BeatFCOS

Beat tracking model that uses a 1-D version of FCOS to detect beats and downbeats.

Results

Installation

1. Clone this repo

2. Install the required packages:

3. Install the python packages:

pip install torch torchsummary numpy torchvision julius torchaudio scipy tqdm soxbindings

Training

The network can be trained using the train.py script.

python train.py --ballroom_audio_dir path/to/ballroom/data --ballroom_annot_dir path/to/ballroom/label --hainsworth_audio_dir path/to/hainsworth/data --hainsworth_annot_dir path/to/hainsworth/label --preload --patience 10 --train_length 2097152 --eval_length 2097152 --act_type PReLU --norm_type BatchNorm --channel_width 32 --channel_growth 32 --augment --batch_size 1 --audio_sample_rate 22050 --num_workers 0

Pre-trained model

Validation

Visualization

Model

CSV datasets

Annotations format

Class mapping format

Acknowledgements

Examples