DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.
Python ai-agentsai-tutorclawbotcli-tooldeepresearchinteractive-learninglearningmulti-agent-systemsrag
Merge dev into PR #1557 and preserve integrated behavior
Bingxi Zhao (Frank)b83592a1072e
diff --cc deeptutor/services/voice/base.pyindex dce58d63f,ba3590f93..b14c89657--- a/deeptutor/services/voice/base.py+++ b/deeptutor/services/voice/base.py@@@ -156,28 -152,7 +156,26 @@@ _WHITESPACE = re.compile(r"[ \t]+"_BLANK_LINES = re.compile(r"\n{3,}")+def _unwrap_emphasis_for_speech(text: str) -> str:+ """Turn Markdown bold/italic/strike into plain words.++ Voice models otherwise speak ``*`` as "asterisk". Math must already have+ been verbalized so TeX ``*`` / ``_`` inside ``$…$`` is not treated as+ emphasis. Leftover unmatched ``**`` markers are dropped; a remaining+ single ``*`` is turned into a space so "asterisk" is never read.+ """+ out = _BOLD_STARS.sub(r"\1", text)+ out = _BOLD_UNDERS.sub(r"\1", out)+ out = _STRIKE.sub(r"\1", out)+ out = _ITALIC_STARS.sub(r"\1", out)+ out = _ITALIC_UNDERS.sub(r"\1", out)+ out = out.replace("**", "")+ out = out.replace("__", "")+ out = out.replace("*", " ")+ return out++- def strip_markdown_for_speech(- text: str, *, max_chars: int = 0, math_speak: bool = True- ) -> str:+ def strip_markdown_for_speech(text: str, *, max_chars: int = 0, math_speak: bool = True) -> str:"""Reduce Markdown to plain prose suitable for TTS.Drops code blocks and tables outright (they read terribly), unwraps links