requirements.txt

# Core dependencies
gradio>=5.35.0
transformers>=4.35.0
torch>=2.0.0
torchaudio>=2.0.0

# Audio processing
librosa>=0.10.0
soundfile>=0.12.0
scipy>=1.9.0

# Performance optimization
accelerate>=0.24.0
numpy>=1.21.0

# Model management
huggingface-hub>=0.17.0
safetensors>=0.4.0

# Additional utilities
datasets>=2.14.0

# Note: Flash Attention 2 is optional and requires manual installation
# If you want Flash Attention 2 for GPU acceleration, install it separately:
# pip install flash-attn --no-build-isolation
# This is commented out to avoid build issues on HuggingFace Spaces

# Optional: Better audio processing (uncomment if needed)
ffmpeg-python>=0.2.0
webrtcvad>=2.0.10
noisereduce>=3.0.0