Update app.py

app.py

@@ -1,23 +1,18 @@
 import gradio as gr
 from PIL import Image
-import torch
-from transformers import TrOCRProcessor, VisionEncoderDecoderModel
 from gtts import gTTS
+import pytesseract
 import tempfile
-import os
-
-# Load TrOCR model (OCR)
-processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-stage1")
-ocr_model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-stage1")
 
 def bangla_reader(image):
     if image is None:
         return "কোনো ছবি দেওয়া হয়নি।", None
 
-    # Step 1: OCR
-    pixel_values = processor(images=image, return_tensors="pt").pixel_values
-    generated_ids = ocr_model.generate(pixel_values)
-    ocr_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    # Step 1: OCR using Tesseract (Bangla)
+    ocr_text = pytesseract.image_to_string(image, lang='ben')
+
+    if not ocr_text.strip():
+        return "কোনো লেখা সনাক্ত করা যায়নি।", None
 
     # Step 2: gTTS Bangla
     try:
@@ -28,7 +23,7 @@ def bangla_reader(image):
     except Exception as e:
         return f"OCR সফল হয়েছে, কিন্তু TTS ব্যর্থ: {str(e)}", None
 
-    return f"OCR ফলাফল:\n{ocr_text}", audio_path
+    return f"OCR ফলাফল:\n{ocr_text.strip()}", audio_path
 
 # Gradio UI
 demo = gr.Interface(