requirements_voice_gemma.txt

# Requirements for DOLPHIN PDF AI - Local Gemma 3n Version with Voice Chat
# Optimized for powerful GPU deployment with local models and voice functionality

# Core PyTorch dependencies (updated versions from voice_chat)
torch==2.3.1
torchvision==0.18.1
torchaudio==2.3.1

# Transformers and HuggingFace ecosystem
transformers>=4.53.0
huggingface-hub>=0.20.0
tokenizers>=0.21.0
accelerate>=0.20.0
timm>=1.0.16
bitsandbytes>=0.41.0
datasets>=2.14.0
safetensors>=0.3.0

# Gradio (using beta version from voice_chat for compatibility)
https://gradio-builds.s3.amazonaws.com/cffe9a7ab7f71e76d7214dc57c6278ffaf5bcdf9/gradio-5.0.0b1-py3-none-any.whl

# Image processing
Pillow>=10.0.0
opencv-python>=4.8.0
numpy>=1.24.0

# PDF processing
PyMuPDF>=1.23.0

# RAG and embeddings
sentence-transformers>=2.2.0
scikit-learn>=1.3.0
faiss-cpu>=1.7.4

# Text processing and markdown
markdown>=3.5.0
python-markdown-math>=0.8

# Voice processing dependencies
librosa==0.10.2.post1
soundfile==0.12.1
pydub==0.25.1
onnxruntime==1.19.0
gtts>=2.3.0

# Voice chat specific dependencies
litgpt==0.4.3
snac==1.2.0
openai-whisper

# Web framework dependencies
fastapi==0.112.4
flask==3.0.3
streamlit==1.37.1

# Utilities
tqdm>=4.66.0
requests>=2.31.0
packaging>=23.0
pyyaml>=6.0
fire

# Optional but recommended for better performance
# flash-attn>=2.0.0; platform_system != "Darwin"
# xformers>=0.0.22; platform_system != "Darwin"
# PyAudio==0.2.14  # Commented out as it can cause installation issues

# Development and debugging
ipython>=8.0.0
jupyter>=1.0.0