requirements.txt

# PyTorch companion packages matching ZeroGPU torch 2.8.0
torchvision==0.23.0
torchaudio==2.8.0

# ------------------------------------------------------------
# ControlFoley core
# ------------------------------------------------------------
antlr4-python3-runtime==4.9.3
av==14.0.1
braceexpand==0.1.7
colorlog==6.9.0
dora_search==0.1.12
einops==0.8.0
flashy==0.0.2
ftfy==6.3.1
h5py==3.15.1
julius==0.2.7
omegaconf==2.3.0
open_clip_torch==2.29.0
submitit==1.5.4
treetable==0.2.6
webdataset==1.0.2
wget==3.2

# ------------------------------------------------------------
# Audio
# ------------------------------------------------------------
audioread==3.1.0
laion_clap==1.1.6
librosa==0.11.0
nnAudio==0.3.3
soundfile==0.13.1
torchdiffeq==0.2.5
torchlibrosa==0.1.0

# laion_clap 1.1.6 requires this exact NumPy version
numpy==1.23.5

# ------------------------------------------------------------
# Audiocraft dependencies bundled inside ControlFoley
# ------------------------------------------------------------
spacy==3.8.7
spacy-legacy==3.0.12
spacy-loggers==1.0.5
num2words==0.5.14

# AudioCraft / NLP helpers
sentencepiece==0.2.1
regex
PyYAML

# ------------------------------------------------------------
# Transformers stack used by ControlFoley / MusicGen
# ------------------------------------------------------------
transformers==4.38.2
tokenizers==0.15.2
safetensors==0.7.0

# ------------------------------------------------------------
# Scientific stack
# ------------------------------------------------------------
scipy==1.15.3
scikit-learn==1.7.2
numba==0.64.0
llvmlite==0.46.0
pandas==2.3.3

# ------------------------------------------------------------
# Vision
# ------------------------------------------------------------
timm==1.0.25

# ------------------------------------------------------------
# Misc runtime dependencies
# ------------------------------------------------------------
cloudpickle==3.1.2
filelock==3.20.0
fsspec==2025.12.0
joblib==1.5.3
packaging==24.2
requests==2.32.5
tqdm==4.67.3
typing_extensions==4.15.0