fix(tts): don't let a self-closing prosody marker swallow the next expr tag

livekit-agents/livekit/agents/tts/_provider_format.py

@@ -634,9 +634,12 @@ def sentence_tokenizer(provider: str, *, expressive: bool) -> tokenize.SentenceT
 _EXPR_CLOSE_RE = re.compile(r"</expr\s*>")
 # self-closing markers only (the trailing / is required)
 _EXPR_SELF_RE = re.compile(r"<expr\b([^>]*?)/\s*>")
-# a wrapping marker (prosody/spell) and its span; non-greedy, instructed not to nest
+# a wrapping marker (prosody/spell) and its span; non-greedy, instructed not to nest.
+# the opening tag must not itself be self-closing, or a self-closing prosody point
+# control ahead of a later wrapping marker gets mistaken for that marker's opener,
+# swallowing everything up to the next </expr> into its span.
 _EXPR_WRAP_RE = re.compile(
-    r'<expr\b(?=[^>]*type="(?:prosody|spell)")([^>]*?)>(.*?)</expr\s*>', re.DOTALL
+    r'<expr\b(?=[^>]*type="(?:prosody|spell)")([^>]*?)(?<!/)>(.*?)</expr\s*>', re.DOTALL
 )
 # a non-wrapping type the LLM forgot to self-close (normalize_markup fixes these)
 _EXPR_UNCLOSED_RE = re.compile(

tests/test_expr_markup.py

@@ -142,6 +142,18 @@ def test_convert_expr_cartesia_prosody_unknown_label_unwraps() -> None:
     assert convert_markup("cartesia", text) == "keep it secret"
 
 
+def test_convert_expr_cartesia_self_closing_prosody_before_spell() -> None:
+    # a self-closing point control ahead of a later wrapping marker must not be
+    # mistaken for that marker's opening tag — each converts independently
+    text = (
+        'Say it slow: <expr type="prosody" label="slow"/> '
+        'and now spell it: <expr type="spell">A7X9</expr> done.'
+    )
+    assert convert_markup("cartesia", text) == (
+        'Say it slow: <speed ratio="0.85"/> and now spell it: <spell>A7X9</spell> done.'
+    )
+
+
 def test_convert_stray_expr_never_reaches_tts() -> None:
     # an unpaired prosody open/close (e.g. split across stream chunks) is dropped,
     # keeping the words