SongFormer / requirements.txt
Ben Osheroff
handler.py for running SongFormer on HF inference endpoint
e31009e
raw
history blame
632 Bytes
# Requirements for HuggingFace Inference Endpoint
# This file contains dependencies needed for the handler.py to work
# Core ML frameworks
transformers>=4.30.0
torch>=2.0.0
# Audio processing
librosa>=0.10.0
soundfile>=0.12.0
audioread>=3.0.0
# Numerical computing
numpy>=1.24.0
scipy>=1.10.0
# Additional dependencies for SongFormer model
# (these may already be installed by the model itself)
einops>=0.7.0
x-transformers>=1.0.0
ema-pytorch>=0.2.0
loguru>=0.7.0
omegaconf>=2.3.0
muq
msaf
# Note: For MP3 support, ffmpeg must be installed in the system
# Add to Dockerfile:
# RUN apt-get update && apt-get install -y ffmpeg