Update app.py

app.py

@@ -1,9 +1,9 @@
 import gradio as gr
 import requests
 import json
+from faster_whisper import WhisperModel
 
 OLLAMA_URL = "http://localhost:11434"
-HF_API_URL = "https://api-inference.huggingface.co/models/openai/whisper-large-v3-turbo"
 
 MODELS = {
     "Qwen2.5-Coder 1.5B (Fastest)": "qwen2.5-coder:1.5b",
@@ -11,6 +11,11 @@ MODELS = {
     "Qwen2.5-Coder 7B (Quality)": "qwen2.5-coder:7b",
 }
 
+# Load Whisper locally (tiny model = ~75MB RAM)
+print("Loading Whisper...")
+whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8")
+print("Whisper ready!")
+
 def check_ollama():
     try:
         r = requests.get(f"{OLLAMA_URL}/api/tags", timeout=5)
@@ -23,16 +28,9 @@ def transcribe_audio(audio):
         return ""
     
     try:
-        with open(audio, "rb") as f:
-            data = f.read()
-        
-        response = requests.post(HF_API_URL, data=data, timeout=60)
-        
-        if response.status_code == 200:
-            result = response.json()
-            return result.get("text", "")
-        else:
-            return f"[STT Error: {response.status_code}]"
+        segments, _ = whisper_model.transcribe(audio)
+        text = " ".join([seg.text for seg in segments])
+        return text.strip()
     except Exception as e:
         return f"[STT Error: {e}]"