Update requirements.txt

requirements.txt

@@ -1,4 +1,20 @@
-gradio
-onnxruntime_genai
-huggingface_hub
-gtts
+# Core Gradio UI library
+gradio>=4.0.0,<5.0.0
+
+# ONNX Runtime GenAI for inference (CPU version)
+# Use --pre as it's often in pre-release
+# NOTE: If targeting a GPU Space, change this to onnxruntime-genai-cuda
+# and update EXECUTION_PROVIDER in app.py to "cuda"
+onnxruntime-genai --pre
+
+# Hugging Face Hub for downloading models/files
+huggingface_hub>=0.20.0
+
+# ONNX Runtime itself (will be installed as a dependency of onnxruntime-genai,
+# but specifying can sometimes help resolve version conflicts if needed)
+# onnxruntime>=1.17.0 # Generally not needed to list explicitly
+
+# Git LFS is needed by huggingface_hub to download large model files.
+# It needs to be installed on the Space environment, which is usually handled
+# by the Hugging Face Spaces infrastructure if not using Docker.
+# If you encounter download issues, ensure git-lfs is available.