diff --git a/cli.py b/cli.py
index 3ed1b73a28ce..899b1f5f6561 100644
--- a/cli.py
+++ b/cli.py
@@ -11913,13 +11913,14 @@ class HermesCLI(CLIAgentSetupMixin, CLICommandsMixin, CLIBillingMixin):
                 pass
 
             # Track consecutive no-speech cycles to avoid infinite restart loops.
+            stop_continuous_restart = False
             if not submitted:
                 self._no_speech_count = getattr(self, '_no_speech_count', 0) + 1
                 if self._no_speech_count >= 3:
                     self._voice_continuous = False
                     self._no_speech_count = 0
                     _cprint(f"{_DIM}No speech detected 3 times, continuous mode stopped.{_RST}")
-                    return
+                    stop_continuous_restart = True
             else:
                 self._no_speech_count = 0
 
@@ -11927,7 +11928,12 @@ class HermesCLI(CLIAgentSetupMixin, CLICommandsMixin, CLIBillingMixin):
             # restart recording so the user can keep talking.
             # (When transcript IS submitted, process_loop handles restart
             # after chat() completes.)
-            if self._voice_continuous and not submitted and not self._voice_recording:
+            if (
+                self._voice_continuous
+                and not submitted
+                and not self._voice_recording
+                and not stop_continuous_restart
+            ):
                 self._voice_restart_recording_async()
 
     def _voice_speak_response_async(self, text: str) -> None: