From 6b2f5555a4a0a1e8c4ab6a262da6e788588773d0 Mon Sep 17 00:00:00 2001
From: James Ketrenos <james_git@ketrenos.com>
Date: Tue, 16 Sep 2025 14:17:20 -0700
Subject: [PATCH] OV cache location change

---
 voicebot/bots/minimal.py | 12 +++++++++++-
 voicebot/bots/whisper.py |  8 +++++---
 2 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/voicebot/bots/minimal.py b/voicebot/bots/minimal.py
index 6b29506..68d9070 100644
--- a/voicebot/bots/minimal.py
+++ b/voicebot/bots/minimal.py
@@ -212,7 +212,7 @@ def create_minimal_bot_tracks(session_name: str, config: Optional[Dict[str, Any]
             - sample_rate: audio sample rate (default 48000)
             - frequency: tone frequency in Hz (default 440)
             - volume: audio volume 0-1 (default 0.5)
-            - static_color: RGB tuple for static mode (default gray)
+            - static_color: RGB color tuple or string (default (128, 128, 128))
 
     Returns:
         Dictionary containing 'video' and 'audio' tracks
@@ -235,6 +235,16 @@ def create_minimal_bot_tracks(session_name: str, config: Optional[Dict[str, Any]
     }
     default_config.update(config)  # type: ignore
 
+    # Parse static_color if it's a string
+    if isinstance(default_config.get('static_color'), str):  # type: ignore
+        try:
+            color_str = default_config['static_color']  # type: ignore
+            r, g, b = map(int, color_str.split(','))  # type: ignore
+            default_config['static_color'] = (r, g, b)
+        except (ValueError, TypeError):
+            logger.warning(f"Invalid static_color format: {default_config.get('static_color')}, using default")  # type: ignore
+            default_config['static_color'] = (128, 128, 128)
+
     media_clock = MediaClock()
 
     video_track = ConfigurableVideoTrack(media_clock, default_config)  # type: ignore
diff --git a/voicebot/bots/whisper.py b/voicebot/bots/whisper.py
index 452adf1..3d69e62 100644
--- a/voicebot/bots/whisper.py
+++ b/voicebot/bots/whisper.py
@@ -156,7 +156,7 @@ class OpenVINOConfig(BaseModel):
 
     device: str = Field(default=_device, description="Target device for inference")
     cache_dir: str = Field(
-        default="./ov_cache", description="Cache directory for compiled models"
+        default="/root/.cache", description="Cache directory for compiled models"
     )
     enable_quantization: bool = Field(
         default=True, description="Enable INT8 quantization"
@@ -510,8 +510,10 @@ class OpenVINOWhisperModel:
         self.model_id = model_id
         self.config = config
         self.device = device
-        self.model_path = Path(model_id.replace("/", "_"))
-        self.quantized_model_path = Path(f"{self.model_path}_quantized")
+        # Ensure cache directory exists
+        Path(self.config.cache_dir).mkdir(parents=True, exist_ok=True)
+        self.model_path = Path(self.config.cache_dir) / model_id.replace("/", "_")
+        self.quantized_model_path = self.model_path / "quantized"
 
         self.processor: Optional[WhisperProcessor] = None
         self.ov_model: Optional[OVModelForSpeechSeq2Seq] = None