Normalize German TTS abbreviations and units

platform/docker/tts-gateway/test_tts_gateway.py

@@ -122,11 +122,15 @@ class LanguageSegmentationTests(unittest.TestCase):
 
     def test_qwen_speaks_temperature_and_time_ranges_with_bis(self):
         spoken = gateway.prepare_for_qwen_speech(
-            "15°–23 °C, 93 %, ca. 12 mm, vor allem um 05–06 Uhr."
+            "15°–23 °C, 93 %, ca. 12 mm, v.a. um 05–06 Uhr, "
+            "Wind max. 16 km/h."
         )
         self.assertIn("15 bis 23 Grad", spoken)
         self.assertIn("93 Prozent", spoken)
+        self.assertIn("circa 12 mm", spoken)
+        self.assertIn("vor allem", spoken)
         self.assertIn("5 bis 6 Uhr", spoken)
+        self.assertIn("Wind maximal 16 Kilometer pro Stunde", spoken)
 
     def test_qwen_keeps_prosody_punctuation(self):
         spoken = gateway.prepare_for_qwen_speech(

platform/docker/tts-gateway/tts_gateway.py

@@ -235,6 +235,9 @@ def _spoken_ipv4(match: re.Match) -> str:
 
 def normalize_for_german_speech(text: str) -> str:
     """Turn common visual notation into unambiguous spoken German."""
+    text = re.sub(r"\bv\.\s*a\.", "vor allem", text, flags=re.IGNORECASE)
+    text = re.sub(r"\bca\.", "circa", text, flags=re.IGNORECASE)
+    text = re.sub(r"\bmax\.", "maximal", text, flags=re.IGNORECASE)
     # Run these before the date rule: otherwise 192.168.1.5 could be partly
     # interpreted as a visual date.
     text = re.sub(
@@ -329,10 +332,13 @@ def prepare_for_qwen_speech(text: str) -> str:
     text = re.sub(r"\[([^]]+)\]\([^)]+\)", r"\1", text)
     text = re.sub(r"(?m)^\s{0,3}#{1,6}\s*", "", text)
     text = re.sub(r"(?m)^\s*[-*+]\s+", "", text)
-    text = text.replace("_", " ").replace("/", ", ")
+    text = text.replace("_", " ")
     text = text.replace("→", ". ").replace("←", ". ")
     if DEFAULT_LANGUAGE == "de":
         text = normalize_for_german_speech(text)
+    # Preserve slashes until after unit normalization so km/h becomes
+    # "Kilometer pro Stunde" instead of the broken "Kilometer, h".
+    text = text.replace("/", ", ")
     text = "".join(
         char for char in text
         if unicodedata.category(char) not in {"So", "Cs"}