Update app.py

app.py

@@ -12,13 +12,20 @@ from transformers import AutoTokenizer, AutoFeatureExtractor
 from streamer import ParlerTTSStreamer                     # local file
 
 from src.detection.factory import get_detector
-from src.alerting.alert_system import get_alerter
+from src.alerting.alert_system import FileAlertSystem
 
 # ──────────────────────────────────────────────────────────
 # CONFIG & BACKEND SET-UP
 # ──────────────────────────────────────────────────────────
 load_dotenv()
 
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s %(levelname)s │ %(message)s",
+    datefmt="%H:%M:%S",
+)
+
+
 with open("config.yaml", "r") as f:
     config = yaml.safe_load(f)
 
@@ -26,57 +33,13 @@ secrets = {"gemini_api_key": os.getenv("GEMINI_API_KEY")}
 
 print("Initializing detector and alerter …")
 detector = get_detector(config)
-alerter = get_alerter(config, secrets["gemini_api_key"])
-print("Backend ready.")
-
-# ──────────────────────────────────────────────────────────
-# TTS MODEL (Parler-TTS mini)
-# ──────────────────────────────────────────────────────────
-device = "cuda:0" if torch.cuda.is_available() else "cpu"
-if device == "cpu":
-    print("\n⚠️  Running TTS on CPU will be slow; only ‘Very Drowsy’ alerts will use it.\n")
-
-model_dtype = torch.float16 if device != "cpu" else torch.float32
-repo_id = "parler-tts/parler_tts_mini_v0.1"
+alerter  = FileAlertSystem(CONFIG)  
 
-print("Loading Parler-TTS …")
-model = ParlerTTSForConditionalGeneration.from_pretrained(repo_id,
-                                                         torch_dtype=model_dtype).to(device)
-tokenizer = AutoTokenizer.from_pretrained(repo_id)
-feature_extractor = AutoFeatureExtractor.from_pretrained(repo_id)
-print("TTS loaded.")
+if alerter.audio_bytes is None:
+    logging.warning("No alert sound loaded; driver will not hear any audio!")
 
-# ──────────────────────────────────────────────────────────
-# AUDIO STREAMER
-# ──────────────────────────────────────────────────────────
-def stream_alert_audio(text_prompt: str):
-    """Yields (sampling_rate, np.ndarray) chunks for Gradio streaming."""
-    sampling_rate = model.config.sampling_rate
-    voice_desc = "Jenny is a female speaker with a clear and urgent voice."
-
-    prompt_ids = tokenizer(text_prompt, return_tensors="pt").input_ids.to(device)
-    desc_ids   = tokenizer(voice_desc, return_tensors="pt").input_ids.to(device)
-
-    streamer = ParlerTTSStreamer(model, device, play_steps=int(sampling_rate * 2.0))
-
-    gen_kwargs = dict(
-        input_ids=desc_ids,
-        prompt_input_ids=prompt_ids,
-        streamer=streamer,
-        do_sample=True,
-        temperature=1.0,
-        repetition_penalty=1.2,
-    )
-
-    thread = Thread(target=model.generate, kwargs=gen_kwargs, daemon=True)
+print("Backend ready.")
 
-    try:
-        thread.start()
-        for chunk in streamer:
-            yield (sampling_rate, chunk)
-    finally:
-        thread.join(timeout=0.1)
-        alerter.reset_alert()
 
 # ──────────────────────────────────────────────────────────
 # FRAME PROCESSOR
@@ -85,26 +48,28 @@ def process_live_frame(frame):
     if frame is None:
         return np.zeros((480, 640, 3), np.uint8), "Status: Inactive", None
 
+    t0 = time.time()
+    
     processed, indicators, _ = detector.process_frame(frame)
     level      = indicators.get("drowsiness_level", "Awake")
     lighting   = indicators.get("lighting", "Good")
     score      = indicators.get("details", {}).get("Score", 0)
 
+    dt_ms = (time.time() - t0) * 1000.0
+    logging.info(f"{dt_ms:6.1f} ms │ {lighting:<4} │ {level:<14} │ score={score:.2f}")
+
     status_txt = f"Lighting: {lighting}\n"
     status_txt += ("Detection paused due to low light."
                    if lighting == "Low"
                    else f"Status: {level}\nScore: {score:.2f}")
 
+    audio_out = None
     audio_out = None
     if level != "Awake" and lighting != "Low":
-        payload = alerter.trigger_alert(level=level)
-        if payload:
-            # Static file path → bytes, Dynamic Gemini path → str
-            if isinstance(payload, bytes):
-                # Return raw bytes (Gradio accepts bytes for .wav / .mp3)
-                audio_out = payload
-            elif isinstance(payload, str):
-                audio_out = stream_alert_audio(payload)
+        audio_bytes = alerter.trigger_alert(level=level)
+        logging.info(f"Printing {audio_bytes}")
+        if audio_bytes:
+            audio_out = audio_bytes
 
     return processed, status_txt, audio_out
 
@@ -133,4 +98,5 @@ with gr.Blocks(theme=gr.themes.Default(primary_hue="blue")) as app:
     )
 
 if __name__ == "__main__":
+    logging.info("Launching Gradio app …")
     app.launch(debug=True)