ——————
+
+# —————— Diarization Formatter ——————
 def format_diarization_html(snippets):
     palette = ["#e74c3c", "#3498db", "#27ae60", "#e67e22", "#9b59b6", "#16a085", "#f1c40f"]
     speaker_colors = {}
-    html = ["<div style='font-family:monospace; line-height:1.5em;'>"]
+    html_lines = []
+    last_spk = None
     for s in snippets:
         if s.startswith("[") and "]" in s:
             spk, txt = s[1:].split("]", 1)
             spk, txt = spk.strip(), txt.strip()
         else:
-            spk, txt = "", s
+            spk, txt = "", s.strip()
+        # hide empty lines
+        if not txt:
+            continue
+        # assign color if new speaker
         if spk not in speaker_colors:
             speaker_colors[spk] = palette[len(speaker_colors) % len(palette)]
         color = speaker_colors[spk]
-        html.append(f"<p style='margin:4px 0; color:{color};'><strong>{spk}:</strong> {txt}</p>")
-    html.append("</div>")
-    return "".join(html)
+        # simplify tag for same speaker
+        if spk == last_spk:
+            display = txt
+        else:
+            display = f"<strong>{spk}:</strong> {txt}"
+        last_spk = spk
+        html_lines.append(
+            f"<p style='margin:4px 0; font-family:monospace; color:{color};'>{display}</p>"
+        )
+    return "<div>" + "".join(html_lines) + "</div>"
 
 # —————— Helpers ——————
 def get_whisper_pipe(model_id: str, device: int):