Update predictions.py

predictions.py

@@ -23,8 +23,19 @@ def get_predictions(uploaded_image):
 
     # Summarize predictions
     text = summarize_predictions_natural_language(pipeline_output)
+    corrected_text = correct_text(text)
 
-    # Generate audio from text
-    narrated_text = tts_pipe(text)
+      # Generate audio from text
+    narrated_text = tts_pipe(corrected_text)
+    audio_data = narrated_text["audio"][0]
+    sample_rate = narrated_text["sampling_rate"]
 
-    return processed_image, text, narrated_text["audio"][0]
+    return processed_image,corrected_text, (sample_rate, audio_data)
+
+
+def correct_text(text):
+    # Rule-based correction
+    # Example: "there are one horse" -> "there is one horse"
+    if "there are one" in text:
+        text = text.replace("there are one", "there is one")
+    return text