fix: disable STT retries and stop continuous mode after 3 silent cycles

- Set max_retries=0 on the STT OpenAI client. The SDK default (2) honors
  Groq's retry-after header (often 53s), blocking the thread for up to
  ~106s on rate limits. Voice STT should fail fast, not retry silently.
- Stop continuous recording mode after 3 consecutive no-speech cycles to
  prevent infinite restart loops when nobody is talking.

cli.py

@@ -3688,6 +3688,17 @@ class HermesCLI:
             except Exception:
                 pass
 
+            # Track consecutive no-speech cycles to avoid infinite restart loops.
+            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
+            else:
+                self._no_speech_count = 0
+
             # If no transcript was submitted but continuous mode is active,
             # restart recording so the user can keep talking.
             # (When transcript IS submitted, process_loop handles restart

tools/transcription_tools.py

@@ -150,7 +150,7 @@ def transcribe_audio(file_path: str, model: Optional[str] = None) -> Dict[str, A
     try:
         from openai import OpenAI, APIError, APIConnectionError, APITimeoutError
 
-        client = OpenAI(api_key=api_key, base_url=base_url, timeout=30)
+        client = OpenAI(api_key=api_key, base_url=base_url, timeout=30, max_retries=0)
 
         with open(file_path, "rb") as audio_file:
             transcription = client.audio.transcriptions.create(