fix(normalizer): expand wh-word 're contractions espeak voices as -ray

api/src/services/text_processing/normalizer.py

@@ -410,6 +410,17 @@ def handle_time(t: re.Match[str]) -> str:
 def normalize_text(text: str, normalization_options: NormalizationOptions) -> str:
     """Normalize text for TTS processing"""
 
+    # Expand the "'re" contractions that espeak mis-phonemizes with a spurious
+    # /ɹeɪ/ ("-ray") ending, e.g. "how're" -> /haʊɹeɪ/ which sounds like "harry".
+    # Only the wh-words and there/these/those break this way; "you're", "we're"
+    # and "they're" already phonemize correctly and are left untouched.
+    text = re.sub(
+        r"\b(how|what|where|who|when|why|there|these|those)['’]re\b",
+        r"\1 are",
+        text,
+        flags=re.IGNORECASE,
+    )
+
     # Handle email addresses first if enabled
     if normalization_options.email_normalization:
         text = EMAIL_PATTERN.sub(handle_email, text)

api/tests/test_normalizer.py

@@ -334,3 +334,30 @@ def test_remaining_symbol():
         )
         == "I love buying products at good store here and at other store"
     )
+
+
+def test_re_contraction_expansion():
+    """Wh-word "'re" contractions are expanded so espeak does not voice a
+    spurious "-ray" ending (e.g. "how're" was phonemized like "harry")."""
+    assert (
+        normalize_text(
+            "Hello there, how're you doing this fine day?",
+            normalization_options=NormalizationOptions(),
+        )
+        == "Hello there, how are you doing this fine day?"
+    )
+    assert (
+        normalize_text(
+            "What're these and where're they going?",
+            normalization_options=NormalizationOptions(),
+        )
+        == "What are these and where are they going?"
+    )
+    # Contractions that already phonemize correctly must be left untouched.
+    assert (
+        normalize_text(
+            "You're sure we're not late and they're here?",
+            normalization_options=NormalizationOptions(),
+        )
+        == "You're sure we're not late and they're here?"
+    )