From ba16e9ca27a0af1eebd13a87bf877ae496664e15 Mon Sep 17 00:00:00 2001
From: Adolfo Reyna <aeroreyna@gmail.com>
Date: Fri, 7 Aug 2026 20:04:34 -0400
Subject: [PATCH] Fix speech text truncation bug where Assistant: prefix caused
 multi-line responses to drop remaining sentences

---
 apple_llm.py    | 6 ++++--
 ollama_llm.py   | 6 ++++--
 opencode_llm.py | 6 ++++--
 3 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/apple_llm.py b/apple_llm.py
index 7bba0aa..a790d96 100644
--- a/apple_llm.py
+++ b/apple_llm.py
@@ -92,8 +92,10 @@ def _clean_spoken_text(text: str) -> str:
     """Clean text for speech output and truncate fake turn generations."""
     if not text:
         return ""
-    # Truncate if model hallucinates fake turn markers
-    for marker in ("User:", "Human:", "Assistant:", "\nUser", "\nHuman", "\nAssistant"):
+    # Strip leading or inline role headers (e.g. "Assistant:")
+    text = re.sub(r"(?i)\b(Assistant|assistant|Bot|bot):\s*", "", text)
+    # Truncate if model hallucinates fake subsequent user turns
+    for marker in ("\nUser:", "\nHuman:", "\nUser", "\nHuman"):
         if marker in text:
             text = text.split(marker)[0]
     # Remove markdown code blocks
diff --git a/ollama_llm.py b/ollama_llm.py
index e33f758..7ee220f 100644
--- a/ollama_llm.py
+++ b/ollama_llm.py
@@ -42,8 +42,10 @@ def _clean_spoken_text(text: str) -> str:
     """Clean text for speech output and truncate fake turn generations."""
     if not text:
         return ""
-    # Truncate if model hallucinates fake turn markers
-    for marker in ("User:", "Human:", "Assistant:", "\nUser", "\nHuman", "\nAssistant"):
+    # Strip leading or inline role headers (e.g. "Assistant:")
+    text = re.sub(r"(?i)\b(Assistant|assistant|Bot|bot):\s*", "", text)
+    # Truncate if model hallucinates fake subsequent user turns
+    for marker in ("\nUser:", "\nHuman:", "\nUser", "\nHuman"):
         if marker in text:
             text = text.split(marker)[0]
     # Remove markdown code blocks
diff --git a/opencode_llm.py b/opencode_llm.py
index 70fd1f2..872f680 100644
--- a/opencode_llm.py
+++ b/opencode_llm.py
@@ -46,8 +46,10 @@ def _clean_spoken_text(text: str) -> str:
     """Clean text for speech output and truncate fake turn generations."""
     if not text:
         return ""
-    # Truncate if model hallucinates fake turn markers
-    for marker in ("User:", "Human:", "Assistant:", "\nUser", "\nHuman", "\nAssistant"):
+    # Strip leading or inline role headers (e.g. "Assistant:")
+    text = re.sub(r"(?i)\b(Assistant|assistant|Bot|bot):\s*", "", text)
+    # Truncate if model hallucinates fake subsequent user turns
+    for marker in ("\nUser:", "\nHuman:", "\nUser", "\nHuman"):
         if marker in text:
             text = text.split(marker)[0]
     # Remove markdown code blocks